Transcript
00:00:00 I remember the reaction where he had drawn these characters
00:00:02 and he was slowly moving around
00:00:04 and like people had no experience with 3D navigation.
00:00:06 It was all still keyboard.
00:00:07 We didn’t even have mice set up at that time,
00:00:10 but slowly moving, going up, picked up a key, go to a wall.
00:00:14 The wall disappears in a little animation
00:00:16 and there’s a monster like right there.
00:00:18 And he practically fell out of his chair.
00:00:20 It was just like, ah, and games just didn’t do that.
00:00:24 You know, the games were the God’s eye view.
00:00:26 You were a little invested in your little guy.
00:00:28 You can be like, you know, happy or sad when things happen,
00:00:32 but you just did not get that kind of startled reaction.
00:00:34 You weren’t inside the game.
00:00:36 Something in the back of your brain,
00:00:38 some reptile brain thing is just going,
00:00:40 oh shit, something just happened.
00:00:42 And that was one of those early points where it’s like,
00:00:44 yeah, this is going to make a difference.
00:00:47 This is going to be powerful and it’s going to matter.
00:00:52 The following is a conversation with John Carmack,
00:00:55 widely considered to be one of the greatest programmers
00:00:58 ever.
00:00:59 He was the cofounder of id Software
00:01:01 and the lead programmer on several games
00:01:03 that revolutionized the technology, the experience,
00:01:07 and the role of gaming in our society,
00:01:09 including Commander Keen, Wolfenstein 3D, Doom, and Quake.
00:01:15 He spent many years as the CTO of Oculus VR,
00:01:19 helping to create portals into virtual worlds
00:01:23 and to define the technological path
00:01:25 to the metaverse and meta.
00:01:28 And now he has been shifting some of his attention
00:01:30 to the problem of artificial general intelligence.
00:01:34 This was the longest conversation on this podcast
00:01:38 at over five hours.
00:01:40 And still, I could talk to John many, many more times,
00:01:43 and we hope to do just that.
00:01:46 This is the Lux Readman podcast.
00:01:48 To support it, please check out our sponsors
00:01:50 in the description.
00:01:52 And now, dear friends, here’s John Carmack.
00:01:56 What was the first program you’ve ever written?
00:02:00 Do you remember?
00:02:01 Yeah, I do.
00:02:02 So I remember being in a radio shack
00:02:04 going up to the TRS 80 computers
00:02:06 and learning just enough to be able to do
00:02:09 10 print John Carmack.
00:02:12 And it’s kind of interesting how, of course,
00:02:15 Carnegie and Ritchie kind of standardized Hello World
00:02:19 as the first thing that you do
00:02:20 in every computer programming language in every computer,
00:02:22 but not having any interaction
00:02:24 with the cultures of Unix or any other standardized things.
00:02:28 It was just like, well, what am I gonna say?
00:02:29 I’m gonna say my name.
00:02:30 And then you learn how to do go to 10
00:02:33 and have it scroll all off the screen.
00:02:34 And that was definitely the first thing
00:02:37 that I wound up doing on a computer.
00:02:39 Can I ask you a programming advice?
00:02:41 I was always told in the beginning
00:02:42 that you’re not allowed to use go to statements.
00:02:44 That’s really bad programming.
00:02:45 Is this correct or not?
00:02:46 Jumping around code, can we look at the philosophy
00:02:50 and the technical aspects of the go to statement
00:02:53 that seems so convenient,
00:02:55 but it’s supposed to be bad programming?
00:02:55 Well, so certainly back in the day
00:02:57 in basic programming languages,
00:02:58 you didn’t have proper loops.
00:03:00 You didn’t have four whiles and repeats.
00:03:02 That was the land of Pascal for people
00:03:04 that kind of generally had access to it back then.
00:03:07 So you had no choice but to use go tos.
00:03:10 And as you made what were big programs back then,
00:03:13 which were a thousand line basic program
00:03:15 is a really big program.
00:03:16 They did tend to sort of degenerate into madness.
00:03:19 You didn’t have good editors or code exploration tools.
00:03:23 So you would wind up fixing things in one place,
00:03:26 add a little patch.
00:03:27 And there’s reasons why structured programming
00:03:29 generally helps understanding,
00:03:31 but go tos aren’t poisonous.
00:03:33 Sometimes they’re the right thing to do.
00:03:35 Usually it’s because there’s a language feature missing
00:03:38 like nested breaks or something
00:03:40 where it can sometimes be better to do a go to cleanup
00:03:44 or go to error rather than having multiple flags,
00:03:48 multiple if statements littered throughout things.
00:03:50 But it is rare.
00:03:51 I mean, if you gripped through all of my code right now,
00:03:55 I don’t think any of my current code bases
00:03:57 would actually have a go to,
00:03:58 but deep within sort of the technical underpinnings
00:04:02 of a major game engine,
00:04:03 you’re gonna have some go tos in a couple of places probably.
00:04:07 Yeah, the infrastructure on top of,
00:04:09 like the closer you get to machine code,
00:04:11 the more go tos you’re gonna see,
00:04:12 the more of these like hacks you’re going to see
00:04:15 because the set of features available to you
00:04:18 in low level programming languages is not, is limited.
00:04:22 So print John Carmack,
00:04:25 when is the first time,
00:04:28 if we could talk about love,
00:04:30 that you fell in love with programming?
00:04:31 You said like, this is really something special.
00:04:34 It really was something
00:04:36 that was one of those love at first sight things
00:04:38 where just really from the time
00:04:39 that I understood what a computer was even,
00:04:42 I mean, I remember looking through old encyclopedias
00:04:45 of the black and white photos of the IBM mainframes
00:04:48 at the reel to reel tape decks.
00:04:50 And for people nowadays,
00:04:52 it can be a little hard to understand
00:04:53 what the world was like then from information gathering
00:04:56 where I would go to the libraries
00:04:58 and there would be a couple books on the shelf
00:05:01 about computers and they would be very out of date
00:05:03 even at that point, just not a lot of information,
00:05:06 but I would grab everything that I could find
00:05:09 and devour everything.
00:05:10 Whenever Time or Newsweek had some article about computers,
00:05:14 I would like cut it out with scissors and put it somewhere.
00:05:16 It just, it felt like this magical thing to me,
00:05:19 this idea that the computer would just do exactly
00:05:23 what you told it to.
00:05:24 I mean, and there’s a little bit of the genie monkey’s paw
00:05:26 sort of issues there where you’d better be really,
00:05:28 really careful with what you’re telling it to do,
00:05:31 but it wasn’t gonna back talk you.
00:05:33 It wasn’t gonna have a different point of view.
00:05:34 It was gonna carry out what you told it to do.
00:05:37 And if you had the right commands,
00:05:39 you could make it do these pretty magical things.
00:05:43 And so what kind of programs did you write at first?
00:05:46 So beyond the print, John Carmack.
00:05:48 So I can remember as going through the learning process
00:05:51 where you find at the start,
00:05:53 you’re just learning how to do
00:05:54 the most basic possible things.
00:05:56 And I can remember stuff like a Superman comic
00:05:59 that RadioShack commissioned to have,
00:06:02 it’s like Superman had lost some of his super brain
00:06:04 and kids had to use RadioShack TRS 80 computers
00:06:07 to do calculations for it to help him
00:06:09 kind of complete his heroics.
00:06:11 And I’d find little things like that
00:06:15 and then get a few basic books
00:06:17 to be able to kind of work my way up.
00:06:20 And again, it was so precious back then.
00:06:22 I had a couple of books
00:06:23 that would teach me important things about it.
00:06:25 I had one book that I could start to learn
00:06:28 a little bit of assembly language from,
00:06:30 and I’d have a few books on basic
00:06:32 and some things that I could get from the libraries.
00:06:34 But my goals in the early days
00:06:37 was almost always making games of various kinds.
00:06:40 I loved the arcade games and the early Atari 2600 games.
00:06:44 And being able to do some of those things myself
00:06:47 on the computers was very much what I aspired to.
00:06:51 And it was a whole journey where
00:06:53 if you learn normal basic,
00:06:54 you can’t do any kind of an action game.
00:06:56 You can write an adventure game.
00:06:57 You can write things where you say, what do you do here?
00:07:00 I get sword attack troll, that type of thing.
00:07:03 And that can be done in the context of basic.
00:07:07 But to do things that had moving graphics,
00:07:09 there were only the most limited things
00:07:11 you could possibly do.
00:07:12 You could maybe do breakout or pong
00:07:14 or that sort of thing in low resolution graphics.
00:07:16 And in fact, one of my first sort of major technical hacks
00:07:20 that I was kind of fond of was on the Apple II computers,
00:07:25 they had a mode called low resolution graphics
00:07:28 where of course all graphics were low resolution back then,
00:07:31 but regular low resolution graphics,
00:07:34 it was a grid of 40 by 40 pixels normally,
00:07:37 but they could have 16 different colors.
00:07:39 And I wanted to make a game kind of like
00:07:42 the arcade game Vanguard, just a scrolling game.
00:07:44 And I wanted to just kind of have it scroll vertically up.
00:07:47 And I could move a little ship around.
00:07:49 You could manage to do that in basic,
00:07:50 but there’s no way you could redraw the whole screen.
00:07:53 And I remember at the time just coming up
00:07:56 with what felt like a brainstorm to me
00:07:58 where I knew enough about the way the hardware was controlled
00:08:02 where the text screen and the low resolution graphics screen
00:08:05 were basically the same thing.
00:08:06 And all those computers could scroll their text screen
00:08:09 reasonably.
00:08:10 You could do a listing and it would scroll things up.
00:08:12 And I figured out that I could kind of tweak just a couple
00:08:16 things that I barely understood to put it into a graphics mode
00:08:19 and I could draw graphics and then I could just
00:08:21 do a line feed at the very bottom of the screen.
00:08:24 And then the system would scroll it all up
00:08:26 using an assembly language routine
00:08:28 that I didn’t know how to write back then.
00:08:30 So that was like this first great hack
00:08:33 that sort of had analogs later on in my career
00:08:36 for a lot of different things.
00:08:37 So I found out that I could draw a screen.
00:08:39 I could do a line feed at the bottom.
00:08:40 It would scroll it up once.
00:08:42 I could draw a couple more lines of stuff at the bottom.
00:08:44 And that was my first way to kind of scroll
00:08:47 the screen, which was interesting in that that played
00:08:50 a big part later on in the id Software days as well.
00:08:53 So do efficient drawing where you
00:08:57 don’t have to draw the whole screen,
00:08:59 but you draw from the bottom using
00:09:01 the thing that was designed in the hardware for text output.
00:09:04 Yeah.
00:09:05 Where so much of, until recently,
00:09:08 game design was limited by what you could actually
00:09:12 get the computer to do.
00:09:13 Where it’s easy to say, OK, I want to scroll the screen.
00:09:16 You just redraw the entire screen at a slight offset.
00:09:19 And nowadays, that works just fine.
00:09:21 Computers are ludicrously fast.
00:09:24 But up until a decade ago or so, there
00:09:28 were all these things everybody wanted to do.
00:09:30 But if they knew enough programming
00:09:31 to be able to make it happen, it would happen too slow
00:09:34 to be a good experience.
00:09:36 Either just ridiculously slow or just slow enough
00:09:39 that it wasn’t fun to experience it like that.
00:09:42 So much of the first couple of decades of the programming work
00:09:45 that I did was largely figuring out
00:09:47 how to do something that everybody
00:09:49 knows how they want it to happen.
00:09:51 It just has to happen 2 to 10 times faster
00:09:54 than the straightforward way of doing things
00:09:57 would make it happen.
00:09:58 And it’s different now because at this point, lots of things
00:10:02 you can just do in the most naive possible way,
00:10:05 and it still works out.
00:10:06 You don’t have nearly the creative limitations
00:10:09 or the incentives for optimizing on that level.
00:10:12 And there’s a lot of pros and cons to that.
00:10:14 But I do generally, I’m not going
00:10:16 to do the angry old man shaking my fist at the clouds bit
00:10:20 where back in my day, programmers
00:10:21 had to do real programming.
00:10:23 It’s amazing that you can just pick an idea
00:10:26 and go do it right now.
00:10:27 And you don’t have to be some assembly language wizard
00:10:30 or deep GPU arcanist to be able to figure out
00:10:33 how to make your wishes happen.
00:10:35 Well, there’s still, see, that’s true.
00:10:38 But let me put on my old man with a fist hat
00:10:41 and say that probably the thing that will define the future
00:10:45 still requires you to operate at the limits
00:10:48 of the current system.
00:10:49 So we’ll probably talk about this.
00:10:51 But if you talk about building the metaverse
00:10:54 and building a VR experience that’s compelling,
00:10:57 it probably requires you to not, to go to assembly
00:11:01 or maybe not literally, but sort of spiritually
00:11:05 to go to the limits of what the system is capable of.
00:11:08 And that really was why virtual reality
00:11:10 was specifically interesting to me
00:11:13 where it had all the ties to,
00:11:15 you could say that even back in the early days,
00:11:17 I have some old magazine articles
00:11:19 that’s talking about Doom as a virtual reality experience
00:11:22 back when just seeing anything in 3D.
00:11:25 So you could say that we’ve been trying to build
00:11:27 those virtual experiences from the very beginning.
00:11:29 And in the modern era of virtual reality,
00:11:32 especially on the mobile side of things,
00:11:34 when it’s standalone and you’re basically using
00:11:36 a cell phone chip to be able to produce
00:11:39 these very immersive experiences,
00:11:41 it does require work.
00:11:43 It’s not at the level of what an old school
00:11:45 console game programmer would have operated at
00:11:48 where you’re looking at hardware registers
00:11:50 and you’re scheduling all the DMA accesses,
00:11:53 but it is still definitely a different level
00:11:55 than what a web developer or even a PC Steam game developer
00:12:00 usually has to work at.
00:12:01 And again, it’s great.
00:12:02 There’s opportunities for people that wanna operate
00:12:04 at either end of that spectrum there
00:12:06 and still provide a lot of value to the world.
00:12:09 Let me ask you sort of a big question about preference.
00:12:15 What would you say is the best programming language?
00:12:19 Your favorite, but also the best.
00:12:22 You’ve seen throughout your career,
00:12:25 you’re considered by many to be
00:12:26 the greatest programmer ever.
00:12:29 I mean, it’s so difficult to place that label on anyone,
00:12:32 but if you put it on anyone, it’s you.
00:12:34 So let me ask you these kind of ridiculous questions
00:12:36 of what’s the best band of all time,
00:12:38 but in your case, what’s the best programming language?
00:12:41 Everything has all the caveats about it.
00:12:43 But so what I use, so nowadays I do program
00:12:47 a reasonable amount of Python for AI, ML sorts of work.
00:12:52 I’m not a native Python programmer.
00:12:54 It’s something I came to very late in my career.
00:12:57 I understand what it’s good for.
00:12:58 But you don’t dream in Python.
00:13:00 I do not.
00:13:00 And it has some of those things
00:13:02 where there’s some amazing stats when you say,
00:13:04 if you just start, if you make a loop,
00:13:07 a triply nested loop and start doing operations in Python,
00:13:10 you can be thousands to potentially a million times slower
00:13:14 than a proper GPU tensor operation.
00:13:17 And these are staggering numbers.
00:13:19 You can be as much slower as we’ve almost gotten faster
00:13:22 in our pace of progress and all this other miraculous stuff.
00:13:26 So your intuitions about inefficiencies
00:13:28 within the Python sort of…
00:13:30 It keeps hitting me upside the face
00:13:32 where it’s gotten to the point now I understand.
00:13:34 It’s like, okay, you just can’t do a loop
00:13:35 if you care about performance in Python.
00:13:38 You have to figure out how you can reformat this
00:13:41 into some big vector operation
00:13:43 or something that’s going to be done completely
00:13:44 within a C++ library.
00:13:47 But the other hand is it’s amazingly convenient
00:13:49 and you just see stuff that people are able
00:13:52 to cobble together by you just import
00:13:54 a few different things and you can do stuff
00:13:56 that nobody on earth could do 10 years ago.
00:13:58 And you can do it in a little cookbook thing
00:14:00 that you copy paste it out of a website.
00:14:02 So that is really great.
00:14:04 When I’m sitting down to do what I consider
00:14:06 kind of serious programming, it’s still in C++.
00:14:10 And it’s really kind of a C flavored C++ at that
00:14:13 where I’m not big into the modern template
00:14:17 metaprogramming sorts of things.
00:14:18 I see a lot of train wrecks coming from
00:14:21 some of that over abstraction.
00:14:23 I spent a few years really going kind of deep
00:14:26 into the kind of the historical Lisp work
00:14:30 and Haskell and some of the functional programming
00:14:32 sides of things.
00:14:33 And there is a lot of value there
00:14:37 in the way you think about things.
00:14:38 And I changed a lot of the way I write my C and C++ code
00:14:42 based on what I learned about the value
00:14:44 that comes out of not having this random mutable state
00:14:48 that you kind of lose track of.
00:14:50 Because something that many people don’t really appreciate
00:14:53 till they’ve been at it for a long time
00:14:55 is that it’s not the writing of the program initially,
00:14:58 it’s the whole lifespan of the program.
00:15:00 And that’s when it’s not necessarily just
00:15:02 how fast you wrote it or how fast it operates,
00:15:05 but it’s how can it bend and adapt
00:15:07 as situations change.
00:15:09 And then the thing that I’ve really been learning
00:15:11 in my time at Meta with the Oculus and VR work
00:15:14 is it’s also how well it hands off
00:15:16 between the continuous kind of revolving door of programmers
00:15:19 taking over maintenance and different things
00:15:21 and how you get people up to speed in different areas.
00:15:24 And there’s all these other different aspects of it.
00:15:27 So C++ is a good language for handover between engineers.
00:15:32 Probably not the best.
00:15:34 And there’s some really interesting aspects to this
00:15:36 where in some cases languages
00:15:39 that are not generally thought well of for many reasons,
00:15:43 like C is derided pretty broadly that yes,
00:15:46 obviously all of these security flaws
00:15:48 that happen with the memory and unsafeness
00:15:50 and buffer overruns and the things that you’ve got there,
00:15:53 but there is this underappreciated aspect
00:15:56 to the language is so simple, anyone can go.
00:15:59 And if you know C, you can generally jump in someplace
00:16:03 and not have to learn what paradigms they’re using
00:16:06 because there just aren’t that many available.
00:16:08 I think there’s some really, really well written C code.
00:16:12 Like I find it great that if I’m messing around
00:16:15 with something in open BSD say,
00:16:17 I mean, I can be walking around in the kernel
00:16:19 and I’m like, I understand everything that’s going on here.
00:16:22 It’s not hard for me to figure out what I need to do
00:16:25 to make whatever change that I need to
00:16:29 while you can have more significant languages.
00:16:32 Like it’s a downside of Lisp
00:16:34 where I don’t regret the time that I spent with Lisp.
00:16:37 I think that it did help my thinking
00:16:40 about programming in some ways,
00:16:42 but the people that are the biggest defenders of Lisp
00:16:45 are saying how malleable of a language it is
00:16:47 that if you write a huge Lisp program,
00:16:50 you’ve basically invented your own kind of language
00:16:53 and structure because it’s not the primitives
00:16:55 of the language you’re using very much.
00:16:57 It’s all of the things you’ve built on top of that.
00:16:59 And then a language like Racket,
00:17:01 kind of one of the more modern Lisp versions,
00:17:03 it’s essentially touted as a language
00:17:05 for building other languages.
00:17:07 And I understand the value of that
00:17:10 for a tiny little project,
00:17:12 but the idea of that for one of these longterm
00:17:14 supported by lots of people kind of horrifies me
00:17:18 where all of those abstractions that you’re like,
00:17:20 okay, you can’t touch this code till you educate yourself
00:17:23 on all of these things that we’ve built on top of that.
00:17:26 And it was interesting to see how when Google made Go,
00:17:30 a lot of the criticisms of that are it’s like,
00:17:33 wow, this is not a state of the art language.
00:17:35 This language is just so simple and almost crude.
00:17:38 And you could see the programming language people
00:17:40 just looking down at it.
00:17:41 But it does seem to be quite popular as basically saying,
00:17:45 this is the good things about C,
00:17:47 everybody can just jump right in and use it.
00:17:49 You don’t need to restructure your brain
00:17:52 to write good code in it.
00:17:53 So I wish that I had more opportunity
00:17:56 for doing some work in Go.
00:17:59 Rust is the other modern language that everybody talks about
00:18:02 that I’m not fit to pass judgment on.
00:18:04 I’ve done a little bit beyond Hello World.
00:18:06 I wrote some like video decompression work in Rust
00:18:09 just as an exercise, but that was a few years ago
00:18:12 and I haven’t really used it since.
00:18:15 The best programming language is the one that works generally
00:18:17 that you’re currently using.
00:18:19 Because that’s another trap is in almost every case
00:18:22 I’ve seen when people mixed languages on a project,
00:18:25 that’s a mistake.
00:18:26 I would rather stay just in one language
00:18:29 so that everybody can work across the entire thing.
00:18:31 And we have, I get meta, we have a lot of projects
00:18:34 that use kind of React frameworks.
00:18:36 So you’ve got JavaScript here
00:18:37 and then you have C++ for real work.
00:18:40 And then you may have Java interfacing
00:18:42 with some other part of the Android system.
00:18:44 And those are all kind of horrible things.
00:18:47 And that was one thing that I remember talking
00:18:51 with Boz at Facebook about it where like,
00:18:54 man, I wish we could have just said,
00:18:56 we’re only hiring C++ programmers.
00:18:59 And he just thought from the Facebook meta perspective,
00:19:02 well, we just wouldn’t be able to find enough.
00:19:05 With the thousands of programmers they’ve got there,
00:19:08 it is not necessarily a dying breed,
00:19:10 but you can sure find a lot more Java
00:19:12 or JavaScript programmers.
00:19:14 And I kind of mentioned that to Elon one time
00:19:17 and he was kind of flabbergasted about that.
00:19:21 It’s like, well, you just,
00:19:22 you go out and you find those programmers
00:19:23 and you don’t hire the other programmers
00:19:25 that don’t do the languages that you wanna use.
00:19:28 But right now, I guess, yeah,
00:19:29 they’re using JavaScript on a bunch of the SpaceX work
00:19:32 for the UI side of things.
00:19:33 When you go find UI programmers,
00:19:35 they’re JavaScript programmers.
00:19:36 I wonder if that’s because there’s a lot
00:19:38 of JavaScript programmers.
00:19:39 Because I do think that great programmers are rare.
00:19:44 That it’s not, you know, if you just look at statistics
00:19:48 of how many people are using different programming languages,
00:19:51 that doesn’t tell you the story
00:19:53 of what the great programmers are using.
00:19:55 And so you have to really look at what you were speaking to,
00:19:59 which is the fundamentals of a language.
00:20:00 What does it encourage?
00:20:01 How does it encourage you to think?
00:20:03 What kind of systems does it encourage you to build?
00:20:05 There is something about C++
00:20:08 that has elements of creativity,
00:20:11 but forces you to be an adult about your programming.
00:20:15 It expects you to be an adult.
00:20:16 It does not force you to.
00:20:18 And so it brings out people that are willing to be creative
00:20:23 in terms of building large systems
00:20:25 and coming up with interesting solutions,
00:20:27 but at the same time have the sort of the good
00:20:31 software engineering practices that amend themselves
00:20:35 to real world systems.
00:20:37 Let me ask you about this other language, JavaScript.
00:20:41 So if we, you know, aliens visit in thousands of years,
00:20:46 and humans are long gone, something tells me
00:20:49 that most of the systems they find will be run by humans.
00:20:53 Will be running JavaScript.
00:20:55 I kind of think that if we’re living in a simulation,
00:20:58 it’s written in JavaScript.
00:21:01 You know, for the longest time, even still,
00:21:05 JavaScript didn’t get any respect,
00:21:08 and yet it runs so much of the world,
00:21:10 and an increasing number of the world.
00:21:12 Is it possible that everything will be written
00:21:15 in JavaScript one day?
00:21:16 So the engineering under JavaScript
00:21:18 is really pretty phenomenal.
00:21:21 The systems that make JavaScript run as fast
00:21:24 as it does right now are kind of miracles
00:21:27 of modern engineering in many ways.
00:21:29 It does feel like it is not an optimal language
00:21:34 for all the things that it’s being used for,
00:21:36 or an optimal distribution system to build huge apps
00:21:40 in something like this without type systems and so on.
00:21:45 But I think for a lot of people,
00:21:47 it does reasonably the necessary things.
00:21:50 It’s still a C flavored language.
00:21:52 It’s still a braces and semicolon language.
00:21:55 It’s not hard for people to be trained in JavaScript
00:21:59 and then understand the roots of where it came from.
00:22:02 I think garbage collection is unequivocally a good thing
00:22:07 for most programs to be written in.
00:22:08 It’s funny that I still, just this morning,
00:22:11 I was seeing a Twitter thread
00:22:13 of a bunch of really senior game dev people
00:22:15 arguing about the virtues and costs of garbage collection.
00:22:18 You will run into some people that are top notch programmers
00:22:21 that just say, no, this is literally not a good thing.
00:22:24 Oh, because it makes you lazy?
00:22:25 Yes, that it makes you not think about things.
00:22:28 And I do disagree.
00:22:29 I think that there is so much objective data
00:22:33 on the vulnerabilities that have happened
00:22:35 in C and C++ programs,
00:22:37 sometimes written by the best programmers in the world.
00:22:40 It’s like nobody is good enough to avoid
00:22:42 ever shooting themselves in the foot with that.
00:22:44 You write enough C code,
00:22:45 you’re going to shoot yourself in the foot.
00:22:47 And garbage collection is a very great thing
00:22:50 for the vast majority of programs.
00:22:51 It’s only when you get into the tightest of real time things
00:22:54 that you start saying, it’s like, no,
00:22:56 the garbage collection has more costs
00:22:58 than it has benefits for me there,
00:22:59 but that’s not 99 plus percent
00:23:02 of all the software in the world.
00:23:04 So JavaScript is not terrible in those ways.
00:23:09 And so much of programming is not the language itself.
00:23:14 It’s the infrastructure around that surrounds it.
00:23:17 All the libraries that you can get
00:23:19 and the different stuff that you can,
00:23:20 ways you can deploy it,
00:23:22 the portability that it gives you.
00:23:24 And JavaScript is really strong on a lot of those things
00:23:27 where for a long time, and it still does if I look at it,
00:23:31 the web stack about everything that has to go
00:23:34 when you do something really trivial in JavaScript
00:23:36 and it shows up on a web browser
00:23:38 to kind of X ray through that
00:23:40 and see everything that has to happen
00:23:42 for your one little JavaScript statement
00:23:44 to turn into something visible in your web browser,
00:23:48 it’s very, very disquieting,
00:23:51 just the depth of that stack
00:23:53 and the fact that so few people can even comprehend
00:23:56 all of the levels that are going on there.
00:23:59 But it’s again, I have to caution myself
00:24:02 to not be the in the good old days old man about it
00:24:05 because clearly there’s enormous value here.
00:24:08 The world does run on JavaScript
00:24:11 to a pretty good approximation there
00:24:13 and it’s not falling apart.
00:24:14 There’s a bunch of scary stuff
00:24:16 where you look at console logs
00:24:18 and you just see all of these bad things that are happening
00:24:20 but it’s still kind of limping along
00:24:22 and nobody really notices.
00:24:24 But so much of my systems design
00:24:28 and systems analysis goes around.
00:24:30 You should understand what the speed of light is,
00:24:32 like what would be the best you could possibly do here.
00:24:36 And it sounds horrible, but in a lot of cases,
00:24:39 you can be a thousand times off your speed of light,
00:24:42 velocity for something and it’s still be okay.
00:24:45 And in fact, it can even sometimes still be
00:24:48 the optimal thing in a larger system standpoint
00:24:51 where there’s a lot of things
00:24:52 that you don’t wanna have to parachute in someone like me
00:24:55 to go in and say, make this webpage run
00:24:59 a thousand times faster,
00:25:01 make this web app into a hardcore native application
00:25:05 that starts up in 37 milliseconds
00:25:08 and everything responds in less than one frame latency.
00:25:11 That’s just not necessary.
00:25:12 And if somebody wants to go pay me millions of dollars
00:25:15 to do software like that,
00:25:16 when they can take somebody right out of a bootcamp
00:25:19 and say, spin up an application for this.
00:25:21 Often being efficient is not really the best metric.
00:25:26 And it’s like, that applies in a lot of areas
00:25:29 where it’s kind of interesting how a lot of our appliances
00:25:32 and everything are all built around energy efficiency,
00:25:36 sometimes at the expense of robustness in some other ways
00:25:39 or higher costs in other ways
00:25:41 where there’s interesting things
00:25:42 where energy or electricity could become much cheaper
00:25:46 in a future world.
00:25:47 And that could change our engineering trade offs
00:25:49 for the way we build certain things
00:25:50 where you could throw away efficiency
00:25:53 and actually get more benefits that actually matter.
00:25:56 I mean, that’s one of the directions
00:25:59 I was considering swerving into was nuclear energy.
00:26:02 When I was kind of like, what do I wanna do next?
00:26:04 It was either gonna be cost effective nuclear fission
00:26:08 or artificial general intelligence.
00:26:10 And one of my pet ideas there is like,
00:26:14 people don’t understand how cheap nuclear fuel is.
00:26:18 And there would be ways that,
00:26:21 you could be a quarter of the efficiency or less,
00:26:24 but if it wound up making your plant 10 times cheaper,
00:26:27 that could be a radical innovation in something like that.
00:26:31 So there’s like some of these thoughts
00:26:32 around like direct fission energy conversion,
00:26:35 fission fragment conversion,
00:26:36 that maybe you build something
00:26:38 that doesn’t require all the steam turbines and everything,
00:26:40 even if it winds up being less efficient.
00:26:42 So that applies a lot in programming
00:26:45 where it’s always good to know what you could do.
00:26:49 If you really sat down and took it far,
00:26:52 because sometimes there’s discontinuities
00:26:53 like around user reaction times,
00:26:56 there are some points where the difference between
00:26:59 operating in one second and 750 milliseconds,
00:27:02 not that huge.
00:27:03 You’ll see it in webpage statistics,
00:27:05 but most of the usability stuff, not that great.
00:27:08 But if you get down to 50 milliseconds,
00:27:11 then all of a sudden this just feels amazing.
00:27:13 It’s just like doing your bidding instantly
00:27:15 rather than you’re giving it a command,
00:27:17 twiddling your thumbs, waiting for it to respond.
00:27:19 So sometimes it’s important to really crunch hard
00:27:23 to get over some threshold,
00:27:25 but there are broad basins in the value metric
00:27:28 for lots of work where it just doesn’t pay
00:27:31 to even go that extra mile.
00:27:33 And there are craftsmen that just don’t wanna buy that
00:27:36 and more power to them.
00:27:37 If somebody just wants to say,
00:27:39 no, I’m going to be, my pride is in my work,
00:27:42 I’m never going to do something
00:27:44 that’s not as good as I could possibly make it,
00:27:46 I respect that and sometimes I am that person,
00:27:50 but I try to focus more on the larger value picture
00:27:53 and you do pick your battles
00:27:55 and you deploy your resources in the play
00:27:57 that’s going to give you the best user value in the end.
00:28:00 Well, if you look at the evolution of life on Earth
00:28:04 as a kind of programming effort,
00:28:10 it seems like efficiency isn’t the thing
00:28:13 that’s being optimized for.
00:28:15 Like natural selection is very inefficient,
00:28:17 but it kind of adapts and through the process
00:28:21 of adaptations building more and more complex systems
00:28:24 that are more and more intelligent,
00:28:25 the final result is kind of pretty interesting.
00:28:28 And so I think of JavaScript the same way.
00:28:30 It’s like this giant mess
00:28:33 that things naturally die off if they don’t work
00:28:36 and if they become useful to people,
00:28:39 they kind of naturally live
00:28:40 and then you build this community,
00:28:42 large community of people that are generating code
00:28:45 and some code is sticky, some is not
00:28:48 and nobody knows the inefficiencies or the efficiencies
00:28:52 or the breaking points, like how reliable this code is
00:28:54 and you kind of just run it, assume it works
00:28:57 and then get unpleasantly surprised
00:29:00 and then that’s very kind of the evolutionary process.
00:29:03 So that’s a really good analogy
00:29:04 and we can go a lot of places with that
00:29:06 where in the earliest days of programming,
00:29:09 when you had finite, you could count the bytes
00:29:11 that you had to work on this,
00:29:13 you had all the kind of hackers playing code golf
00:29:15 to be one less instruction
00:29:17 than the other person’s multiply routine
00:29:19 to kind of get through and it was so perfectly crafted.
00:29:22 It was a crystal piece of artwork when you had a program
00:29:26 because there just were not that many,
00:29:28 you couldn’t afford to be lazy in different ways
00:29:31 and in many ways I see that as akin to the symbolic AI work
00:29:34 where again, if you did not have the resources to just say,
00:29:38 well, we’re gonna do billions and billions
00:29:40 of programmable weights here,
00:29:42 you have to turn it down into something
00:29:44 that is symbolic and crafted like that
00:29:47 but that’s definitely not the way DNA and life
00:29:51 and biological evolution and things work.
00:29:55 On the one hand, it’s almost humbling
00:29:58 how little programming code is in our bodies.
00:30:00 We’ve got a couple of billion base pairs
00:30:02 and it’s like this all fits on a thumb drive for years now
00:30:05 and then our brains are even a smaller section of that.
00:30:08 You’ve got maybe 50 megabytes
00:30:10 and this is not like Shannon Limit perfectly
00:30:13 information dense conveyances here.
00:30:17 It’s like these are messy codes,
00:30:19 they’re broken up into amino acids.
00:30:21 A lot of them don’t do important things
00:30:23 or they do things in very awkward ways
00:30:26 but it is this process of just accumulation
00:30:29 on top of things and you need scale,
00:30:33 both you need scale for the population for that to work out
00:30:36 and in the early days in the 50s and 60s,
00:30:39 the kind of ancient era of computers
00:30:42 where you could count when they say like
00:30:44 when the internet started even in the 70s,
00:30:46 there were like 18 hosts or something on it.
00:30:48 It was this small finite number
00:30:50 and you were still optimizing everything
00:30:52 to be as good as you possibly could be
00:30:54 but now it’s billions and billions of devices
00:30:57 and everything going on and you can have
00:31:00 this very much natural evolution going on
00:31:03 where lots of things are tried,
00:31:06 lots of things are blowing up,
00:31:07 venture capitalists lose their money
00:31:09 when a startup invested in the wrong tech stack
00:31:11 and things completely failed or failed the scale
00:31:14 but good things do come out of it
00:31:17 and it’s interesting to see the mimetic evolution
00:31:21 of the way different things happen
00:31:22 like mentioning hello world at the beginning.
00:31:24 It’s funny how some little thing like that
00:31:26 where everybody, every programmer knows hello world now
00:31:29 and that was a completely arbitrary sort of decision
00:31:32 that just came out of the dominance of Unix and C
00:31:35 and early examples of things like that.
00:31:38 So millions of experiments are going on all the time
00:31:42 but some things do kind of rise to the top
00:31:44 and win the fitness war for whether it’s mind space
00:31:48 or programming techniques or anything.
00:31:50 Like there’s a site on Stack Exchange called CodeGolf
00:31:54 where people compete to write the shortest possible program
00:31:57 for a particular task in all the different kinds
00:32:00 of languages and it’s really interesting to see
00:32:04 folks kind of, they’re masters of their craft,
00:32:10 really play with the limits of programming languages.
00:32:13 It’s really beautiful to see
00:32:14 and across all the different programming languages,
00:32:17 you get to see some of these weird programming languages
00:32:20 and mainstream ones, difference between Python 2 and 3,
00:32:25 you get to see the difference between C and C++ and Java
00:32:27 and you get to see JavaScript, all of that
00:32:30 and it’s kind of inspiring to see how much depth
00:32:38 of possibility there is within programming languages
00:32:41 that CodeGolf kind of tasks reveal.
00:32:44 Most of us, if you do any kind of programming,
00:32:46 you kind of do boring kind of very vanilla type of code.
00:32:51 That’s the way to build large systems
00:32:52 but it’s nice to see that the possibility
00:32:54 of creative genius is still within those languages.
00:32:57 It’s laden within those languages.
00:33:00 So given that you are once again,
00:33:03 one of the greatest programmers ever,
00:33:05 what do you think makes a good programmer?
00:33:08 Maybe a good modern programmer.
00:33:10 So I just gave a long rant slash lecture at Meta
00:33:15 to the TPM organization and my biggest point was
00:33:20 everything that we’re doing really should flow
00:33:23 from user value, all the good things that we’re doing.
00:33:26 It’s like, we’re not technical people.
00:33:28 It’s like, you shouldn’t be taking pride
00:33:30 just in the specific thing.
00:33:32 Like CodeGolf is the sort of thing, it’s a fun puzzle game
00:33:34 but that really should not be a major motivator for you.
00:33:38 It’s like, we’re solving problems for people
00:33:40 or we’re providing entertainment to people.
00:33:41 We’re doing something of value to people
00:33:44 that’s displacing something else in their life.
00:33:46 So we wanna be providing a net value
00:33:49 over what they could be doing
00:33:51 but instead they’re choosing to use our products.
00:33:53 And that’s where, I mean, it sounds trite or corny
00:33:56 but I fundamentally do think that’s how you make
00:33:59 the world a better place.
00:34:00 If you have given more value to people
00:34:02 than it took you and your team to create,
00:34:05 then the world’s a better place.
00:34:07 People have, they’ve gone from something of lesser value,
00:34:10 chosen to use your product
00:34:11 and their life feels better for that.
00:34:13 And if you’ve produced that economically,
00:34:16 that’s a really good thing.
00:34:19 On the other hand, if you spent ridiculous amounts of money
00:34:22 you’ve just kind of shoveled a lot of cash
00:34:24 into a wood chipper there
00:34:25 and you should maybe not feel so good
00:34:28 about what you’re doing.
00:34:29 So being proud about like a specific architecture
00:34:33 or a specific technology or a specific code sequence
00:34:36 that you’ve done, it’s great to get a little smile
00:34:39 like a tiny little dopamine hit for that
00:34:41 but the top level metrics should be
00:34:43 that you’re building things of value.
00:34:45 Now you can get into the argument about how you,
00:34:48 what is user value?
00:34:49 How do you actually quantify that?
00:34:51 And there can be big arguments about that
00:34:53 but it’s easy to be able to say,
00:34:55 okay, this pissed off user there is not getting value
00:34:57 from what you’re doing.
00:34:59 This user over there with the big smile on their face,
00:35:01 the moment of delight when something happened,
00:35:04 there’s a value that’s happened there.
00:35:05 I mean, you have to at least accept
00:35:07 that there is a concept of user value
00:35:09 even if you have trouble exactly quantifying it,
00:35:12 you can usually make relative arguments about it.
00:35:15 Well, this was better than this, we’ve improved things.
00:35:18 So being a servant to the user is your job
00:35:23 when you’re a developer, you wanna be producing something
00:35:26 that other people are gonna find valuable.
00:35:30 And if you are technically inclined
00:35:32 then finding the right levers to be able to pull
00:35:35 to be able to make a design that’s going to produce
00:35:39 the most value for the least amount of effort.
00:35:41 And it always has to be kind of divided,
00:35:44 there’s a ratio there where you,
00:35:45 it’s a problem at the big tech companies,
00:35:48 whether it’s Meta Google, Apple, Microsoft, Amazon,
00:35:52 companies that have almost infinite money.
00:35:55 I mean, I know their CFO will complain
00:35:57 that it’s not infinite money
00:35:58 but from most developers standpoints
00:36:00 it really does feel like it.
00:36:02 And it’s almost counterintuitive
00:36:04 that if you’re working hard as a developer on something,
00:36:07 there’s always this thought,
00:36:08 if only I had more resources, more people, more RAM,
00:36:12 more megahertz, then my product will be better.
00:36:16 And that sense that at certain points, it’s certainly true
00:36:19 that if you are really hamstrung by this,
00:36:22 removing an obstacle will make a better product,
00:36:25 make more value.
00:36:26 But if you’re not making your core design decisions
00:36:30 in this fiercely competitive way
00:36:33 where you’re saying feature A or feature B,
00:36:35 you can’t just say, let’s do both.
00:36:38 Because then you’re not making a value judgment about them.
00:36:41 You’re just saying, well, they both seem good.
00:36:43 I don’t wanna necessarily have to pick out
00:36:45 which one is better or how much better
00:36:47 and tell team B that, sorry, we’re not gonna do this
00:36:51 because A is more important.
00:36:53 But that notion of always having to really critically value
00:36:57 what you’re doing, your time, the resources you expend,
00:36:59 even the opportunity cost of doing something else,
00:37:02 that’s super important.
00:37:04 Well, let me ask you about the big debates
00:37:07 that you’re mentioning of how to measure value.
00:37:12 Is it possible to measure it kind of numerically
00:37:19 or can you do the sort of Johnny Ive,
00:37:22 the designer route of imagining
00:37:26 sort of somebody using a thing
00:37:28 and imagining a smile on their face,
00:37:30 imagining the experience of love and joy
00:37:34 that you have when you use the thing.
00:37:35 That’s from a design perspective
00:37:37 or if you’re building more like a lower level thing
00:37:40 for like Linux, you imagine a developer
00:37:43 that might come across this and use it
00:37:45 and become happy and better off because of it.
00:37:50 So where do you land on those things?
00:37:52 Is it measurable?
00:37:53 So I imagine like Meta and Google
00:37:57 will probably try to measure the thing.
00:37:58 They’ll try to, it’s like you try to optimize engagement
00:38:01 or something, let’s measure engagement.
00:38:03 And then I think there is a kind of,
00:38:05 I mean, I admire the designer ethic of like,
00:38:09 think of a future that’s immeasurable
00:38:13 and you try to make somebody in that future
00:38:15 that’s different from today happy.
00:38:17 So I do usually favor,
00:38:20 if you can get any kind of a metric that’s good,
00:38:23 by all means, listen to the data.
00:38:25 But you can go too far there where we’ve had problems
00:38:28 where it’s like, hey, we had a performance regression
00:38:30 because our fancy new telemetry system
00:38:33 is doing a bazillion file writes
00:38:36 to kind of archive this stuff
00:38:37 because we needed to collect information
00:38:38 to determine if we were doing,
00:38:40 if our plans were good.
00:38:42 So when information is available,
00:38:45 you should never ignore it.
00:38:46 I mean, all of it.
00:38:47 For actual users using the thing,
00:38:49 human beings using the thing,
00:38:51 large number of human beings,
00:38:52 and you get to see sort of a lot of large numbers.
00:38:54 So there’s the zero to one problem
00:38:55 of when you’re doing something really new,
00:38:57 you do kind of have to make a guess.
00:38:59 But one of the points that I’ve been making at Meta
00:39:02 is we have more than enough users now
00:39:05 that anything somebody wants to try in VR,
00:39:08 we have users that will be interested in that.
00:39:10 You do not get to make a completely
00:39:13 greenfield blue sky pitch and say,
00:39:15 I’m going to do this because I think it might be interesting.
00:39:18 I challenge everyone.
00:39:20 There are going to be people,
00:39:21 whether it’s working in VR on your,
00:39:24 like on your desktop replacement
00:39:26 or communicating with people in different ways
00:39:29 or playing the games.
00:39:31 There are going to be probably millions of people,
00:39:35 or at least if you pick some tiny niche
00:39:37 that we’re not in right now,
00:39:38 there’s still going to be thousands of people out there
00:39:40 that have the headsets that would be your target market.
00:39:43 And I tell people, pay attention to them.
00:39:46 Don’t invent fictional users.
00:39:47 Don’t make an Alice Bob Charlie
00:39:50 that fits whatever matrix of tendencies
00:39:53 that you want to break the market down to,
00:39:55 because it’s a mistake to think about imaginary users
00:39:58 when you’ve got real users that you could be working with.
00:40:01 But on the other hand,
00:40:03 there is value to having a kind of wholeness
00:40:07 of vision for a product.
00:40:09 And companies like Meta have,
00:40:13 they understand the trade offs where you can have a company
00:40:16 like SpaceX or Apple in the Steve Jobs era,
00:40:20 where you have a very powerful leading personality
00:40:23 that can micromanage at a very low level
00:40:26 and can say it’s like, no, that handle needs to be different
00:40:29 or that icon needs to change the tint there.
00:40:32 And they clearly get a lot of value out of it.
00:40:34 They also burn through a lot of employees
00:40:37 that have horror stories to tell
00:40:39 about working there afterwards.
00:40:41 My position is that you’re at your best
00:40:45 when you’ve got a leader that is at their limit
00:40:48 of what they can kind of comprehend
00:40:49 of everything below them.
00:40:51 And they can have an informed opinion
00:40:53 about everything that’s going on.
00:40:55 And you take somebody, you’ve got to believe
00:40:57 that somebody that has 30, 40 years of experience,
00:41:01 you would hope that they’ve got wisdom
00:41:02 that the just out of bootcamp person contributing
00:41:06 doesn’t have.
00:41:07 And that if they’re like, well, that’s wrong there,
00:41:09 you probably shouldn’t do it that way
00:41:10 or even just don’t do it that way, do it another way.
00:41:14 So there’s value there,
00:41:15 but it can’t go beyond a certain level.
00:41:17 I mean, I have Steve Jobs stories of him saying things
00:41:21 that are just wrong right in front of me
00:41:23 about technical things because he was not operating
00:41:25 at that level.
00:41:28 But when it does work and you do get
00:41:30 that kind of passionate leader
00:41:32 that’s thinking about the entire product
00:41:34 and just really deeply cares
00:41:35 about not letting anything slip through the cracks,
00:41:39 I think that’s got a lot of value.
00:41:40 But the other side of that is the people saying that,
00:41:42 well, we wanna have these independent teams
00:41:44 that are bubbling up the ideas
00:41:46 because it’s almost it’s anti capitalist
00:41:49 or anti free market to say,
00:41:50 it’s like I want my grand, my great leader
00:41:53 to go ahead and dictate all these points there
00:41:55 where clearly free markets bring up things
00:41:57 that you don’t expect.
00:41:59 Like in VR, we saw a bunch of things
00:42:01 like it didn’t turn out at all
00:42:03 the way the early people thought
00:42:04 were gonna be the key applications
00:42:06 and things that would not have been approved
00:42:08 by the dark cabal making the decisions
00:42:12 about what gets into the store turn out
00:42:14 to in some cases be extremely successful.
00:42:17 So yeah, I definitely kind of wanted to be,
00:42:20 there was a point where I did make a pitch.
00:42:22 It’s like, hey, make me VR dictator
00:42:23 and I’ll go in and get shit done.
00:42:25 I am, and that’s just, it’s not in the culture at Meta.
00:42:28 You know, and they understand the trade offs they,
00:42:31 and that’s just not the way,
00:42:32 that’s not the company that they want,
00:42:34 the team that they wanna do.
00:42:37 It’s fascinating because VR and we’ll talk about it more.
00:42:40 It’s still unclear to me
00:42:44 in what way VR will change the world
00:42:46 because it does seem clear that VR
00:42:48 will somehow fundamentally transform this world
00:42:51 and it’s unclear to me how yet.
00:42:54 Let me know when you wanna get into that.
00:42:56 Who will?
00:42:57 Well, hold on a second.
00:42:58 So in the stick to the you being the best programmer ever.
00:43:02 Okay, in the early days when you didn’t have,
00:43:05 when you didn’t have adult responsibilities
00:43:07 of leading teams and all that kind of stuff
00:43:09 and you can focus on just being a programmer,
00:43:13 what did the productive day in the life
00:43:15 of John Carmack look like?
00:43:17 How many hours is the keyboard?
00:43:19 How much sleep?
00:43:20 What was the source of calories that fueled the brain?
00:43:24 What was it like?
00:43:25 What times did you wake up?
00:43:26 So I was able to be remarkably consistent
00:43:29 about what was good working conditions for me
00:43:32 for a very long time.
00:43:34 I was never one of the programmers
00:43:37 that would do all nighters going through work
00:43:40 for 20 hours straight.
00:43:41 It’s like my brain generally starts turning to mush
00:43:43 after 12 hours or so.
00:43:45 But the hard work is really important
00:43:48 and I would work for decades.
00:43:51 I would work 60 hours a week.
00:43:52 I would work a 10 hour day, six days a week
00:43:55 and try to be productive at that.
00:43:58 Now my schedule shifted around a fair amount
00:44:00 when I was young without any kids
00:44:02 and any other responsibilities.
00:44:04 I was on one of those cycling schedules
00:44:06 where I’d kind of get in an hour later each day
00:44:08 and roll around through the entire time
00:44:11 and I’d wind up kind of pulling in at two or three
00:44:13 in the afternoon sometimes and then working again
00:44:16 past midnight or two in the morning.
00:44:19 And that was, when it was just me trying
00:44:22 to make things happen and I was usually isolated off
00:44:26 in my office, people generally didn’t bother me much at in
00:44:30 and I could get a lot of programming work done that way.
00:44:34 I did settle into a more normal schedule
00:44:36 when I was taking kids to school and things like that.
00:44:39 So kids were the forcing function that got you to wake up
00:44:42 at the same time each day.
00:44:43 It’s not clear to me that there was much of a difference
00:44:46 in the productivity with that where I kind of feel,
00:44:50 if I just get up when I feel like it,
00:44:52 it’s usually a little later each day
00:44:54 but I just recently made the focusing decision
00:44:57 to try to push my schedule back a little bit earlier
00:44:59 to getting up at eight in the morning
00:45:01 and trying to shift things around.
00:45:04 Like I’m often doing experiments with myself
00:45:07 about what should I be doing to be more productive
00:45:10 and one of the things that I did realize was happening
00:45:13 in recent months where I would go for a walk or a run.
00:45:18 I cover like four miles a day and I would usually do that
00:45:22 just as the sun’s going down here in Texas now
00:45:25 and it’s still really damn hot but I’d go out
00:45:27 at 8.30 or something and cover the time there
00:45:31 and then the showering and it was putting a hold in my day
00:45:34 where I would have still a couple hours after that
00:45:36 and sometimes my best hours were at night
00:45:38 when nobody else is around, nobody’s bothering me
00:45:41 but that hole in the day was a problem
00:45:43 so just a couple weeks ago, I made the change
00:45:46 to go ahead and say, all right,
00:45:47 I’m gonna get up a little earlier,
00:45:48 I’m gonna do a walk or get out there first
00:45:51 so I can have more uninterrupted time.
00:45:53 So I’m still playing with factors like this
00:45:55 as I kind of optimize my work efforts
00:45:59 but it’s always been, it was 60 hours a week
00:46:03 for a very long time.
00:46:04 To some degree, I had a little thing in the back of my head
00:46:06 where I was almost jealous of some of the programmers
00:46:08 that would do these marathon sessions
00:46:10 and I had like Dave Taylor, one of the guys that he had,
00:46:13 he would be one of those people
00:46:14 that would fall asleep under his desk sometimes
00:46:16 and all the kind of classic hacker tropes about things
00:46:19 and a part of me was like always a little bothered
00:46:22 that that wasn’t me, that I wouldn’t go program
00:46:25 20 hours straight because I’m falling apart
00:46:28 and not being very effective after 12 hours.
00:46:31 I mean, yeah, 12 hour programming,
00:46:32 that’s fine when you’re doing that
00:46:34 but it never, you’re not doing smart work much after,
00:46:38 at least I’m not, but there’s a range of people.
00:46:40 I mean, that’s something that a lot of people
00:46:41 don’t really get in their gut
00:46:43 where there are people that work on four hours of sleep
00:46:46 and are smart and can continue to do good work
00:46:48 but then there’s a lot of people that just fall apart.
00:46:51 So I do tell people that I always try
00:46:54 to get eight hours of sleep.
00:46:55 It’s not this, you know, push yourself harder,
00:46:58 get up earlier, I just do worse work
00:47:00 where, you know, you can work a hundred hours a week
00:47:04 and still get eight hours of sleep
00:47:05 if you just kind of prioritize things correctly
00:47:08 but I do believe in working hard, working a lot.
00:47:12 There was a comment that a game dev made
00:47:14 that I know there’s a backlash against really hard work
00:47:18 in a lot of cases and I get into online arguments
00:47:21 about this all the time but he was basically saying,
00:47:24 yeah, 40 hours a week, that’s kind of a part time job
00:47:27 and if you were really in it,
00:47:29 you’re doing what you think is important,
00:47:31 what you’re passionate about, working more gets more done
00:47:34 and it’s just really not possible to argue with that
00:47:39 if you’ve been around the people
00:47:40 that work with that level of intensity
00:47:42 and just say, it’s like, no, they should just stop.
00:47:45 And we had, I kind of came back around to that
00:47:49 a couple of years ago where I was using
00:47:51 the fictional example of, all right, some people say,
00:47:55 they’ll say with a straight face, they think, no,
00:47:57 you are less productive if you work
00:47:59 more than 40 hours a week.
00:48:01 And they’re generally misinterpreting things
00:48:03 where your marginal productivity for an hour
00:48:05 after eight hours is less than in one of your peak hours
00:48:08 but you’re not literally getting less done.
00:48:11 There is a point where you start breaking things
00:48:13 and getting worse behavior and everything out of it
00:48:16 where you’re literally going backwards
00:48:18 but it’s not at eight or 10 or 12 hours.
00:48:21 And the fictional example I would use was,
00:48:23 imagine there’s an asteroid coming to impact,
00:48:27 to crash into Earth, destroy all of human life.
00:48:30 Do you want Elon Musk or the people working at SpaceX
00:48:34 that are building the interceptor
00:48:35 that’s going to deflect the asteroid,
00:48:38 do you want them to clock out at five
00:48:40 because damn it, they’re just gonna go do worse work
00:48:42 if they work another couple hours.
00:48:44 And it seems absurd.
00:48:47 And that’s a hypothetical though
00:48:48 and everyone can dismiss that.
00:48:49 But then when coronavirus was hitting
00:48:52 and you have all of these medical personnel
00:48:54 that are clearly pushing themselves really, really hard
00:48:58 and I’d say, it’s like, okay,
00:48:59 do you want all of these scientists working on treatments
00:49:03 and vaccines and caring for all of these people?
00:49:05 Are they really screwing everything up
00:49:07 by working more than eight hours a day?
00:49:09 And of course people say, I’m just an asshole
00:49:11 to say something like that, but it’s the truth.
00:49:15 Working longer gets more done.
00:49:17 Well, so that’s kind of the layer one.
00:49:20 But I’d like to also say that,
00:49:22 at least I believe, depending on the person,
00:49:25 depending on the task, working more and harder
00:49:30 will make you better for the next week
00:49:35 in those peak hours.
00:49:37 So there’s something about a deep dedication to a thing
00:49:42 that kind of gets deep in you.
00:49:44 So the hard work isn’t just about the raw hours
00:49:48 of productivity, it’s the thing it does to you
00:49:54 in the weeks and months after too.
00:49:56 You’re tempering yourself in some ways.
00:49:59 And I think, it’s like Jiro dreams of sushi.
00:50:01 If you really dedicate yourself completely
00:50:03 to making the sushi, to really putting in the long hours,
00:50:08 day after day after day, you become a true craftsman
00:50:12 of the thing you’re doing.
00:50:14 Now there’s, of course, discussions about
00:50:16 are you sacrificing a lot of personal relationships?
00:50:18 Are you sacrificing a lot of other possible things
00:50:21 you could do with that time?
00:50:22 But if you’re talking about purely being a master
00:50:28 or a craftsman of your art,
00:50:30 that more hours isn’t just about doing more,
00:50:34 it’s about becoming better at the thing you’re doing.
00:50:37 Yeah, and I don’t gainsay anybody
00:50:38 that wants to work the minimum amount.
00:50:41 They’ve got other priorities in their life.
00:50:42 My only argument that I’m making,
00:50:44 it’s not that everybody should work hard,
00:50:46 it’s that if you want to accomplish something,
00:50:49 working longer and harder
00:50:50 is the path to getting it accomplished.
00:50:53 Well, let me ask you about this then,
00:50:56 the mythical work life balance.
00:51:01 For an engineer, it seems like
00:51:03 that’s one of the professions for a programmer
00:51:07 where working hard does lead to greater productivity in it.
00:51:12 But it also raises the question of
00:51:17 sort of personal relationships and all that kind of stuff,
00:51:19 family, and how are you able to find work life balance?
00:51:24 Is there advice you can give,
00:51:25 maybe even outside of yourself?
00:51:27 Have you been able to arrive at any wisdom
00:51:29 on this part in your years of life?
00:51:32 I do think that there’s a wide range of people
00:51:34 where different people have different needs.
00:51:36 It’s not a one size fits all.
00:51:38 I am certainly what works for me.
00:51:40 I can tell enough that I’m different
00:51:45 than a typical average person
00:51:46 in the way things impact me,
00:51:48 the things that I want to do,
00:51:50 my goals are different,
00:51:51 and sort of the levers to impact things are different,
00:51:55 where I have literally never felt burnout,
00:51:59 and I know there’s lots of brilliant, smart people
00:52:02 that do world leading work that get burned out,
00:52:05 and it’s never hit me.
00:52:08 I’ve never been at a point where I’m like,
00:52:11 I just don’t care about this.
00:52:13 I don’t want to do this anymore.
00:52:14 But I’ve always had the flexibility
00:52:16 to work on lots of interesting things.
00:52:18 I can always just turn my gaze to something else
00:52:21 and have a great time working on that.
00:52:23 And so much of the ability to actually work hard
00:52:27 is the ability to have multiple things to choose from
00:52:29 and to use your time on the most appropriate thing.
00:52:33 There are time periods where it’s the best time
00:52:36 for me to read a new research paper
00:52:38 that I need to really be thinking hard about it.
00:52:41 Then there’s a time that maybe I should just scan
00:52:43 and organize my old notes
00:52:44 because I’m just not on top of things.
00:52:47 Then there’s the time that,
00:52:48 all right, let’s go bang out a few hundred lines of code
00:52:51 for something.
00:52:52 So switching between them has been real valuable.
00:52:57 So you’ll always have kind of joy in your heart
00:52:59 for all the things you’re doing,
00:53:01 and that is a kind of work life balance
00:53:03 as a first sort of step.
00:53:04 Yeah, I do. So you’re always happy.
00:53:06 I do.
00:53:07 Well, happy.
00:53:08 Yeah, I mean, that’s like a lot of people would say
00:53:10 that often I look like kind of a grim person,
00:53:12 you know, with just sitting there with a neutral expression
00:53:15 or even like knitted brows and a frown on my face
00:53:17 as I’m staring at something.
00:53:18 That’s what happiness looks like for you.
00:53:20 Yeah, it’s kind of true where that it’s like,
00:53:24 okay, I’m pushing through this, I’m making progress here.
00:53:27 I know that doesn’t work for everyone.
00:53:30 I know it doesn’t work for most people.
00:53:33 But what I am always trying to do in those cases
00:53:35 is I don’t want to let somebody
00:53:37 that might be a person like that be told by someone else
00:53:40 that no, don’t even try that out as an option
00:53:44 where work life balance versus kind of your life’s work
00:53:48 where there’s a small subset of the people
00:53:51 that can be very happy being obsessive about things.
00:53:55 And, you know, obsession can often get things done
00:53:58 that just practical, prudent, pedestrian work won’t
00:54:03 or at least won’t for a very long time.
00:54:05 There’s legends of your nutritional intake
00:54:10 in the early days.
00:54:11 What can you say about sort of as a, you know,
00:54:15 being a programmer as a kind of athlete?
00:54:17 So what was the nutrition that fueled?
00:54:21 I have never been that great on really paying attention
00:54:25 to it where I’m good enough that I don’t eat a lot.
00:54:29 You know, I’ve never been like a big heavy guy,
00:54:31 but it was interesting where one of the things
00:54:34 that I can remember being an unhappy teenager,
00:54:36 not having enough money and like,
00:54:38 one of the things that bothered me
00:54:40 about not having enough money
00:54:41 is I couldn’t buy pizza whenever I wanted to.
00:54:43 So I got rich and then I bought a whole lot of pizza.
00:54:46 So that was defining, like, that’s what being rich felt like.
00:54:50 A lot of the little things, like I could buy all the pizza
00:54:52 and comic books and video games that I wanted to.
00:54:55 And it really didn’t take that much,
00:54:58 but the pizza was one of those things
00:55:00 that it’s absolutely true that for a long time
00:55:03 at id Software, I had a pizza delivered every single day.
00:55:06 You know, the delivery guy knew me by name
00:55:09 and I didn’t find out until years later
00:55:11 that apparently I was such a good customer
00:55:13 that they just never raised the price on me.
00:55:15 And I was using this six year old price for the pizzas
00:55:18 that they were still kind of sending my way every day.
00:55:21 So you were doing eating once a day or were you?
00:55:25 It would be spread out.
00:55:26 You know, you have a few pieces of pizza,
00:55:28 you have some more later on
00:55:29 and I’d maybe have something at home.
00:55:31 It was one of the nice things that Facebook Meta
00:55:34 is they feed you quite well.
00:55:36 You get a different, I guess now it’s DoorDash
00:55:39 sorts of things delivered,
00:55:40 but they take care of making sure
00:55:42 that everybody does get well fed.
00:55:44 And I probably had better food those six years
00:55:47 that I was working in the Meta office there
00:55:49 than I used to before.
00:55:51 But it’s worked out okay for me.
00:55:53 My health has always been good.
00:55:55 I get a pretty good amount of exercise
00:55:57 and I don’t eat to excess
00:55:59 and I avoid a lot of other kind of not so good
00:56:02 for you things.
00:56:03 So I’m still doing quite well at my age.
00:56:05 Did you have a kind of, I don’t know,
00:56:10 spiritual experience with food or coffee
00:56:13 or any of that kind of stuff?
00:56:15 I mean, you know, the programming experience,
00:56:17 you know, with music or like I listen to brown noise
00:56:21 on a program or like creating an environment
00:56:24 and the things you take into your body,
00:56:26 just everything you construct
00:56:27 can become a kind of ritual
00:56:29 that empowers the whole process, the programming.
00:56:32 Did you have that relationship with pizza or?
00:56:34 It would really be with Diet Coke.
00:56:35 I mean, there still is that sense of, you know,
00:56:38 drop the can down, crack open the can of Diet Coke.
00:56:40 All right, now I mean business.
00:56:41 We’re getting to work here.
00:56:44 Still to this day?
00:56:45 Yeah.
00:56:45 Is Diet Coke still a part of it?
00:56:46 Yeah, probably eight or nine a day.
00:56:49 Nice.
00:56:50 Okay, what about your setup?
00:56:52 How many screens?
00:56:53 What kind of keyboard?
00:56:55 Is there something interesting?
00:56:56 What kind of IDE, Emacs Vim or something modern?
00:57:02 Linux, what operating system laptop
00:57:04 or any interesting thing that brings you joy?
00:57:06 So I kind of migrated cultures
00:57:09 where early on through all of game dev,
00:57:11 there was sort of one culture there,
00:57:13 which was really quite distinct
00:57:14 from the more the Silicon Valley venture,
00:57:17 you know, culture for things.
00:57:19 They’re different groups
00:57:20 and they have pretty different mores
00:57:21 and the way they think about things where,
00:57:24 and I still do think a lot of the big companies
00:57:26 can learn things from the hardcore game development
00:57:30 side of things where it still boggles my mind
00:57:32 how hostile to debuggers and IDEs that so much of them,
00:57:39 the kind of big money, get billions of dollars,
00:57:41 Silicon Valley venture backed funds are.
00:57:44 Oh, that’s interesting.
00:57:45 Sorry, so you’re saying like big companies
00:57:48 like Google and Meta are hostile to?
00:57:50 They are not big on debuggers and IDEs.
00:57:52 Like so much of it is like Emacs, Vim for things.
00:57:55 And we just assume that debuggers don’t work
00:57:58 most of the time for the systems.
00:58:01 And a lot of this comes from a sort of Linux bias
00:58:03 on a lot of things where I did come up
00:58:06 through the personal computers and then the DOS
00:58:09 and then Windows and it was Borland tools
00:58:14 and then Visual Studio and.
00:58:16 Do you appreciate the buggers?
00:58:18 Very much so.
00:58:19 I mean, a debugger is how you get a view into a system
00:58:22 that’s too complicated to understand.
00:58:23 I mean, anybody that thinks just read the code
00:58:25 and think about it.
00:58:26 That’s an insane statement in the,
00:58:28 you can’t even read all the code on a big system.
00:58:31 You have to do experiments on the system
00:58:34 and doing that by adding log statements,
00:58:36 recompiling and rerunning it
00:58:39 is an incredibly inefficient way of doing it.
00:58:41 I mean, yes, you can always get things done
00:58:43 even if you’re working with stone knives and bear skins.
00:58:46 That is the mark of a good programmer
00:58:48 is that given any tools,
00:58:50 you will figure out a way to get it done,
00:58:52 but it’s amazing what you can do
00:58:54 with sometimes much, much better tools
00:58:57 where instead of just going through this iterative
00:58:59 compile run debug cycle,
00:59:02 you have the old Lisp direction of like,
00:59:05 you’ve got a REPL and you’re working interactively
00:59:07 and doing amazing things there.
00:59:08 But in many cases, a debugger
00:59:10 as a very powerful user interface that can stop,
00:59:13 examine all the different things in your program,
00:59:15 set all of these different breakpoints.
00:59:16 And of course you can do that with GDB or whatever there,
00:59:20 but this is one of the user interface
00:59:22 fundamental principles
00:59:23 where when something is complicated to do,
00:59:26 you won’t use it very often.
00:59:28 There’s people that will break out GDB
00:59:30 when they’re at their wits end
00:59:31 and they just have beat their head
00:59:33 against a problem for so long.
00:59:35 But for somebody that kind of grew up in game dev,
00:59:37 it’s like they were running into the debugger anyways
00:59:40 before they even knew there was a problem.
00:59:42 And you would just stop and see what was happening.
00:59:44 And sometimes you could fix things
00:59:46 even before you did one compile cycle.
00:59:50 You could be in the debugger and you would say,
00:59:52 well, I’m just going to change this right here.
00:59:54 And yep, that did the job and fix it and go on.
00:59:57 And for people who don’t know,
00:59:57 GDB is a sort of popular, I guess, Linux debugger
01:00:03 primarily for C++.
01:00:05 They handle most of the languages,
01:00:06 but it’s based on C as the original kind of Unix heritage.
01:00:10 And it’s kind of like command line.
01:00:11 It’s not user friendly.
01:00:13 It doesn’t allow for clean visualizations
01:00:15 and you’re exactly right.
01:00:17 So that you’re using this kind of debugger
01:00:18 usually when you’re at what’s end
01:00:21 and there’s a problem that you can’t figure out why
01:00:23 by just looking at the codes, you have to find it.
01:00:25 That’s how I guess normal programmers use it.
01:00:28 But you’re saying there should be tools
01:00:30 that kind of visualize and help you
01:00:32 as part of the programming process,
01:00:35 just the normal programming process
01:00:37 to understand the code deeper.
01:00:39 Yeah, when I’m working on like my C, C++ code,
01:00:42 I’m always running it from the debugger.
01:00:45 Just I type in the code, I run it many times.
01:00:48 The first thing I do after writing code
01:00:50 is set a break point and step through the function.
01:00:53 Now other people will say, it’s like,
01:00:54 oh, I do that in my head.
01:00:55 Well, your head is a faulty interpreter
01:00:57 of all those things there.
01:00:59 And I’ve written brand new code.
01:01:01 I wanna step in there
01:01:01 and I’m gonna single step through that,
01:01:03 examine lots of things
01:01:04 and see if it’s actually doing what I expected it to.
01:01:08 It is a kind of companion, the debugger.
01:01:12 Like you’re now coding in an interactive way
01:01:14 with another being.
01:01:17 The debugger is a kind of dumb being,
01:01:19 but it’s a reliable being.
01:01:21 That is an interesting question
01:01:22 of what role does AI play in that kind of,
01:01:25 with codecs and these kind of ability to generate code,
01:01:29 that might be, you might start having tools
01:01:32 that understand the code in interesting deep ways
01:01:35 that can work with you.
01:01:37 There’s a whole spectrum there
01:01:39 from static code analyzers and various kind of dynamic tools
01:01:43 there up to AI that can conceivably grok these programs
01:01:46 that literally no human can understand.
01:01:49 They’re too big, too intertwined and too interconnected,
01:01:51 but it’s not beyond the possibility of understanding.
01:01:54 It’s just beyond what we can hold in our heads
01:01:57 as kind of mutable state while we’re working on things.
01:02:00 And I’m a big proponent, again,
01:02:03 of things like static analyzers and some of that stuff
01:02:06 where you’ll find some people that don’t like being scolded
01:02:10 by a program for how they’ve written something
01:02:12 where it’s like, oh, I know better.
01:02:14 And sometimes you do, but that was something
01:02:16 that it was very, very valuable for me
01:02:20 when, and not too many people get an opportunity
01:02:23 like this to have.
01:02:24 This is almost one of those spiritual experiences
01:02:26 as a programmer, an awakening to,
01:02:28 the id software code bases
01:02:30 were a couple of million lines of code.
01:02:32 And at one point I had used a few
01:02:34 of the different analysis tools,
01:02:36 but I made a point to really go through
01:02:39 and scrub the code base using every tool that I could find.
01:02:43 And it was eyeopening where we had a reputation
01:02:45 for having some of the most robust, strongest code,
01:02:48 where there were some great things
01:02:50 that I remember hearing from Microsoft
01:02:52 telling us about crashes on Xbox.
01:02:54 And we had this tiny number that they said
01:02:56 were probably literally hardware errors.
01:02:59 And then you have other significant titles
01:03:01 that just have millions of faults
01:03:03 that are getting recorded all the time.
01:03:04 So I was proud of our code on a lot of levels,
01:03:07 but when I took this code analysis squeegee
01:03:10 through everything, it was shocking
01:03:14 how many errors there were in there.
01:03:16 Things that you can say, okay, this was a copy paste,
01:03:19 not changing something right here.
01:03:21 Lots of things that were, the most common problem
01:03:25 was something in a printf format string
01:03:27 that was the wrong data type that could cause crashes there.
01:03:30 And you really want the warnings for things like that.
01:03:33 Then the next most common was missing a check for null
01:03:36 that could actually happen, that could blow things up.
01:03:38 And those are obviously like top C, C++ things.
01:03:41 Everybody has those problems.
01:03:43 But the long tail of all of the different little things
01:03:46 that could go wrong there, and we had good programmers
01:03:49 and my own code, stuff that I’d be looking at,
01:03:51 it’s like, oh, I wrote that code, that’s definitely wrong.
01:03:53 We’ve been using this for a year
01:03:56 and it’s this submarine, this mine sitting there
01:03:59 waiting for us to step on.
01:04:01 And it was humbling.
01:04:03 It was, and I reached the conclusion
01:04:05 that anything that can be syntactically allowed
01:04:09 in your language, if it’s gonna show up eventually
01:04:13 in a large enough code base,
01:04:15 you’re not gonna, good intentions
01:04:16 aren’t going to keep it from happening.
01:04:18 You need automated tools and guardrails for things.
01:04:21 And those start with things like static types
01:04:24 and or even type hints in the more dynamic languages.
01:04:26 But the people that rebel against that,
01:04:30 that basically say that slows me down doing that.
01:04:33 There’s something to that, I get that.
01:04:35 I’ve written, I’ve cobbled things together in a notebook.
01:04:38 I’m like, wow, this is great that it just happened,
01:04:41 but yeah, that’s kind of sketchy,
01:04:42 but it’s working fine, I don’t care.
01:04:44 It does come back to that value analysis
01:04:47 where sometimes it’s right to not care.
01:04:49 But when you do care, if it’s going to be something
01:04:52 that’s going to live for years
01:04:53 and it’s gonna have other people working on it,
01:04:56 and it’s gonna be deployed to millions of people,
01:04:59 then you wanna use all of these tools.
01:05:01 You wanna be told it’s like, no,
01:05:02 you’ve screwed up here, here and here.
01:05:04 And that does require kind of an ego check about things
01:05:08 where you have to be open to the fact
01:05:11 that everything that you’re doing
01:05:12 is just littered with flaws.
01:05:14 It’s not that, oh, you occasionally have a bad day.
01:05:16 It’s just whatever stream of code you output,
01:05:19 there is going to be a statistical regularity of things
01:05:21 that you just make mistakes on.
01:05:24 And I do think there’s the whole argument
01:05:27 about test driven design and unit testing
01:05:29 versus kind of analysis and different things.
01:05:33 I am more in favor of the analysis and the stuff
01:05:36 that just like you can’t run your program
01:05:37 until you fix this rather than you can run it
01:05:40 and hopefully a unit test will catch it in some way.
01:05:42 Yeah, in my private code, I have asserts everywhere.
01:05:46 Yeah.
01:05:48 There’s something pleasant to me, pleasurable to me
01:05:52 about sort of the dictatorial rule of like,
01:05:55 this should be true at this point.
01:05:58 And too many times I’ve made mistakes
01:06:03 that shouldn’t have been made.
01:06:05 And I would assume I wouldn’t be the kind of person
01:06:08 that would make that mistake,
01:06:09 but I keep making that mistake.
01:06:10 Therefore, an assert really catches me,
01:06:13 really helps all the time.
01:06:15 So my code, I would say like 10 to 20% of my private code
01:06:19 just for personal use is probably asserts.
01:06:21 And they’re active comments.
01:06:22 That’s one of those things that in theory,
01:06:24 they don’t make any difference to the program.
01:06:27 And if it was all operating the way you expected
01:06:29 it would be, then they will never fire.
01:06:32 But even if you have it right
01:06:34 and you wrote the code right initially,
01:06:36 then circumstances change.
01:06:38 The world outside your program changes.
01:06:40 And in fact, that’s one of the things
01:06:42 where I’m kind of fond in a lot of cases
01:06:45 of static array size declarations,
01:06:47 where I went through this period where it’s like,
01:06:49 okay, now we have general collection classes.
01:06:51 We should just make everything variable
01:06:54 because I had this history of in the early days,
01:06:57 you get doom, which had some fixed limits on it.
01:07:00 Then everybody started making crazier and crazier things.
01:07:02 And they kept bumping up the different limits,
01:07:04 this many lines, this many sectors.
01:07:06 And it seemed like a good idea,
01:07:09 well, we should just make this completely generic.
01:07:10 It can go kind of go up to whatever.
01:07:13 And there’s cases where that’s the right thing to do,
01:07:17 but it also, the other aspect of the world
01:07:19 changing around you is it’s good to be informed
01:07:23 when the world has changed more than you thought it would.
01:07:25 And if you’ve got a continuously growing collection,
01:07:28 you’re never gonna find out.
01:07:29 You might have this quadratic slow down on something
01:07:32 where you thought, oh, I’m only ever gonna have
01:07:34 a handful of these, but something changes
01:07:36 and there’s a new design style.
01:07:38 And all of a sudden you’ve got 10,000 of them.
01:07:40 So I kind of like in many cases, picking a number,
01:07:44 some nice brown power of two number
01:07:47 and setting it up in there and having an assert saying,
01:07:49 it’s like, hey, you hit this limit.
01:07:52 You should probably think are the choices
01:07:54 that you’ve made around all of this still relevant
01:07:56 if somebody’s using 10 times more
01:07:59 than you thought they would.
01:08:00 Yeah, this code was originally written
01:08:02 with this kind of worldview,
01:08:04 with this kind of set of constraints.
01:08:06 You were thinking of the world in this way.
01:08:09 If something breaks, that means you gotta rethink
01:08:11 the initial stuff.
01:08:13 And it’s nice for it to do that.
01:08:16 Is there any stuff like a keyboard or monitors?
01:08:21 I’m a fairly pedestrian on a lot of that
01:08:23 where I did move to triple monitors
01:08:26 like in the last several years ago.
01:08:27 I had been dual monitor for a very long time.
01:08:30 And it was one of those things where probably years later
01:08:34 than I should have, I’m just like, well,
01:08:36 the video cards now generally have three output ports.
01:08:38 I should just put the third monitor up there.
01:08:40 That’s been a pure win.
01:08:41 I’ve been very happy with that.
01:08:44 But no, I don’t have fancy keyboard or mouse
01:08:47 or anything really going on.
01:08:48 The key things is an IDE that has helpful debuggers,
01:08:52 has helpful tools.
01:08:54 So it’s not the Emacs Vim route and then Dicode.
01:08:56 Yeah, so I did spend,
01:08:58 I spent one of my week long retreats where I’m like,
01:09:01 okay, I’m gonna make myself use,
01:09:04 it was actually classic VI,
01:09:05 which I know people will say you should never have done that.
01:09:07 You should have just used Vim directly.
01:09:09 But I gave it the good try.
01:09:11 It’s like, okay, I’m being in kind of classic
01:09:13 Unix developer mode here.
01:09:15 And I worked for a week on it.
01:09:18 I used Anki to like teach myself
01:09:20 the different little key combinations
01:09:22 for things like that.
01:09:23 And in the end, it was just like, all right,
01:09:26 this was kind of like my civil war reenactment phase.
01:09:28 It’s like, I’m going out there doing it
01:09:30 like they used to in the old days.
01:09:31 And it was kind of fun in that regard.
01:09:34 So many people right now.
01:09:35 They’re screaming as they’re listening to this.
01:09:38 So again, the out is that this was not modern Vim,
01:09:41 but still, yes, I was very happy to get back
01:09:44 to my visual studio at the end.
01:09:46 Yeah, I’m actually, I struggle with this a lot
01:09:49 because I use a Kinesis keyboard
01:09:52 and I use Emacs primarily.
01:09:56 And I feel like I can, exactly as you said,
01:09:59 I can understand the code.
01:10:00 I can navigate the code.
01:10:01 There’s a lot of stuff you could build with an Emacs
01:10:03 with using Lisp.
01:10:04 You can customize a lot of things for yourself
01:10:07 to help you introspect the code,
01:10:09 like to help you understand the code
01:10:11 and visualize different aspects of the code.
01:10:13 You can even run debuggers,
01:10:14 but it’s work and the world moves past you
01:10:18 and the better and better ideas are constantly being built.
01:10:21 And that puts a kind of,
01:10:25 I need to take the same kind of retreat
01:10:27 as you’re talking about,
01:10:28 but now I’m still fighting the civil war.
01:10:30 I need to kind of move into the 21st century.
01:10:33 And it does seem like the world is,
01:10:34 or a large chunk of the world is moving
01:10:37 towards visual studio code,
01:10:38 which is kind of interesting to me.
01:10:40 Again, it’s the JavaScript ecosystem on the one hand
01:10:43 and IDs are one of those things
01:10:45 that you want to be infinitely fast.
01:10:47 You want them to just kind of immediately respond.
01:10:50 And like, I mean, heck, I’ve got,
01:10:52 there’s someone I know,
01:10:53 I’m an old school game dev guy
01:10:55 that still uses visual studio six.
01:10:57 And on a modern computer,
01:10:59 everything is just absolutely instant
01:11:01 on something like that,
01:11:02 because it was made to work on a computer
01:11:04 that’s 10,000 or a hundred thousand times slower.
01:11:07 So just everything happens immediately.
01:11:10 And all the modern systems just feel,
01:11:13 they feel so crufty when it’s like,
01:11:15 oh, why is this refreshing the screen
01:11:17 and moving around and updating over here?
01:11:19 And something blinks down there
01:11:20 and you should update this.
01:11:21 And there are things that we’ve lost
01:11:25 with that incredible flexibility,
01:11:27 but lots of people get tons of value from it.
01:11:31 And I am super happy that that seems to be winning
01:11:33 over even a lot of the old Vim and Emacs people
01:11:36 that they’re kind of like,
01:11:37 hey, visual studio codes, maybe not so bad.
01:11:40 I am, that may be the final peacekeeping solution
01:11:43 where everybody is reasonably happy
01:11:45 with something like that.
01:11:47 So can you explain what a dot plan file is
01:11:50 and what role that played in your life?
01:11:53 Does it still continue to play a role?
01:11:55 Back in the early, early days of id Software,
01:11:58 one of our big things that was unique with what we did
01:12:01 is I had adopted next stations
01:12:04 or kind of next step systems
01:12:05 from Steve Jobs’s out in the woods away from Apple company.
01:12:11 And they were basically, it was kind of interesting
01:12:15 because I did not really have a background
01:12:17 with the Unix system.
01:12:18 So many of the people they get immersed in that in college
01:12:21 and that sets a lot of cultural expectations for them.
01:12:27 And I didn’t have any of that,
01:12:29 but I knew that my background was,
01:12:31 I was a huge Apple II fan boy.
01:12:34 I was always a little suspicious of the Mac.
01:12:36 I was not really what kind of I wanted to go with.
01:12:41 But when Steve Jobs left Apple and started Next,
01:12:44 this computer did just seem like one of those amazing things
01:12:46 from the future where it had all of this cool stuff in it.
01:12:50 And we were still back in those days working on DOS,
01:12:53 everything blew up.
01:12:54 You had reset buttons
01:12:55 because your computer would just freeze.
01:12:56 If you’re doing development work,
01:12:58 literally dozens of times a day,
01:12:59 your computer was just rebooting constantly.
01:13:02 And so this idea of, yes, any of the Unix workstations
01:13:06 would have given a stable development platform
01:13:08 where you don’t crash and reboot all the time.
01:13:11 But Next also had this really amazing graphical interface
01:13:15 and it was great for building tools.
01:13:17 And it used Objective C as kind of an interesting dead end
01:13:22 for things like that.
01:13:23 So Next was Unix based, it said Objective C.
01:13:26 So it has a lot of the elements.
01:13:27 That became Mac.
01:13:28 I mean, the kind of reverse acquisition of Apple by Next
01:13:31 where that took over
01:13:32 and became what the modern Mac system is.
01:13:35 And to find some of the developer,
01:13:38 like the tools and the whole community.
01:13:41 Yeah, you’ve still got,
01:13:42 if you’re programming on Apple stuff now,
01:13:43 there’s still all these NS somethings,
01:13:45 which was originally Next step objects
01:13:47 of different kinds of things.
01:13:49 But one of the aspects of those Unix systems
01:13:52 was they had this notion of a.plan file
01:13:56 where a.file is an invisible file.
01:14:00 Usually in your home directory or something.
01:14:02 And there was a trivial server running
01:14:03 on most Unix systems at the time
01:14:05 that when somebody ran a trivial little command
01:14:09 called finger, you could do finger
01:14:11 and then somebody’s address,
01:14:13 it could be anywhere on the internet
01:14:15 if you were connected correctly.
01:14:16 Then all that server would do was read the.plan file
01:14:20 in that user’s home directory
01:14:22 and then just spit it out to you.
01:14:24 And originally the idea was that could be
01:14:27 whether you’re on vacation, what your current project was.
01:14:30 It’s supposed to be like the plan of what you’re doing.
01:14:32 And people would use it for various purposes,
01:14:35 but all it did was dump that file over
01:14:39 to the terminal of whoever issued the finger command.
01:14:42 And at one point I started just keeping a list
01:14:46 of what I was doing in there,
01:14:48 which would be what I was working on in the day.
01:14:51 And I would have this little syntax.
01:14:53 I kind of got to myself about,
01:14:55 here’s something that I’m working on.
01:14:57 I put a star when I finish it.
01:14:58 I could have a few other little bits of punctuation.
01:15:01 And at the time it was,
01:15:02 it started off as being just like my to do list.
01:15:05 And it would be these trivial, obscure little things
01:15:08 like I fixed something with collision detection code,
01:15:12 made Fireball do something different
01:15:14 and just little one liners that people
01:15:16 that were following the games could kind of decipher.
01:15:20 But I did wind up starting to write much more in depth things.
01:15:24 I would have little notes of thoughts and insights
01:15:28 and then I would eventually start having little essays.
01:15:30 I would sometimes dump into the dot plan files
01:15:33 interspersed with the work logs of things that I was doing.
01:15:36 So in some ways it was like a super early proto blog
01:15:39 where I was just kind of dumping out what I was working on.
01:15:42 But it was interesting enough that there were a lot
01:15:45 of people that were interested in this.
01:15:48 So most of the people didn’t have Unix workstation.
01:15:51 So there were the websites back in the day
01:15:53 that would follow the Doom and Quake development
01:15:55 that would basically make a little service
01:15:58 that would go grab all the changes
01:15:59 and then people could just get it with a web browser.
01:16:02 And there was a period where like all of the little
01:16:04 kind of Dallas gaming diaspora of people
01:16:07 that were at all in that orbit,
01:16:09 there were a couple dozen plan files going on,
01:16:12 which was, and this was some years before blogging
01:16:15 really became kind of a thing.
01:16:17 And it was kind of a premonition
01:16:20 of sort of the way things would go.
01:16:22 And there was, it’s all been collected.
01:16:25 It’s available online in different places.
01:16:27 And it’s kind of fun to go back and look through
01:16:29 what I was thinking, what I was doing in the different areas.
01:16:32 Have you had a chance to look back?
01:16:33 Is there some interesting, very low level specific
01:16:37 to do items, maybe things you’ve never completed,
01:16:40 all that kind of stuff and high level philosophical essay
01:16:44 type of stuff that stands out?
01:16:46 Yeah, there’s some good stuff on both
01:16:48 where a lot of it was low level nitpicky details
01:16:51 about game dev and I’ve learned enough things
01:16:55 where there’s no project that I worked on
01:16:58 that I couldn’t go back and do a better job on now.
01:17:00 I mean, you just, you learn things,
01:17:02 hopefully if you’re doing it right,
01:17:03 you learn things as you get older
01:17:05 and you should be able to do a better job
01:17:07 at all of the early things.
01:17:08 And there’s stuff in Wolfenstein Doomquake that’s like,
01:17:12 oh, clearly I could go back and do a better job at this,
01:17:15 whether it’s something in the rendering engine side
01:17:17 or how I implemented the monster behaviors
01:17:20 or managed resources or anything like that.
01:17:22 Do you see the flaws in your thinking now?
01:17:24 Like looking back?
01:17:26 Yeah, I do.
01:17:27 I mean, sometimes I’ll get the,
01:17:29 I’ll look at it and say, yeah, I had a pretty clear view
01:17:31 of I was doing good work there.
01:17:33 And I haven’t really hit the point
01:17:35 where there was another programmer, Graham Devine,
01:17:38 who was, he had worked at ID and Seventh Guest
01:17:41 and he made some comment one time
01:17:43 where he said he looked back at some of his old notes
01:17:45 and he was like, wow, I was really smart back then.
01:17:48 And I don’t hit that so much where,
01:17:52 I mean, I look at it and I always know that,
01:17:54 yeah, there’s all the, you know, with aging,
01:17:56 you get certain changes in how you’re able to work problems,
01:18:00 but all of the problems that I’ve worked,
01:18:02 I’m sure that I could do a better job on all of them.
01:18:06 Oh, wow.
01:18:07 So you can still step right in.
01:18:08 If you could travel back in time and talk to that guy,
01:18:10 you would teach him a few things.
01:18:12 Yeah, absolutely.
01:18:14 That’s awesome.
01:18:15 What about the high level philosophical stuff?
01:18:18 Is there some insights that stand out that you remember?
01:18:20 There’s things that I was understanding about development
01:18:25 and, you know, in the industry and so on
01:18:28 that were in a more primitive stage
01:18:31 where I definitely learned a lot more in the later years
01:18:36 about business and organization and team structure.
01:18:41 There were, I mean, there were definitely things
01:18:44 that I was not the best person
01:18:46 or even a very good person about managing,
01:18:48 like how a team should operate internally,
01:18:51 how people should work together.
01:18:53 I was just, you know, just get out of my way
01:18:57 and let me work on the code and do this.
01:18:59 And more and more, I’ve learned how,
01:19:03 in the larger scheme of things,
01:19:04 how sometimes relatively unimportant
01:19:07 some of those things are,
01:19:08 where it is this user value generation
01:19:11 that’s the overarching importance for all of that.
01:19:14 And I didn’t necessarily have my eye on that ball correctly
01:19:17 through a lot of my, you know, my earlier years.
01:19:21 And there’s things that, you know,
01:19:23 I could have gotten more out of people
01:19:25 handling things in different ways.
01:19:27 I am, you know, I could have made, you know,
01:19:29 in some ways more successful products
01:19:32 by following things in different ways.
01:19:33 There’s mistakes that we made
01:19:35 that we couldn’t really have known
01:19:37 how things would have worked out,
01:19:38 but it was interesting to see in later years companies
01:19:41 like Activision showing that,
01:19:42 hey, you really can just do the same game,
01:19:44 make it better every year.
01:19:46 And you can look at that from a negative standpoint
01:19:48 and say, it’s like, oh, that’s just being derivative
01:19:50 and all that.
01:19:51 But if you step back again and say, it’s like, no,
01:19:53 are the people buying it still enjoying it?
01:19:55 Are they enjoying it more
01:19:56 than what they might have bought otherwise?
01:19:59 And you can say, no, that’s actually
01:20:00 a great value creation engine to do that
01:20:03 if you’re in a position where you can.
01:20:05 I, you know, don’t be forced into reinventing everything
01:20:09 just because you think that you need to.
01:20:12 You know, lots of things about business and team stuff
01:20:15 that could be done better, but the technical work,
01:20:18 the kind of technical visionary type stuff
01:20:20 that I laid out, I still feel pretty good about.
01:20:23 There are some classic old ones about my defending
01:20:26 of OpenGL versus D3D,
01:20:29 which turned out to be one of the more
01:20:32 probably important momentous things there
01:20:34 where it never, it was always a rear guard action on Windows
01:20:38 where Microsoft was just not gonna let that win.
01:20:42 But when I look back on it now,
01:20:44 that fight to keep OpenGL relevant
01:20:46 for a number of years there
01:20:48 meant that OpenGL was there when mobile started happening
01:20:52 and OpenGL ES was the thing that drove
01:20:55 all of the acceleration of the mobile industry.
01:20:58 And it’s really only in the last few years
01:21:00 as Apple’s moved to Metal
01:21:02 and some of the other companies have moved to Vulkan
01:21:04 that that’s moved away.
01:21:06 But really stepping back and looking at it,
01:21:09 it’s like, yeah, I sold tens of millions of games
01:21:11 for different things.
01:21:13 But billions and billions of devices wound up
01:21:17 with an appropriate capable graphics API
01:21:21 due in no small part to me thinking
01:21:23 that that was really important
01:21:25 that we not just give up and use Microsoft’s
01:21:30 at that time really terrible API.
01:21:32 The thing about Microsoft is the APIs don’t stay terrible.
01:21:35 They were terrible at the start, but a few versions on,
01:21:38 they were actually quite good.
01:21:40 And there was a completely fair argument to be made
01:21:42 that by the time DX9 was out,
01:21:45 it was probably a better programming environment
01:21:47 than OpenGL, but it was still a wonderful good thing
01:21:51 that we had an open standard
01:21:52 that could show up on Linux and Android and iOS
01:21:55 and eventually WebGL still to this day.
01:21:58 So that was one, that would be on my greatest hits list
01:22:02 of things that I kind of pushed with.
01:22:04 The impact it had on billions of devices, yes.
01:22:07 So let’s talk about it.
01:22:08 Can you tell the origin story of id Software?
01:22:12 Again, one of the greatest game developer companies ever.
01:22:16 It created Wolfenstein 3D games
01:22:18 that defined my life also in many ways
01:22:22 as a thing that made me realize
01:22:24 what computers are capable of in terms of graphics,
01:22:26 in terms of performance, it just unlocked something deep
01:22:31 in me and understanding what these machines are all about
01:22:34 as games can do that.
01:22:35 So Wolfenstein 3D, Doom, Quake,
01:22:37 and just all the incredible engineering innovation
01:22:40 that went into that.
01:22:41 So how did it all start?
01:22:44 So I’ll caveat upfront
01:22:45 that I usually don’t consider myself the historian
01:22:49 of the software side of things.
01:22:51 I usually do kind of point people at John Romero
01:22:55 for stories about the early days where I’ve never been,
01:23:00 like I’ve commented that I’m a remarkably
01:23:01 unsentimental person in some ways
01:23:03 where I don’t really spend a lot of time
01:23:05 unless I’m explicitly prodded to go back
01:23:07 and think about the early days of things.
01:23:10 And I didn’t necessarily make the effort
01:23:14 to archive everything exactly in my brain.
01:23:17 And the more that I work on machine learning and AI
01:23:19 and the aspects of memory
01:23:20 and how when you go back and polish certain things,
01:23:23 it’s not necessarily exactly the way it happened.
01:23:25 But having said all of that, from my view,
01:23:29 the way everything happened that led up to that was
01:23:34 after I was an adult
01:23:36 and kind of taking a few college classes
01:23:38 and deciding to drop out, I was doing,
01:23:40 I was hardscrabble contract programming work,
01:23:43 really struggling to kind of keep groceries
01:23:46 and pay my rent and things.
01:23:48 And the company that I was doing the most work for
01:23:50 was a company called Softdisk Publishing,
01:23:53 which had the sounds bizarre now business model
01:23:57 of monthly subscription software.
01:23:59 Before there was an internet that people could connect to
01:24:02 and get software, you would pay a certain amount
01:24:05 and every month they would send you a disc
01:24:07 that had some random software on it.
01:24:09 And people that were into computers
01:24:11 thought this was kind of cool.
01:24:12 And they had different ones for the Apple II,
01:24:14 the 2GS, the PC, the Mac, the Amiga,
01:24:18 lots of different things here.
01:24:20 So quirky little business,
01:24:21 but I was doing a lot of contract programming for them
01:24:24 where I’d write tiny little games
01:24:26 and sell them for 300, $500.
01:24:29 And one of the things that I was doing again
01:24:32 to keep my head above water here was
01:24:35 I decided that I could make one program
01:24:38 and I could port it to multiple systems.
01:24:41 So I would write a game like Dark Designs or Catacombs
01:24:44 and I would develop it on the Apple II, the 2GS
01:24:47 and the IBM PC, which apparently was the thing
01:24:51 that really kind of piqued the attention
01:24:54 of the people working down there.
01:24:56 Like Jay Wilbur was my primary editor
01:24:58 and Tom Hall was a secondary editor.
01:25:01 And they kept asking me, it’s like,
01:25:02 hey, you should come down and work for us here.
01:25:06 And I pushed it off a couple of times
01:25:08 because I was really enjoying my freedom
01:25:10 of kind of being off on my own.
01:25:11 Even if I was barely getting by, I loved it.
01:25:14 I was doing nothing but programming all day,
01:25:17 but I did have enough close scrapes with like,
01:25:20 damn, I’m just really out of money
01:25:21 that maybe I should get an actual job
01:25:25 rather than contracting these kind of one at a time things.
01:25:27 And Jay Wilbur was great.
01:25:29 He was like FedExing me the checks when I would need them.
01:25:32 I had to kind of get over whatever hump I was at.
01:25:35 So I took the, I finally took them up on their offer
01:25:38 to come down to Shreveport, Louisiana.
01:25:41 I was in Kansas city at the time,
01:25:43 drove down to through the Ozarks and everything
01:25:47 down to Louisiana and saw the soft disk offices,
01:25:51 went through, talked to a bunch of people,
01:25:53 met the people I had been working with remotely
01:25:56 at that time.
01:25:57 But the most important thing for me
01:25:59 was I met two programmers there,
01:26:01 John Romero and Lane Roth that for the first time ever,
01:26:04 I had met programmers that knew more cool stuff than I did
01:26:08 where the world was just different back then.
01:26:11 I was in Kansas city.
01:26:12 It was one of those smartest kid in the school
01:26:14 does all the computer stuff.
01:26:15 The teachers don’t have anything to teach him.
01:26:18 But all I had to learn from was these few books
01:26:20 at the library.
01:26:21 It was not much at all.
01:26:23 And there were some aspects of programming
01:26:25 that were kind of black magic to me.
01:26:27 Like, it’s like, oh, he knows how to format a track on,
01:26:30 I am on a low level drive programming interface.
01:26:34 And this was, I was still not at all sure
01:26:37 I was gonna take the job,
01:26:38 but I met these awesome programmers
01:26:40 that were doing cool stuff.
01:26:42 And Romero had worked at Origin Systems
01:26:44 and he had done like, so many different games ahead of time
01:26:49 that I did kind of quickly decide,
01:26:50 it’s like, yeah, I’ll go take the job down there.
01:26:53 And I settled down there, moved in
01:26:57 and started working on more little projects.
01:26:59 And the first kind of big change that happened down there
01:27:03 was the company wanted to make a gaming focused,
01:27:05 a PC gaming focused subscription.
01:27:08 Just like all their other,
01:27:09 it was the same formula that they used for everything.
01:27:11 Pay a monthly fee and you’ll get a disc
01:27:14 with one or two games just every month
01:27:16 and no choice in what you get, but we think it’ll be fun.
01:27:19 And that was the model they were comfortable with
01:27:21 and they said, all right,
01:27:22 we’re gonna start this gamers edge department.
01:27:25 And all of us that were interested in that,
01:27:27 like me, Romero, Tom Hall was kind of helping us
01:27:31 from his side of things, Jay would peek in
01:27:34 and we had a few other programmers working with us
01:27:37 at the time.
01:27:38 And we were going to just start making games,
01:27:41 just the same model.
01:27:43 And we dived in and it was fantastic.
01:27:45 Do you have to make new games?
01:27:47 Every month.
01:27:48 Every month.
01:27:49 Yeah.
01:27:50 And this, in retrospect, looking back at it,
01:27:52 that sense that I had done all this contract programming
01:27:55 and John Romero had done like far more of this
01:27:58 where he had done one of his teaching himself efforts
01:28:00 was he made a game for every letter of the alphabet.
01:28:03 It’s that sense of like,
01:28:04 I’m just gonna go make 26 different games,
01:28:06 give them a different theme.
01:28:07 And you learn so much when you go through
01:28:10 and you crank these things out
01:28:12 like on a biweekly, monthly basis,
01:28:14 something like that. From start to finish.
01:28:16 It’s not like just an idea.
01:28:17 It’s not just from the very beginning to the very end.
01:28:21 It’s done.
01:28:22 It has to be done.
01:28:24 There’s no delaying.
01:28:25 It’s done.
01:28:26 And you’ve got deadlines.
01:28:26 And that kind of rapid iteration pressure cooker environment
01:28:32 was super important for all of us developing the skills
01:28:35 that brought us to where we eventually went to.
01:28:38 I mean, people would say like,
01:28:39 like in the history of the Beatles,
01:28:41 like it wasn’t them being the Beatles.
01:28:43 It was them playing all of these other early works
01:28:46 that that opportunity to craft all of their skills
01:28:48 before they were famous,
01:28:49 that was very critical to their later successes.
01:28:53 And I think there’s a lot of that here
01:28:54 where we did these games that nobody remembers,
01:28:58 lots of little things that contributed
01:29:00 to building up the skillset for the things
01:29:02 that eventually did make us famous.
01:29:05 Yeah, Dostoevsky wrote The Gambler.
01:29:08 I had to write it in a month, just to make money.
01:29:12 And nobody remembers that probably
01:29:14 because he had to figure out,
01:29:15 because it’s literally,
01:29:17 he didn’t have enough time to write it fast enough.
01:29:21 So he had to come up with hacks.
01:29:23 Literally,
01:29:23 it again comes down to that point
01:29:25 where pressure and limitation of resources
01:29:28 is surprisingly important.
01:29:30 And it’s counterintuitive in a lot of ways
01:29:32 where you just think that if you’ve got all the time
01:29:34 in the world and you’ve got all the resources in the world,
01:29:36 of course, you’re gonna get something better.
01:29:38 But sometimes it really does work out
01:29:40 that the innovations, mother necessity,
01:29:44 where you can, or resource constraints,
01:29:46 and you have to do things when you don’t have a choice.
01:29:49 It’s surprising what you can do.
01:29:50 Is there any good games written in that time?
01:29:52 Would you say?
01:29:53 Some of them are still fun to go back and play
01:29:55 where you get the,
01:29:58 they were all about kind of,
01:29:59 the more modern term is game feel,
01:30:01 about how just the exact feel that things,
01:30:04 it’s not the grand strategy of the design,
01:30:06 but how running and jumping and shooting
01:30:08 and those things I feel in the moment.
01:30:11 And some of those are still,
01:30:13 if you sat down on them, you kind of go,
01:30:15 it’s a little bit different.
01:30:16 It doesn’t have the same movement feel,
01:30:17 but you move over and you’re like, bang, jump, bang.
01:30:20 It’s like, hey, that’s kind of cool still.
01:30:23 So you can get lost in the rhythm of the game.
01:30:25 Is that what you mean by feel?
01:30:27 Just like there’s something about it that pulls you in?
01:30:31 Nowadays, again, people talk about compulsion loops
01:30:33 and things where it’s that sense of exactly
01:30:37 what you’re doing, what your fingers are doing
01:30:38 on the keyboard, what your eyes are seeing,
01:30:41 and there are gonna be these sequences of things.
01:30:43 Grab the loot, shoot the monster,
01:30:44 jump over the obstacle, get to the end of the level.
01:30:47 These are eternal aspects of game design in a lot of ways,
01:30:50 but there are better and worse ways to do all of them.
01:30:53 And we did so many of these games
01:30:55 that we got a lot of practice with it.
01:30:58 So one of the kind of weird things
01:31:01 that was happening at this time
01:31:02 is John Romero was getting some strange fan mail.
01:31:06 And back in the days, this was before email.
01:31:09 So we literally got letters sometimes and telling him,
01:31:12 it’s like, oh, I wanna talk to you about your games.
01:31:14 I wanna reach out, different things.
01:31:17 And eventually it turned out
01:31:19 that these were all coming from Scott Miller
01:31:21 at Apogee Software.
01:31:23 And he was reaching out through,
01:31:26 he didn’t think he could contact John directly,
01:31:28 that he would get intercepted.
01:31:29 So he was trying to get him to contact him
01:31:31 through like back channel fan mail,
01:31:34 because he basically was saying,
01:31:35 hey, I’m making all this money on shareware games.
01:31:38 I want you to make shareware games,
01:31:40 because he had seen some of the games
01:31:42 that Romero had done.
01:31:44 And we looked at Scott Miller’s games
01:31:47 and we didn’t think they were very good.
01:31:50 We’re like, that can’t be making the kind of money
01:31:53 that he’s saying he’s making 10 grand or something
01:31:56 off of this game.
01:31:57 And we really thought that he was full of shit,
01:31:59 that it was a lie trying to get him into this.
01:32:03 But so that was kind of going on at one level.
01:32:07 And it was funny the moment when Romero realized
01:32:10 that he had some of these letters pinned up on his wall,
01:32:12 like all of his fans.
01:32:13 And then we noticed that they all had the same return
01:32:15 address with different names on them,
01:32:17 which was a little bit of a two edged sword there.
01:32:19 Trying to figure out the puzzle laid out before him.
01:32:23 Yeah, what happened after I kind of coincident with that
01:32:26 was I was working on a lot of the new technologies
01:32:29 where I was now full on the IBM PC for the first time,
01:32:33 where I was really a long hold out on Apple II forever.
01:32:37 And I loved my Apple II.
01:32:38 It was the computer I always wished I had
01:32:40 when I was growing up.
01:32:41 And when I finally did have one,
01:32:43 I was kind of clinging on to that well past it
01:32:46 sort of good use by day.
01:32:47 Is it the best computer ever made you would you say?
01:32:50 I wouldn’t make judgments like that about it,
01:32:53 but it was positioned in such a way,
01:32:54 especially in the school systems that it impacted
01:32:57 a whole lot of American programmers at least,
01:33:00 where there was programs that the Apple IIs
01:33:02 got into the schools and they had enough capability
01:33:05 that lots of interesting things happened with them.
01:33:08 In Europe, it was different.
01:33:09 You had your Amigas and Ataris and Acorns in the UK
01:33:13 and things that had different things.
01:33:16 But in the United States,
01:33:17 it was probably the Apple II made the most impact
01:33:19 for a lot of programmers of my generation.
01:33:23 But so I was really digging into the IBM
01:33:26 and this was even more so with the total focus
01:33:29 because I had moved to another city
01:33:30 where I didn’t know anybody that I wasn’t working with.
01:33:33 I had a little apartment and then at Softdisk,
01:33:36 again, the things that drew me to it,
01:33:38 I had a couple of programmers that knew more than I did
01:33:42 and they had a library.
01:33:43 They had a set of books and a set of magazines.
01:33:46 They had a couple of years of magazines,
01:33:47 the old Dr. Dobbs journal and all of these magazines
01:33:50 that had information about things.
01:33:53 And so I was just in total immersion mode.
01:33:56 It was eat, breathe, sleep, computer programming,
01:33:59 particularly the IBM for everything that I was doing.
01:34:03 And I was digging into a lot of these
01:34:05 low level hardware details
01:34:06 that people weren’t usually paying attention to,
01:34:08 the way the IBM EGA cards worked, which was fun for me.
01:34:13 I hadn’t had experience with things at that level.
01:34:16 And back then you could get hardware documentation
01:34:19 just down at the register levels.
01:34:21 This is where the CRTC register is.
01:34:23 This is how the color registers work
01:34:25 and how the different things are applied.
01:34:27 And they were designed for a certain reason.
01:34:29 They were designed for an application.
01:34:31 They had an intended use in mind,
01:34:33 but I was starting to look at other ways
01:34:36 that they could perhaps be exploited
01:34:38 that they weren’t initially intended for.
01:34:40 Could you comment on like, first of all, what operating system was there?
01:34:44 What instruction set was it?
01:34:45 Like, what are we talking about?
01:34:48 So this was DOS and x86.
01:34:50 So 16 bit, 8086.
01:34:52 The 286s were there and 386s existed.
01:34:55 They were rare.
01:34:56 We had a couple for our development systems,
01:34:59 but we were still targeting the more broad.
01:35:02 It was all DOS 16 bit.
01:35:04 None of this was kind of DOS extenders and things.
01:35:07 How different is it from the systems of today?
01:35:09 Is it kind of a precursor that’s similar?
01:35:12 Very little.
01:35:13 If you open up command.exe or com on Windows,
01:35:17 you see some of the remnants of all of that,
01:35:19 but it was a different world.
01:35:21 It was the 640K is enough world, and nothing was protected.
01:35:25 It crashed all the time.
01:35:26 You had TSRs or terminate and stay resident hacks
01:35:29 on top of things that would cause configuration problems.
01:35:33 All the hardware was manually configured in your auto exec.
01:35:37 So it was a very different world.
01:35:39 But the code is still the same, similar.
01:35:41 You could still write it.
01:35:42 My earliest code there was written in Pascal.
01:35:44 That was what I had learned at an earlier point.
01:35:47 So between BASIC and C++, there was Pascal.
01:35:51 So when BASIC assembly language, some of my intermediate stuff
01:35:55 was, well, you had to for performance.
01:35:57 BASIC was just too slow.
01:35:58 So most of the work that I was doing
01:36:00 as a contract programmer in my teenage years
01:36:02 was assembly language.
01:36:05 Wait, you wrote games in assembly?
01:36:07 Yeah, complete games in assembly language.
01:36:10 And it’s thousands and thousands of lines
01:36:12 of three letter acronyms for the instructions.
01:36:16 You don’t earn the once again greatest programmer ever label
01:36:21 without being able to write a game in assembly.
01:36:23 OK.
01:36:23 That’s good.
01:36:24 Everybody serious wrote their games in assembly language.
01:36:27 It was kind of a.
01:36:28 Everybody serious.
01:36:28 See what he said?
01:36:29 Everybody serious.
01:36:31 It was an outlier to use Pascal a little bit,
01:36:33 where there was one famous program called wizardry.
01:36:36 It was one of the great early role playing games
01:36:39 that was written in Pascal.
01:36:40 But it was almost nothing used Pascal there.
01:36:43 But I did learn Pascal.
01:36:45 And I remember doing all of my.
01:36:47 To this day, I sketch in data structures.
01:36:49 When I’m thinking about something, I’ll open up a file
01:36:52 and I’ll start writing struct definitions for how
01:36:55 data is going to be laid out.
01:36:56 And Pascal was kind of formative to that
01:36:58 because I remember designing my RPGs in Pascal record
01:37:02 structures and things like that.
01:37:04 And so I had gotten the Pascal compiler for the Apple 2GS
01:37:07 that I could work on.
01:37:08 And the first IBM game that I developed, I did in Pascal.
01:37:12 And that’s actually kind of an interesting story,
01:37:14 again, talking about the constraints and resources
01:37:17 where I had an Apple 2GS.
01:37:19 I didn’t have an IBM PC.
01:37:21 I wanted to port my applications to IBM
01:37:24 because I thought I could make more money on it.
01:37:27 So what I wound up doing is I rented a PC for a week
01:37:31 and bought a copy of Turbo Pascal.
01:37:33 And so I had a hard one week and this was cutting into
01:37:36 what minimal profit margin I had there.
01:37:38 But I had this computer for a week.
01:37:40 I had to get my program ported before I had to return
01:37:43 the PC and that was kind of what the first thing
01:37:46 that I had done on the IBM PC and what led me
01:37:49 to taking the job at Softdisk.
01:37:51 And Turbo Pascal, how’s that different from regular Pascal?
01:37:55 Is it a different compiler or something like that?
01:37:57 It was a product of Borland, which before Microsoft
01:38:00 kind of killed them, they were the hot stuff
01:38:02 developer tools company.
01:38:04 You had Borland, Turbo Pascal and Turbo C
01:38:06 and Prologue, I mean, all the different things.
01:38:09 But what they did was they took a supremely pragmatic
01:38:12 approach of making something useful.
01:38:14 It was one of these great examples where Pascal
01:38:17 was an academic language and you had things like
01:38:20 the UCSDP system that Wizardry was actually written in
01:38:23 that they did manage to make a game with that,
01:38:27 but it was not a super practical system.
01:38:30 While Turbo Pascal was, it was called Turbo
01:38:33 because it was blazingly fast to compile.
01:38:35 I mean, really ridiculously 10 to 20 times faster
01:38:39 than most other compilers at the time.
01:38:41 But it also had very pragmatic access to look,
01:38:44 you can just poke at the hardware in these different ways
01:38:46 and we have libraries that let you do things.
01:38:49 And it was a pretty good, it was a perfectly good way
01:38:51 to write games and this is one of those things
01:38:53 where people have talked about different paths
01:38:56 that computer development could have taken
01:38:58 where C took over the world for reasons
01:39:01 that came out of Unix and eventually Linux.
01:39:05 And that was not a foregone conclusion at all.
01:39:07 And people can make real reasoned rational arguments
01:39:10 that the world might’ve been better
01:39:12 if it had gone a Pascal route.
01:39:14 I’m somewhat agnostic on that where I do know
01:39:17 from experience, it was perfectly good enough to do that.
01:39:21 And it had some fundamental improvements
01:39:23 like it had range checked arrays as an option there,
01:39:26 which could avoid many of C’s real hazards
01:39:29 that happened in a security space.
01:39:31 But C1, they were basically offering it
01:39:33 about the same level of abstraction.
01:39:35 It was a systems programming language, but.
01:39:38 But you said Pascal had a more emphasis on data structures.
01:39:41 I actually, in the tree of languages,
01:39:44 did Pascal come before C?
01:39:48 They were pretty contemporaneous.
01:39:49 So Pascal’s lineage went to Modula 2 and eventually Oberon,
01:39:53 which was another Nicholas word, kind of experimental language.
01:39:58 But they were all good enough at that level.
01:40:01 Now, some of the classic academic oriented Pascals
01:40:04 were just missing fundamental things like, oh, you can’t
01:40:06 access this core system thing because we’re just
01:40:08 using it to teach students.
01:40:10 But Turbo Pascal showed that only modest changes to it
01:40:14 really did make it a completely capable language.
01:40:17 And it had some reasons why you could implement it
01:40:19 as a single pass compiler so it could be way, way faster,
01:40:22 although less scope for optimizations
01:40:24 if you do it that way.
01:40:26 And it did have some range checking options.
01:40:28 It had a little bit better typing capability.
01:40:30 You’d have properly typed enums, sorts of things,
01:40:33 and other stuff that C lacked.
01:40:35 But C was also clearly good enough.
01:40:37 And it wound up with a huge inertia
01:40:39 from the Unix ecosystem and everything that came with that.
01:40:42 Garbage collection?
01:40:43 No, it was not garbage collected.
01:40:45 It’s just the same kind of thing as C.
01:40:46 Same manual.
01:40:47 So you could still have your use after freeze
01:40:49 and all those other problems.
01:40:50 But just getting rid of array overruns,
01:40:53 at least if you were compiled with that debugging option,
01:40:56 certainly would have avoided a lot of problems
01:40:58 and could have a lot of benefits.
01:40:59 But so anyways, that was the next thing.
01:41:01 I had to learn C because C was where it seemed like most
01:41:04 of the things were going.
01:41:07 So I abandoned Pascal and I started working in C.
01:41:09 I started hacking on these hardware things dealing
01:41:11 with the graphics controllers and the EGA systems.
01:41:16 And what we most wanted to do, so at that time
01:41:19 we were sitting in our darkened office
01:41:22 playing all the different console video games.
01:41:24 And we were figuring out what games
01:41:27 do we want to make for our Gamer’s Edge product there.
01:41:30 And so we had one of the first Super Nintendos sitting there.
01:41:34 And we had an older Nintendo looking at all those games.
01:41:37 And the core thing that those consoles did that you just
01:41:40 didn’t get on the PC games was this ability
01:41:42 to have a massive scrolling world where
01:41:45 most of the games that you would make on the PC
01:41:47 and earlier personal computers would be a static screen.
01:41:51 You move little things around on it and you interact like that.
01:41:55 Maybe you go to additional screens as you move.
01:41:58 But arcade games and consoles had this wonderful ability
01:42:01 to just have a big world that you’re slowly
01:42:04 moving your window through.
01:42:06 And that was for those types of games,
01:42:08 that kind of action exploration adventure games,
01:42:10 that was a super, super important thing.
01:42:13 And PC games just didn’t do that.
01:42:16 And what I had come across was a couple
01:42:18 of different techniques for implementing that on the PC.
01:42:22 And they’re not hard, complicated things.
01:42:25 When I explain them now, they’re pretty straightforward.
01:42:28 But just nobody was doing it.
01:42:29 You sound like Einstein describing his five papers
01:42:31 is pretty straightforward.
01:42:33 I understand.
01:42:34 But they’re nevertheless revolutionary.
01:42:35 So side scrolling is a game changer.
01:42:38 Yeah, and scrolling is a genius invention.
01:42:40 It’s either vertical.
01:42:41 And some of the consoles had different limitations
01:42:43 about you could do one but not the other.
01:42:45 And there were similar things going on as advancements,
01:42:48 even in the console space, where you’d have the original Mario
01:42:51 game was just horizontal scrolling.
01:42:54 And then later, Mario games added vertical aspects to it
01:42:57 and different things that you were
01:42:59 doing to explore, kind of expand the capabilities there.
01:43:02 And so much of the early game design for decades
01:43:05 was removing limitations, letting you do things
01:43:08 that you envisioned as a designer,
01:43:10 you wanted the player to experience,
01:43:11 but the hardware just couldn’t really
01:43:14 or you didn’t know how to make it happen.
01:43:16 It felt impossible.
01:43:17 You can imagine that you want to create like this big world
01:43:21 through which you can side scroll,
01:43:23 like through which you can walk.
01:43:26 And then you ask yourself a question,
01:43:28 how do I actually build that in a way that’s
01:43:31 like the latency is low enough, the hardware can actually
01:43:35 deliver that in such a way that it’s a compelling experience.
01:43:38 Yeah, and we knew what we wanted to do
01:43:40 because we were playing all of these console games,
01:43:42 playing all these Nintendo games and arcade games.
01:43:45 Clearly, there was a whole world of awesome things
01:43:47 there that we just couldn’t do on the PC, at least initially.
01:43:51 Because every programmer can tell,
01:43:52 it’s like if you want to scroll, you can just
01:43:54 redraw the whole screen.
01:43:55 But then it turns out, well, you’re
01:43:56 going five frames per second.
01:43:58 That’s not an interactive fun experience.
01:44:01 You want to be going 30 or 60 frames per second or something.
01:44:04 And it just didn’t feel like that was possible.
01:44:06 It felt like the PCs had to get five times faster for you
01:44:10 to make a playable game there.
01:44:12 And interestingly, I wound up with two completely different
01:44:16 solutions for the scrolling problem.
01:44:18 And this is a theme that runs through everything, where
01:44:23 all of these big technical advancements, it turns out
01:44:25 there’s always a couple different ways of doing them.
01:44:28 And it’s not like you found the one true way of doing it.
01:44:31 And we’ll see this as we go into 3D games and things later.
01:44:35 But so the scrolling, the first set of scrolling tricks
01:44:38 that I got was the hardware had this ability to,
01:44:43 you could shift inside the window of memory.
01:44:47 So the EGA cards at the time had 256 kilobytes of memory.
01:44:51 And it was awkwardly set up in this planar format,
01:44:55 where instead of having 256 or 24 million colors,
01:45:00 you had 16 colors, which is four bits.
01:45:03 So you had four bit planes, 64k a piece.
01:45:06 Of course, 64k is a nice round number
01:45:08 for 16 bit addressing.
01:45:10 So your graphics card had a 16 bit window
01:45:13 that you could look at.
01:45:15 And you could tell it to start the video scan out
01:45:17 anywhere inside there.
01:45:19 So there were a couple games that
01:45:21 had taken this approach of you could make a 2 by 2 screen
01:45:23 or a 1 by 4 screen.
01:45:25 And you could do scrolling really easily like that.
01:45:28 You could just lay it all out and just pan around there.
01:45:30 But you just couldn’t make it any bigger
01:45:32 because that’s all the memory that was there.
01:45:35 The first insight to the scrolling that I had
01:45:37 was, well, if we make a screen that’s just one tile larger,
01:45:42 and we usually had tiles that were 16 pixels by 16 pixels,
01:45:45 the little classic Mario block that you run into,
01:45:48 lots of art gets drawn that way.
01:45:50 And your screen is a certain number of tiles.
01:45:52 But if you had one little buffer region outside of that,
01:45:56 you could easily pan around inside that 16 pixel region.
01:45:59 That could be perfectly smooth.
01:46:01 But then what happens if you get to the edge
01:46:04 and you want to keep going?
01:46:05 The first way we did scrolling was
01:46:08 what I called adaptive tile refresh, which was really
01:46:11 just a matter of you get to the edge
01:46:13 and then you go back to the original point
01:46:16 and then only change the tiles that have actually
01:46:19 that are different between where it was.
01:46:21 In most of the games at the time,
01:46:23 if you think about your classic Super Mario Brothers game,
01:46:26 you’ve got big fields of blue sky, long rows
01:46:30 of the same brick texture.
01:46:33 And there’s a lot of commonality.
01:46:34 It’s like a data compression thing.
01:46:36 If you take the screen and you set it down
01:46:38 on top of each other, in general, only about 10%
01:46:42 of the tiles were actually different there.
01:46:45 So this was a way to go ahead and say,
01:46:47 well, I’m going to move it back, and then I’m only
01:46:49 going to change those 10%, 20%, whatever percent tiles there.
01:46:54 And that meant that it was essentially five times faster
01:46:57 than if you were redrawing all of the tiles.
01:46:59 And that worked well enough for us
01:47:01 to do a bunch of these games for Gamer’s Edge.
01:47:04 We had a lot of these scrolling games like Slurred Axe
01:47:07 and Shadow Knights and things like that
01:47:09 that we were cranking out at this high rate that
01:47:11 had this scrolling effect on it.
01:47:13 And it worked well enough.
01:47:14 There were design challenges there
01:47:16 where if you made the worst case,
01:47:18 if you made a checkerboard over the entire screen,
01:47:20 you scroll over one and every single tile changes
01:47:23 and your frame rate’s now five frames per second
01:47:25 because it had to redraw everything.
01:47:27 So the designers had a little bit
01:47:28 that they had to worry about.
01:47:29 They had to make these relatively plain looking
01:47:32 levels, but it was still pretty magical.
01:47:34 It was something that we hadn’t seen before.
01:47:37 And the first thing that we wound up doing with that
01:47:41 was I had just gotten this working,
01:47:43 and Tom Hall was sitting there with me,
01:47:46 and we were looking over at our Super Nintendo on the side
01:47:49 there with Super Mario 3 running.
01:47:52 And we had the technology, we had the tools set up there,
01:47:56 and we stayed up all night.
01:47:57 And we basically cloned the first level
01:47:59 of Super Mario Brothers.
01:48:01 Performance wise as well?
01:48:02 Yeah.
01:48:02 And we had our little character running and jumping in there.
01:48:05 It was close to pixel accurate as far
01:48:09 as all the backgrounds and everything,
01:48:11 but the gaming was just stuff that we cobbled together
01:48:13 from previous games that I had written.
01:48:15 I just really kitbashed the whole thing together
01:48:18 to make this demo.
01:48:19 And that was one of the rare cases
01:48:21 when I said I don’t usually do these all night programming
01:48:24 things.
01:48:24 There’s probably only two memorable ones
01:48:26 that I can think about.
01:48:28 One was the all nighter to go ahead and get
01:48:31 our Dangerous Dave and copyright infringement
01:48:34 is how we titled it.
01:48:35 Because we had a game called Dangerous Dave running around
01:48:38 with a shotgun shooting things, and we were just
01:48:41 taking our most beloved game at the time there, the Super Mario
01:48:44 3, and sticking Dave inside that with this new scrolling
01:48:47 technology that was going perfectly smooth for them
01:48:53 as it ran.
01:48:54 And Tom and I just blearily the next morning left,
01:48:58 and we left a disk on the desk for John Romero and Jay
01:49:02 Wilbur to see and just said, run this.
01:49:05 And we eventually made it back in later in the day,
01:49:08 and it was like they grabbed us and pulled us into the room.
01:49:13 And that was the point where they were like,
01:49:15 we got to do something with this.
01:49:17 We’re going to make a company.
01:49:19 We’re going to go make our own games, where
01:49:21 this was something that we were able to just kind of hit them
01:49:24 with a hammer of an experience like, wow, this
01:49:27 is just like so much cooler than what we thought
01:49:30 was possible there.
01:49:31 And initially, we tried to get Nintendo
01:49:33 to let us make Super Mario 3 on the PC.
01:49:36 That’s really what we wanted to do.
01:49:38 We were like, hey, we can finish this.
01:49:40 It’s line of sight.
01:49:41 This will be great.
01:49:42 And we sent something to Nintendo,
01:49:45 and we heard that it did get looked at in Japan,
01:49:48 and they just weren’t interested in that.
01:49:50 But that’s another one of those life could
01:49:52 have gone a very different way, where
01:49:53 we could have been like Nintendo’s house PC team
01:49:57 at that point.
01:49:58 And define the direction of Wolfenstein and Doom and Quake
01:50:06 could have been a Nintendo creation.
01:50:08 Yeah.
01:50:09 So at the same time that we were just
01:50:11 doing our first scrolling demos, we
01:50:14 reached out to Scott Miller at Apogee and said,
01:50:17 it’s like, hey, we do want to make some games.
01:50:20 These things that you think you want, those are nothing.
01:50:22 What do you see what we can actually do now?
01:50:24 This is going to be amazing.
01:50:26 And he just popped right up and sent a check to us,
01:50:29 where at that point, we still thought
01:50:31 he might be a fraud, that he was just lying about all of this.
01:50:34 But he was totally correct on how much money
01:50:36 he was making with his shareware titles.
01:50:39 And this was his real brainstorm about this,
01:50:43 where shareware was this idea that software
01:50:45 doesn’t have a fixed price.
01:50:47 If you use it, you send out of the goodness of your heart
01:50:49 some money to the creator.
01:50:51 And there were a couple of utilities
01:50:53 that did make some significant success like that.
01:50:55 But for the most part, it didn’t really work.
01:50:58 Now, there wasn’t much software in a pure shareware model
01:51:01 that was successful.
01:51:03 The Apogee innovation was to take something,
01:51:07 call it shareware, split it into three pieces.
01:51:10 You always made a trilogy.
01:51:12 And you would put the first piece out,
01:51:14 but then you buy the whole trilogy
01:51:16 for some shareware amount, which in reality,
01:51:19 it meant that the first part was a demo,
01:51:21 where you kind of like the demo went everywhere for free
01:51:24 and you paid money to get the whole set.
01:51:26 But it was still play to shareware.
01:51:28 And we were happy to have the first one go everywhere.
01:51:31 And it wasn’t a crippled demo, where the first episode
01:51:33 of all these trilogies, it was a real complete game.
01:51:36 And probably 20 times as many people played that part of it,
01:51:39 thought they had a great game, had fond memories of it,
01:51:43 but never paid us a dime.
01:51:45 But enough people were happy with that,
01:51:48 where it was really quite successful.
01:51:50 And these early games that we didn’t think very much of
01:51:53 compared to commercial quality games,
01:51:55 but they were doing really good business,
01:51:57 some fairly crude things.
01:51:59 And people, it was good business.
01:52:01 People enjoyed it.
01:52:02 And it wasn’t like you were taking a crapshoot
01:52:04 on what you were getting.
01:52:05 You just played a third of the experience.
01:52:07 And you loved it enough to hand write out a check
01:52:10 and put it in an envelope and address it and send it out
01:52:13 to Apogee to get the rest of them.
01:52:16 So it was a really pretty feel good business prospect there,
01:52:20 because everybody was happy.
01:52:23 They knew what they were getting when they sent it in.
01:52:26 They would send in fan mail.
01:52:27 If you’re going to the trouble of addressing a letter
01:52:29 and filling out an envelope, you write something in it.
01:52:32 And there were just the literal bags of fan mail
01:52:35 for the shareware games, so people loved them.
01:52:38 I should mention that for you, the definition of wealth
01:52:42 is being able to have pizza whenever you want.
01:52:46 For me, there was a dream that I would play shareware games
01:52:50 over and over, the part that’s free, over and over.
01:52:53 And it was very deeply fulfilling experience.
01:52:56 But I dreamed of a time when I could actually
01:53:00 afford the full experience.
01:53:02 And this is kind of this dreamland beyond the horizon
01:53:05 when you could find out what else is there.
01:53:09 In some sense, even just playing the shareware was,
01:53:14 it’s the limitation of that.
01:53:17 You know, life is limited, eventually we all die.
01:53:21 In that way, shareware was somehow really fulfilling
01:53:26 to have this kind of mysterious thing beyond what’s free
01:53:31 always there.
01:53:32 It’s kind of, I don’t know, maybe it’s because
01:53:35 a part of my childhood is playing shareware games.
01:53:37 That was a really fulfilling experience.
01:53:39 It’s so interesting how that model still brought joy
01:53:43 to so many people.
01:53:44 The 20X people that played it.
01:53:45 I felt very good about that.
01:53:47 I would run into people that would say,
01:53:49 oh, I loved that game that you had early on,
01:53:52 Commander Keen, whatever.
01:53:53 And no, they meant just the first episode
01:53:57 that they got to see everywhere.
01:53:58 That’s me, I played the crap out of Commander Keen.
01:54:01 And that was all good.
01:54:03 Yeah, yeah.
01:54:04 But so we were in this position where Scott Miller
01:54:06 was just fronting us cash, saying, yeah, make a game.
01:54:10 But we did not properly pull the trigger and say,
01:54:14 all right, we’re quitting our jobs.
01:54:16 We were like, we’re gonna do both.
01:54:17 We’re gonna keep working at Softdisk, working on this.
01:54:21 And then we’re going to go ahead and make a new game
01:54:24 for Apogee at the same time.
01:54:27 And this eventually did lead to some legal problems.
01:54:29 And we had trouble.
01:54:31 It all got worked out in the end,
01:54:32 but it was not a good call at the time there.
01:54:35 And your legal mind at the time was not stellar.
01:54:39 You were not thinking in terms of the illegal terms.
01:54:42 No, I definitely wasn’t.
01:54:44 None of us were.
01:54:46 And in hindsight, yeah, it’s like,
01:54:47 how did we think we were gonna get away
01:54:48 with even using our work computers
01:54:51 to write software for our breakaway new company?
01:54:56 It was not a good plan.
01:54:58 How did Commander Keen come to be?
01:55:00 So the design process, we would start from,
01:55:04 we had some idea of what we wanted to do.
01:55:05 We wanted to do a Mario like game.
01:55:08 It was gonna be a side scroller.
01:55:10 It was gonna use the technology.
01:55:11 We had some sense of what it would have to look like
01:55:14 because of the limitations
01:55:15 of this adaptive tile refresh technology.
01:55:17 It had to have fields of relatively constant tiles.
01:55:21 You couldn’t just paint up a background
01:55:23 and then move that around.
01:55:25 The early design or all the design for Commander Keen
01:55:28 really came from Tom Hall,
01:55:30 where he was kind of the main creative mind
01:55:34 for the early id software stuff,
01:55:36 where we had an interesting division of things
01:55:39 where Tom was all creative and design.
01:55:42 I was all programming.
01:55:43 John Romero was an interesting bridge
01:55:45 where he was both a very good programmer
01:55:47 and also a very good designer and artist
01:55:50 and kind of straddled between the areas.
01:55:52 But Commander Keen was very much Tom Hall’s baby.
01:55:55 And he came up with all the design and backstory
01:55:59 for the different things of kind of a mad scientist,
01:56:02 little kid with building a rocket ship
01:56:06 and a zap gun and visiting alien worlds
01:56:09 and doing all of this that the background
01:56:11 that we lay the game inside of.
01:56:13 And there’s not a whole lot to any of these things.
01:56:16 Design for us was always just what we needed to do
01:56:19 to make the game that was gonna be so much fun to play.
01:56:23 And we laid out our first trilogy of games,
01:56:26 the shareware formula, it’s gonna be three pieces.
01:56:29 We make Commander Keen one, two, and three.
01:56:31 And we just really started busting on all that work.
01:56:35 And it went together really quickly.
01:56:37 It was like three months or something
01:56:39 that while we were still making games every month
01:56:41 for Gamer’s Edge, we were sharing technology between that.
01:56:45 I’d write a bunch of code for this
01:56:46 and we’d just kind of use it for both.
01:56:48 Again, not a particularly good idea there
01:56:50 that had consequences for us.
01:56:53 But in three months, we got our first game out
01:56:57 and all of a sudden it was three times as successful
01:57:00 as the most successful thing Apogee had had before.
01:57:03 And we were making like $30,000 a month
01:57:05 immediately from the Commander Keen stuff.
01:57:09 And that was again, a surprise to us.
01:57:11 It was more than we thought that that was gonna make.
01:57:15 And we said, well, we’re gonna certainly roll
01:57:17 into another set of titles from this.
01:57:19 And in that three months, I had come up
01:57:21 with a much better way of doing the scrolling technology
01:57:24 that was not the adaptive tile refresh,
01:57:27 which in some ways was even simpler.
01:57:29 And these things, so many of the great ideas of technology
01:57:33 are things that are back of the envelope designs.
01:57:36 I make this comment about modern machine learning
01:57:38 where all the things that are really important
01:57:40 practically in the last decade are,
01:57:42 each of them fits on the back of an envelope.
01:57:44 There are these simple little things.
01:57:46 They’re not super dense, hard to understand technologies.
01:57:51 And so the second scrolling trick was just a matter of like,
01:57:55 okay, we know we’ve got this 64K window.
01:57:58 And the question was always like,
01:58:00 well, you could make a two by two,
01:58:02 but you can’t go off the edge.
01:58:05 But I finally asked, well, what actually happens
01:58:07 if you just go off the edge?
01:58:09 If you take your start and you say, it’s like,
01:58:12 okay, I can move over, I’m scrolling,
01:58:14 I can move over, I can move down, I’m scrolling.
01:58:16 I get to what should be the bottom of the memory window.
01:58:19 It’s like, well, what if I just keep going?
01:58:21 And I say, I’m gonna start at,
01:58:24 what happens if I start at FFFE
01:58:26 at the very end of the 64K block?
01:58:29 And it turns out it just wraps back around to the top
01:58:32 of the block.
01:58:33 And I’m like, oh, well, this makes everything easy.
01:58:35 You can just scroll the screen everywhere
01:58:37 and all you have to draw is just one new line of tiles,
01:58:40 which everything you expose,
01:58:42 it might be unaligned off various parts
01:58:44 of the screen memory, but it just works.
01:58:48 That no longer had the problem of you had to have fields
01:58:50 of the similar colors because it doesn’t matter
01:58:53 what you’re doing, you could be having
01:58:54 a completely unique world and you’re just drawing
01:58:57 the new strip as it comes on.
01:58:59 But it might be, like you said, unaligned,
01:59:02 so it can be all over the place.
01:59:03 Yes, and it turns out it doesn’t matter.
01:59:04 I would have two page flipped screens,
01:59:06 as long as they didn’t overlap, they moved in series
01:59:09 through this two dimensional window of graphics.
01:59:12 And that was one of those like, well, this is so simple.
01:59:15 This just works, it’s faster,
01:59:19 there it seemed like there was no downside.
01:59:21 Funny thing was, it turned out after we shipped titles
01:59:25 with this, there were what they called super VGA cards,
01:59:29 the cards that would allow higher resolutions
01:59:31 and different features that the standard ones didn’t.
01:59:35 And on some of those cards,
01:59:38 this was a weird compatibility quirk again,
01:59:40 because nobody thought this was not
01:59:41 what it was designed to do.
01:59:43 And some of those cards had more memory.
01:59:45 They had more than just 256K and four planes.
01:59:48 They had 512K or a megabyte.
01:59:51 And on some of those cards, I scroll my window down
01:59:55 and then it goes into uninitialized memory
01:59:57 that actually exists there
01:59:59 rather than wrapping back around at the top.
02:00:01 And then I was in the tough position
02:00:04 of do I have to track every single one of these?
02:00:06 And it was a mad house back then
02:00:07 with there were 20 different video card vendors
02:00:10 with all slightly different implementations
02:00:12 of their nonstandard functionality.
02:00:14 So either I needed to natively program
02:00:17 all of the VGA cards there to map in that memory
02:00:22 and keep scrolling down through all of that,
02:00:24 or I kind of punted and took the easy solution
02:00:27 of when you finally did run to the edge of the screen,
02:00:30 I accepted a hitch and just copied the whole screen up there.
02:00:33 So on some of those cards, it was a compatibility mode.
02:00:38 In the normal ones, when it all worked fine,
02:00:40 everything was just beautifully smooth.
02:00:42 But if you had one of those cards
02:00:43 where it did not wrap the way I wanted it to,
02:00:46 you’d be scrolling around, scrolling around,
02:00:48 and then eventually you’d have a little hitch
02:00:50 where 200 milliseconds or something
02:00:52 that was not super smooth as it froze a little bit.
02:00:56 And this is the binary thing.
02:00:57 Is it one of the standard screens
02:00:59 or is it one of the weird ones, the Super VGA ones?
02:01:02 Yeah. Okay.
02:01:03 And so we would default to,
02:01:04 and I think that was one of those
02:01:05 that changed over the kind of course of deployment
02:01:09 where early on we would have a normal mode
02:01:11 and then you would enable the compatibility flag
02:01:13 if your screen did this crazy flickery thing
02:01:16 when you got to a certain point in the game.
02:01:18 And then later, I think it probably got enabled by default
02:01:21 as just more and more of the cards
02:01:23 kind of did not do exactly the right thing.
02:01:26 And that’s the two edged sword
02:01:27 of doing unconventional things with technology
02:01:30 where you can find something that nobody thought about
02:01:33 doing that kind of scrolling trick
02:01:34 when they set up those cards.
02:01:36 But the fact that nobody thought that was the primary reason
02:01:39 when I was relying on that,
02:01:40 then I wound up being broken on some of the later cards.
02:01:44 Let me take a bit of a tangent,
02:01:45 but ask you about the hacker ethic.
02:01:50 Cause you mentioned shareware.
02:01:51 It’s an interesting world.
02:01:53 The world of people that make money,
02:01:56 the business and the people that build systems,
02:02:00 the engineers.
02:02:02 And what is the hacker ethic?
02:02:05 You’ve been a man of the people
02:02:08 and you’ve embodied at least the part of that ethic.
02:02:12 What does it mean?
02:02:13 What did it mean to you at the time?
02:02:14 What does it mean to you today?
02:02:15 So Steven Levy’s book, Hackers,
02:02:17 was a really formative book for me as a teenager.
02:02:21 I mean, I read it several times
02:02:23 and there was all of the great lore
02:02:26 of the early MIT era of hackers
02:02:28 and ending up at the end with,
02:02:31 it kind of went through the early MIT hackers
02:02:34 and then the Silicon Valley hardware hackers
02:02:36 and then the game hackers in part three.
02:02:39 And at that time as a teenager,
02:02:41 I really was kind of bitter in some ways.
02:02:44 I thought I was born too late.
02:02:46 I thought I missed the window there.
02:02:48 And I really thought I belonged in that third section
02:02:52 of that book with the game hackers.
02:02:53 And they were talking about the Williams at Sierra
02:02:56 and origin systems with Richard Garriott.
02:02:59 And it’s like, I really wanted to be there.
02:03:02 And I knew that was now a few years in the past.
02:03:05 It was not to be,
02:03:08 but the early days, especially the early MIT hacker days,
02:03:12 talking a lot about this sense of the hacker ethic,
02:03:16 that there was this sense that
02:03:17 it was about sharing information, being good,
02:03:20 not keeping it to yourself,
02:03:22 and that it’s not a zero sum game,
02:03:24 that you can share something with another programmer
02:03:28 and it doesn’t take it away from you.
02:03:30 You then have somebody else doing something.
02:03:33 And I also think that there’s an aspect of it
02:03:35 where it’s this ability to take joy
02:03:39 in other people’s accomplishments,
02:03:41 where it’s not the cutthroat bit of like,
02:03:43 I have to be first,
02:03:44 I have to be recognized as the one that did this
02:03:47 in some way,
02:03:48 but being able to see somebody else do something
02:03:51 and say, holy shit, that’s amazing.
02:03:53 And just taking joy in the ability of something amazing
02:03:56 that somebody else does.
02:03:58 And the big thing that I was able to do through ID Software
02:04:03 was this ability to eventually release the source code
02:04:07 for most of our,
02:04:08 like all of our really seminal game titles.
02:04:10 And that was a, it was a stepping stone process
02:04:13 where we were kind of surprised early on
02:04:16 where people were able to hack the existing games.
02:04:20 And of course I had experience with that.
02:04:21 I remember hacking my copies of Ultima.
02:04:23 So I’d give myself 9999 gold and raise my levels
02:04:27 and break out the sector editor.
02:04:29 And so I was familiar with all of that.
02:04:31 So it was just, it was with a smile
02:04:33 when I started to see people doing that to our games,
02:04:35 I am making level editors for Commander Keen
02:04:38 or hacking up Wolfenstein 3D,
02:04:41 but I made the pitch internally
02:04:44 that we should actually release our own tools
02:04:47 for like what we did, what we use to create the games.
02:04:51 And that was a little bit debatable about,
02:04:55 well, will this let other, we’ll give people leg up.
02:04:57 It’s always like,
02:04:58 what’s that gonna mean for the competition?
02:05:00 But the really hard pitch was to actually release
02:05:04 the full source code for the games.
02:05:06 And it was a balancing act with the other people
02:05:10 inside the company where it’s interesting
02:05:13 how the programmers generally did get,
02:05:17 certainly the people that I worked closely with,
02:05:20 they did kind of get that hacker ethic bit
02:05:22 where you wanted to share your code.
02:05:24 You were proud of it.
02:05:25 You wanted other people to take it, do cool things with it.
02:05:28 But interestingly, the broader game industry
02:05:33 is a little more hesitant to embrace that
02:05:36 than like the group of people
02:05:38 that we happen to have at id Software,
02:05:40 where it was always a little interesting to me
02:05:42 seeing how a lot of people in the game modding community
02:05:45 were very possessive of their code.
02:05:47 They did not wanna share their code.
02:05:49 They wanted it to be theirs.
02:05:50 It was their claim to fame.
02:05:52 And that was much more like what we tended to see
02:05:55 with artists where the artists understand something
02:05:58 about credit and wanting it to be known as their work.
02:06:02 And a lot of the game programmers
02:06:05 felt a little bit more like artists
02:06:07 than like hacker programmers,
02:06:08 in that it was about building something
02:06:11 that maybe felt more like art to them
02:06:13 than the more tool based and exploration based
02:06:16 kind of hacking culture side of things.
02:06:19 Yeah, it’s so interesting that this kind of fear
02:06:23 that credit will not be sufficiently attributed to you.
02:06:27 And that’s one of the things that I do bump into a lot
02:06:30 because I try not to go…
02:06:34 I mean, it’s easy for me to say
02:06:35 because so much credit is heaped on me
02:06:37 for the id Software side of things.
02:06:39 But when people come up and they wanna pick a fight
02:06:42 and say, no, it’s like that wasn’t where first person gaming
02:06:45 came from and you can point to some of like things
02:06:48 on obscure titles that I was never aware of
02:06:51 or like the old Playdoh systems
02:06:53 or each personal computer had something that was 3Dish
02:06:57 and moving around.
02:06:58 And I’m happy to say it’s like, no, I mean,
02:07:01 I saw Battlezone and Star Wars in the arcades.
02:07:03 I had seen 3D graphics, I had seen all these things
02:07:06 that are standing on the shoulders of lots of other people.
02:07:09 But sometimes these examples they pull out,
02:07:10 it’s like, no, I didn’t know that existed.
02:07:12 I mean, I had never heard of that before then
02:07:15 that didn’t contribute to what I made,
02:07:17 but there’s plenty of stuff that did.
02:07:19 And I think there’s good cases to be made
02:07:23 that obviously Doom and Quake and Wolfenstein
02:07:26 were formative examples for what everything
02:07:29 that came after that.
02:07:31 But I don’t feel the need to go fight
02:07:34 and say claim primacy or initial invention
02:07:37 of anything like that.
02:07:38 But a lot of people do want to.
02:07:40 I think when you fight for the credit in that way
02:07:43 and it does go against the hacker ethic,
02:07:46 you destroy something fundamental about the culture,
02:07:48 about the community that builds cool stuff.
02:07:51 I think credit ultimately,
02:07:53 so I had this sort of,
02:07:58 there’s a famous wrestler and freestyle wrestling
02:08:00 called Buvai Sarasateeva.
02:08:03 And he always preached that you should just focus
02:08:07 on the art of the wrestling and let people
02:08:12 write your story however they want.
02:08:15 The highest form of the art is just focusing on the art.
02:08:20 And that is something about the hacker ethic
02:08:23 is just focus on building cool stuff,
02:08:26 sharing it with other cool people,
02:08:29 and credit will get assigned correctly
02:08:35 in the long arc of history.
02:08:37 Yeah, and I generally think that’s true.
02:08:39 And you’ve got,
02:08:42 there’s some things, there’s a graphics technique
02:08:44 that got labeled CarMax Reverse.
02:08:46 I am literally named,
02:08:48 and it turned out that I wasn’t the first person
02:08:50 to figure that out.
02:08:51 Most scientific things or mathematical things you wind up,
02:08:54 it’s like, oh, this other person
02:08:56 had actually done that somewhat before.
02:08:58 And then there’s things that get attributed to me
02:09:00 like the inverse square root hack that I actually didn’t do.
02:09:03 I flat out, that wasn’t me.
02:09:04 And it’s like, it’s weird how the memetic power
02:09:07 of the internet, I cannot convince people of that.
02:09:09 You’re like the Mark Twain of programming.
02:09:10 Yes.
02:09:12 Everything just gets attributed to you now,
02:09:14 even though you’ve never sought the credit of things.
02:09:18 But part of the fact of the humility behind that
02:09:21 is what attracts the attributions.
02:09:25 Let’s talk about a game.
02:09:27 I mean, one of the greatest games ever made.
02:09:29 I know you could talk about doing Quake and so on,
02:09:31 but to me, Wolfenstein 3D was like, whoa.
02:09:35 It blew my mind that a world like this could exist.
02:09:38 So how did Wolfenstein 3D come to be
02:09:42 in terms of the programming, in terms of the design,
02:09:44 in terms of some of the memorable technical challenges?
02:09:47 And also, actually just something you haven’t mentioned
02:09:50 is how did these ideas come to be inside your mind,
02:09:58 the adaptive side scrolling,
02:10:00 the solutions to these technical challenges?
02:10:03 So I usually can introspectively pull back
02:10:06 pretty detailed accounts of how technology solutions
02:10:11 and design choices on my part came to be,
02:10:14 where technically we had done two games,
02:10:17 3D games like that before,
02:10:19 where Hover Tank was the first one,
02:10:20 which had flat shaded walls,
02:10:22 but did have the scaled enemies inside it.
02:10:25 And then Catacombs 3D, which had textured walls,
02:10:28 scaled enemies, and some more functionality,
02:10:34 like the disappearing walls and some other stuff.
02:10:37 But what’s really interesting
02:10:38 from a game development standpoint is those games,
02:10:41 Catacombs 3D, Hover Tank, and Wolfenstein,
02:10:45 they literally used the same code
02:10:48 for a lot of the character behavior
02:10:50 that a 2D game that I had made earlier called Catacombs did,
02:10:54 where it was an overhead view game, kind of like Gauntlet,
02:10:56 you’re running around and you can open up doors,
02:10:58 pick up items, basic game stuff.
02:11:01 And the thought was that this exact same game experience
02:11:06 just presented in a different perspective.
02:11:09 It could be literally the same game
02:11:11 just with a different view into it,
02:11:13 would have a dramatically different impact on the players.
02:11:16 So it wasn’t a true 3D,
02:11:20 you’re saying that you could kind of fake it,
02:11:22 you can like scale enemies,
02:11:24 meaning things that are farther away,
02:11:25 you can make them smaller.
02:11:27 So the game was a 2D map,
02:11:29 like all of our games use the same tool for creation.
02:11:33 We use the same map editor for creating Keen as Wolfenstein
02:11:36 and Hover Tank and Catacombs and all this stuff.
02:11:39 So the game was a 2D grid made out of blocks.
02:11:42 And you could say, well, these are walls,
02:11:44 these are where the enemies start,
02:11:45 then they start moving around.
02:11:47 And these early games like Catacombs,
02:11:49 you played it strictly in a 2D view.
02:11:51 It was a scrolling 2D view,
02:11:53 and that was kind of using an adaptive tile refresh
02:11:55 at the time to be able to do something like that.
02:11:58 And then the thought that these early games,
02:12:01 all it did was take the same basic enemy logic,
02:12:04 but instead of seeing it from the God’s eye view on top,
02:12:07 you were inside it and turning from side to side,
02:12:10 yawing your view and moving forwards and backwards
02:12:12 and side to side.
02:12:14 And it’s a striking thing where you always talk about
02:12:17 wanting to isolate and factor changes in values.
02:12:20 And this was one of those most pure cases there
02:12:22 where the rest of the game changed very little.
02:12:25 It was our normal kind of change the colors on something
02:12:27 and draw a different picture for it,
02:12:29 but it’s kind of the same thing.
02:12:30 But the perspective changed in a really fundamental way.
02:12:33 And it was dramatically different.
02:12:36 I can remember the reactions where the artist, Adrian,
02:12:40 that had been drawing the pictures for it.
02:12:42 We had a cool big troll thing in Catacombs 3D,
02:12:45 and we had these walls that you could get a key
02:12:47 and you could make the blocks disappear,
02:12:49 get really simple stuff.
02:12:51 Blocks could either be there or not there.
02:12:52 So our idea of a door was being able to make a set of blocks
02:12:55 just disappear.
02:12:57 And I remember the reaction
02:12:58 where he had drawn these characters
02:13:00 and he was slowly moving around
02:13:01 and like people had no experience with 3D navigation.
02:13:04 It was all still keyboard.
02:13:05 We didn’t even have mice set up at that time,
02:13:08 but slowly moving, going up, picked up a key, go to a wall.
02:13:12 The wall disappears in a little animation
02:13:14 and there’s a monster like right there.
02:13:16 And he practically fell out of his chair.
02:13:18 It was just like, ah!
02:13:19 And games just didn’t do that.
02:13:22 The games were the God’s eye view.
02:13:24 You were a little invested in your little guy.
02:13:26 You can be like happy or sad when things happen,
02:13:29 but you just did not get that kind of startle reaction.
02:13:32 You weren’t inside the game.
02:13:33 Something in the back of your brain,
02:13:35 some reptile brain thing is just going,
02:13:37 oh shit, something just happened.
02:13:39 And that was one of those early points where it’s like,
02:13:42 yeah, this is gonna make a difference.
02:13:44 This is going to be powerful and it’s gonna matter.
02:13:47 Were you able to imagine that in the idea stage or no?
02:13:51 So not that exact thing.
02:13:53 So again, we had cases like the arcade games,
02:13:56 Battle Zone and Star Wars
02:13:58 that you could kind of see a 3D world
02:14:00 and things coming at you and you get some sense of it,
02:14:03 but nothing had done the kind of worlds that we were doing
02:14:06 and the sort of action based things.
02:14:08 3D at the time was really largely
02:14:12 about the simulation thoughts.
02:14:14 And this is something that really
02:14:16 might have trended differently
02:14:18 if not for the id Software approach in the games
02:14:20 where there were flight simulators,
02:14:23 there were driving simulators,
02:14:24 you had like hard drive in and Microsoft flight simulator.
02:14:28 And these were doing 3D and general purpose 3D
02:14:31 in ways that were more flexible
02:14:33 than what we were doing with our games,
02:14:35 but they were looked at as simulations.
02:14:38 They weren’t trying to necessarily be fast or responsive
02:14:42 or letting you do kind of exciting maneuvers
02:14:44 because they were trying to simulate reality
02:14:46 and they were taking their cues from the big systems,
02:14:49 the Evans and Sutherlands and the Silicon Graphics
02:14:51 that were doing things.
02:14:52 But we were taking our cues
02:14:54 from the console and arcade games.
02:14:56 We wanted things that were sort of quarter eaters
02:14:59 that were doing fast paced things
02:15:01 that you could smack you around
02:15:02 rather than just smoothly gliding you from place to place.
02:15:06 So.
02:15:07 Quarter eaters.
02:15:07 Yeah.
02:15:08 And you know, a funny thing is,
02:15:10 so much that that built into us
02:15:12 that Wolfenstein still had lives
02:15:15 and you had like one of the biggest power ups
02:15:16 in all these games like was an extra life
02:15:18 because you started off with three lives
02:15:20 and you lose your lives and then it’s game over
02:15:23 and there weren’t save games in most of this stuff.
02:15:26 It was, it sounds almost crazy to say this,
02:15:28 but it was an innovation in Doom
02:15:30 to not have lives.
02:15:32 You know, you could just play Doom as long as you wanted.
02:15:33 You just restart at the start of the level and why not?
02:15:36 This is like, we aren’t trying to take people’s quarters.
02:15:39 They’ve already paid for the entire game.
02:15:41 We want them to have a good time
02:15:43 and you would have some, you know,
02:15:45 some old timer purists that might think
02:15:47 that there’s something to the epic journey
02:15:48 of making it to the end,
02:15:50 having to restart all the way from the beginning
02:15:52 after a certain number of tries.
02:15:53 But now more fun is had when you just let people
02:15:56 kind of keep trying when they’re stuck
02:15:58 rather than having to go all the way back
02:15:59 and learn different things.
02:16:01 So you’ve recommended the book,
02:16:03 Game Engine Black Book Wolfenstein 3D
02:16:05 for technical exploration of the game.
02:16:07 So looking back 30 years,
02:16:10 what are some memorable technical innovations
02:16:13 that made this perspective shift into this world
02:16:16 that’s so immersive that scares you when a monster appears
02:16:20 or some things you have to solve?
02:16:22 So one of the interesting things
02:16:24 that come back to the theme of deadlines
02:16:26 and resource constraints,
02:16:27 the game Catacombs 3D,
02:16:31 we shipped, we were supposed to be shipping this
02:16:33 for Gamers Edge on a monthly cadence and I had slipped.
02:16:36 I was actually late.
02:16:37 It slipped like six weeks
02:16:39 because this was texture mapped walls
02:16:41 doing stuff that I hadn’t done before.
02:16:44 And at the six week point,
02:16:46 it was still kind of glitchy and buggy.
02:16:48 There were things that I knew that if you had a wall
02:16:51 that was like almost edge on, you could slide over to it
02:16:54 and you could see some things freak out
02:16:56 or vanish or not work.
02:16:57 And I hated that,
02:16:59 but I was up against the wall.
02:17:01 We had to ship the game.
02:17:03 It was still a lot of fun to play.
02:17:04 It was novel, nobody had seen it.
02:17:06 It gave you that startle reflex reaction.
02:17:09 So it was worth shipping,
02:17:11 but it had these things that I knew were kind of flaky
02:17:13 and janky and not what I was really proud of.
02:17:17 So one of the things that I did very differently
02:17:20 in Wolfenstein was I went,
02:17:24 Catacombs used almost a conventional thing
02:17:27 where you had segments that were one dimensional polygons
02:17:30 basically that were clipped and back faced
02:17:33 and done kind of like a very crude 3D engine
02:17:36 from the professionals,
02:17:37 but I wasn’t getting it done right.
02:17:39 I was not doing a good enough job.
02:17:41 I didn’t really have line of sight to fix it right.
02:17:45 There’s stuff that of course I look back,
02:17:46 it’s like, oh, it’s obvious how to do this
02:17:48 and do the math right, do your clipping right,
02:17:51 check all of this, how you handle the precision.
02:17:53 But I did not know how to do that at that time.
02:17:55 Was that the first 3D engine you wrote Catacombs 3D?
02:17:59 Yeah, Hover Tank had been a little bit before that,
02:18:00 but that had the flat shaded walls.
02:18:02 So the texture mapping on the walls
02:18:04 was what was bringing in some of these challenges
02:18:06 that was hard for me
02:18:08 and I couldn’t solve it right at the time.
02:18:11 Can you describe what flat shading is and texture mapping?
02:18:13 So the walls were solid color,
02:18:16 one of 16 colors in Hover Tank.
02:18:19 So that’s easy, it’s fast,
02:18:20 you just draw the solid color for everything.
02:18:23 Texture mapping is what we all see today
02:18:25 where you have an image that is stretched and distorted
02:18:28 onto the walls or the surfaces that you’re working with.
02:18:32 And it was a long time for me to just figure out
02:18:35 how to do that without it distorting in the wrong ways.
02:18:38 And I did not get it all exactly right in Catacombs
02:18:42 and I had these flaws.
02:18:44 So that was important enough to me
02:18:46 that rather than continuing to bang my head on that
02:18:49 when I wasn’t positive I was gonna get it,
02:18:51 I went with a completely different approach
02:18:53 for drawing for figuring out where the walls were
02:18:56 which was a ray casting approach,
02:18:58 which I had done in Catacombs 3D I had a bunch of C code
02:19:02 trying to make this work right and it wasn’t working right.
02:19:05 In Wolfenstein I wound up going to a very small amount
02:19:09 of assembly code.
02:19:11 So in some ways there should be a slower way of doing it
02:19:14 but by making it a smaller amount of work
02:19:16 that I could more tightly optimize it worked out
02:19:19 and Wolfenstein 3D was just absolutely rock solid.
02:19:22 It was nothing glitched in there.
02:19:25 The game just was pretty much flawless through all of that
02:19:28 and I was super proud of that.
02:19:31 But eventually like in the later games
02:19:33 I went back to the more span based things
02:19:35 where I could get more total efficiency
02:19:37 once I really did figure out how to do it.
02:19:40 So there were two sort of key technical things
02:19:42 to Wolfenstein, one was this ray casting approach
02:19:45 which you still to this day you see people go and say
02:19:48 let’s write a ray casting engine
02:19:50 because it’s an understandable way of doing things
02:19:53 that lets you make games very much like that.
02:19:55 So you see ray casters in JavaScript,
02:19:57 ray casters in Python, people that are basically going
02:20:00 and re implementing that approach to taking a tiled world
02:20:04 and casting out into it.
02:20:06 It works pretty well but it’s not the fastest way
02:20:08 of doing it.
02:20:09 Can you describe what ray casting is?
02:20:11 So you start off and you’ve got your screen
02:20:13 which is 320 pixels across at the time
02:20:15 if you haven’t sized down the window for greater speed
02:20:19 and at every pixel there’s gonna be an angle
02:20:22 from you’ve got your position in the world
02:20:24 and you’re going to just run along that angle
02:20:26 and keep going until you hit a block.
02:20:29 So up to 320 times across there
02:20:32 it’s gonna throw a cast array out into the world
02:20:35 from wherever your origin is until it runs into a wall
02:20:38 and then it can figure out exactly where on the wall it hits.
02:20:42 The performance challenge of that is as it’s going out
02:20:45 every block it’s crossing it checks is this a solid wall.
02:20:49 So that means that in like the early Wolfenstein levels
02:20:52 you’re in a small jail cell going out into a small hallway
02:20:55 it’s super efficient for that
02:20:57 because you’re only stepping across three or four blocks
02:21:00 but then if somebody makes a room that covers
02:21:02 our maps were limited to 64 by 64 blocks.
02:21:05 If you made one room that was nothing but walls
02:21:08 at the far space it would go pretty slow
02:21:11 because it would be stepping across 80 tile tests
02:21:14 or something along the way.
02:21:15 By the way the physics of our universe
02:21:17 seems to be competing in this very thing.
02:21:19 So this maps nicely to the actual physics of our world.
02:21:23 Yeah you get like I follow a little bit of something
02:21:25 like Steven Wolfram’s work on interconnected
02:21:28 network information states of that
02:21:30 and that’s it’s beyond what I can have an informed opinion
02:21:34 on but it’s interesting that people are considering
02:21:37 things like that and have things that can back it up.
02:21:42 Yeah there’s whole different sets
02:21:44 of interesting stuff there.
02:21:45 So Wolfenstein 3D had ray casting.
02:21:48 Ray casting and then the other kind of key aspect
02:21:51 was what I called compiled scalers
02:21:53 where the idea of you’d saw this in the earlier
02:21:58 classic arcade games like Space Harrier and stuff
02:22:01 where you would take a picture which is normally drawn
02:22:04 directly on the screen and then if you have the ability
02:22:06 to make it bigger or smaller big chunky pixels
02:22:09 or fizzly small drop sampled pixels.
02:22:12 That’s the fundamental aspect of what our characters
02:22:15 were doing in these 3D games.
02:22:16 You would have it’s just like you might have drawn
02:22:18 a tiny little character but now we can make them
02:22:20 really big and make them really small and move it around.
02:22:23 That was the limited kind of 3D that we had for characters.
02:22:26 To make them turn there were literally
02:22:27 eight different views of them.
02:22:29 You didn’t actually have a 3D model that would rotate.
02:22:31 You just had these cardboard cutouts.
02:22:33 But that was good enough for that startle fight reaction
02:22:36 and it was kind of what we had to deal with there.
02:22:40 So a straightforward approach to do that
02:22:42 you could just write out your doubly nested loop
02:22:44 of you’ve got your stretch factor
02:22:47 and it’s like you’ve got a point,
02:22:48 you stretch by a little bit,
02:22:49 it might be on the same pixel,
02:22:50 it might be on the next pixel,
02:22:51 it might have skipped a pixel.
02:22:53 You can write that out but it’s not gonna be fast enough
02:22:56 where especially you get a character for that
02:22:58 right in your face monster covering
02:23:00 almost the entire screen.
02:23:02 Doing that with a general purpose scaling routine
02:23:05 would have just been much too slow.
02:23:06 It would have worked when they’re small characters
02:23:08 but then it would get slower and slower
02:23:09 as they got closer to you until right at the time
02:23:12 when you most care about having a fast reaction time,
02:23:15 the game would be chunking down.
02:23:17 So the fastest possible way to draw pixels at that time
02:23:22 was to, instead of saying I’ve got a general purpose
02:23:28 version that can handle any scale,
02:23:31 I used a program to make essentially
02:23:35 a hundred or more separate little programs
02:23:37 that was optimized for I will take an image
02:23:40 and I will draw it 12 pixels tall.
02:23:42 I’ll take an image, I’ll draw it 14 pixels tall
02:23:44 up by every two pixels even for that.
02:23:47 So you would have the most optimized code
02:23:50 so that in the normal case
02:23:52 where most of the world is fairly large,
02:23:55 like the pixels are big,
02:23:57 we did not have a lot of memory.
02:23:58 So in most cases that meant that you would load
02:24:01 a pixel color and then you would store it multiple times.
02:24:05 So that was faster than even copying an image
02:24:09 in a normal conventional case
02:24:10 because most of the time the image is expanded.
02:24:13 So instead of doing one read, one write for a simple copy,
02:24:16 you might be doing one read and three or four writes
02:24:19 as it got really big.
02:24:20 And that had the beneficial aspect
02:24:22 of just when you needed the performance most
02:24:24 when things are covering the screen,
02:24:25 it was giving you the most acceleration for that.
02:24:28 By the way, were you able to understand this
02:24:32 through thinking about it or were you testing
02:24:34 like the right speed and like that?
02:24:36 This again comes back to, I can find the antecedents
02:24:40 for things like this.
02:24:41 So back in the Apple II days,
02:24:45 the graphics were essentially single bits at a time.
02:24:49 And if you wanted to make your little spaceship,
02:24:51 if you wanted to make it smoothly go across the world,
02:24:54 if you just took the image and you drew it out
02:24:56 at the next location, you would move by seven pixels
02:24:59 at a time, so it’d go chunk, chunk, chunk.
02:25:01 If you wanted to make it move smoothly,
02:25:03 you actually had to make seven versions of the ship
02:25:05 that were pre shifted.
02:25:07 You could write a program that would shift it dynamically,
02:25:09 but on a one megahertz processor,
02:25:11 that’s not going anywhere fast.
02:25:13 So if you wanted to do a smooth moving fast action game,
02:25:16 you made separate versions of each of these sprites.
02:25:20 Now there were a few more tricks you could pull
02:25:22 that if it still wasn’t fast enough,
02:25:24 you could make a compiled shape
02:25:26 where instead of this program that normally copies an image
02:25:31 and says like, get this byte from here, stored here,
02:25:33 get this byte, store this byte.
02:25:35 If you’ve got the memory space, you could say,
02:25:38 I’m going to write the program
02:25:39 that does nothing but draw this shape.
02:25:41 It’s going to be like,
02:25:43 I’m going to load the immediate value 25,
02:25:46 which is some bit pattern,
02:25:47 and then I’m going to store that at this location.
02:25:51 Rather than loading something from memory
02:25:53 that involved indexing registers and this other slow stuff,
02:25:56 you could go ahead and say,
02:25:57 no, I’m gonna hard code the exact values
02:26:00 of all of the image right into the program.
02:26:02 And this was always a horrible trade off there,
02:26:04 which you didn’t have much memory
02:26:05 and you didn’t have much speed.
02:26:07 But if you had something that you wanted to go really fast,
02:26:09 you could turn it into a program.
02:26:11 And that was, knowing about that technique
02:26:14 is what made me think about some of these unwinding it
02:26:17 for the PC where people that didn’t come
02:26:19 from that background were less likely to think about that.
02:26:23 I mean, there’s some deep parallels
02:26:25 probably to human cognition as well.
02:26:27 There’s something about optimizing and compressing
02:26:34 the processing of a new information
02:26:36 that requires you to predict the possible ways
02:26:39 in which the game or the world might unroll.
02:26:44 And you have something like compiled scaler is always there.
02:26:47 So you have like, like you have a prediction
02:26:51 of how the world will unroll
02:26:52 and you have some kind of optimized data structure
02:26:57 for that prediction.
02:26:58 And then you can modify if the world turns out
02:27:00 to be different, you can modify a slight way.
02:27:02 And as far as building out techniques,
02:27:04 so much of the brain is about the associative context.
02:27:06 You know, they’re just, when you learn something,
02:27:09 it’s in the context of something else
02:27:10 and you can have faint, tiny little hints of things.
02:27:13 And I do think there are some deep things
02:27:16 around like sparse distributed memories and boosting.
02:27:18 That’s like, if you can just be slightly
02:27:20 above the noise floor of having some hint of something,
02:27:23 you can have things refined into pulling the memory back up.
02:27:26 So having a, being a programmer and having a toolbox
02:27:29 of like all of these things that,
02:27:31 things that I did in all of these previous lives
02:27:33 of programming tasks that still matters to me
02:27:36 about how I’m able to pull up some of these things.
02:27:38 Like in that case, it was something I did on the Apple 2
02:27:41 then being relevant for the PC.
02:27:43 And I have still cases when I would,
02:27:46 when I would work on mobile development then be like,
02:27:48 okay, I did something like this back in the doom days,
02:27:51 but now it’s a different environment,
02:27:53 but I have still had that tie.
02:27:55 I can bring it in and I can transform it
02:27:56 into what the world needs right now.
02:27:59 And I do think that’s actually one of the very core things
02:28:01 with human cognition and brain like,
02:28:04 you know, brain like functioning
02:28:06 is finding these ways about you’ve got,
02:28:08 your brain is kind of everything everywhere all at once.
02:28:10 You know, it’s, it is just a set of all of this stuff
02:28:13 that is just fetched back by these queries that go into it.
02:28:16 And they can just be slightly above the noise floor
02:28:18 with a random noise in your neurons and synapses
02:28:21 that are affecting exactly what gets pulled up.
02:28:23 So you’re saying some of these very specific solutions
02:28:26 for different games,
02:28:28 you find that there’s a kernel of a deep idea
02:28:31 that’s generalizable to other, to other things.
02:28:34 Yeah, you can’t predict what it’s going to be,
02:28:36 but that idea of like,
02:28:37 I called out that compiled shaders in the forward
02:28:40 that I wrote for that, the game engine black book,
02:28:43 as you know, this is,
02:28:44 it’s kind of an end point of unrolling code,
02:28:48 but that’s one of those things that thinking about that
02:28:50 and having that in your mind.
02:28:51 And I’m sure there are some programmers
02:28:53 that, you know, hear about that.
02:28:54 Think about it a little bit.
02:28:55 It’s kind of the mind blown moment.
02:28:57 It’s like, oh, you can just turn all of that data into code.
02:29:00 And nowadays, you know, you have instruction cache issues
02:29:03 and that’s not necessarily the best idea,
02:29:05 but there are different,
02:29:07 it’s an idea that has power
02:29:09 and has probably relevance in some other areas.
02:29:11 Maybe it’s in a hardware point of view
02:29:13 that there’s a way you approach building hardware
02:29:15 that has that same,
02:29:16 you don’t even have to think about iterating.
02:29:18 You just bake everything all the way into it in one place.
02:29:22 What is the story of how you came to program Doom?
02:29:25 What are some memorable technical challenges
02:29:27 or innovations within that game?
02:29:29 So the path that we went after Wolfenstein got out
02:29:33 and we were on this crazy arc
02:29:35 where Keen one through three, more success than we thought,
02:29:38 Keen four through six, even more success,
02:29:41 Wolfenstein even more success.
02:29:42 So we were on this crazy trajectory for things.
02:29:46 So actually our first box commercial project
02:29:48 was the Commander Keen game,
02:29:50 but then Wolfenstein was going to have a game called
02:29:52 Spear of Destiny, which was a commercial version,
02:29:55 60 new levels.
02:29:57 So the rest of the team took the game engine
02:29:59 pretty much as it was and started working on that.
02:30:02 We got new monsters,
02:30:03 but it’s basically reskins of the things there.
02:30:07 And there’s a really interesting aspect about that
02:30:09 that I didn’t appreciate until much, much later
02:30:12 about how Wolfenstein clearly did tap out its limit
02:30:16 about what you want to play,
02:30:18 all the levels and a couple of our license things.
02:30:21 There was a hard creative wall
02:30:24 that you did not really benefit much
02:30:25 by continuing to beat on it.
02:30:27 But a game like Doom and other more modern games
02:30:31 like Minecraft or something,
02:30:33 there’s kind of a Turing completeness level
02:30:35 of design freedom that you get in games
02:30:37 that Wolfenstein clearly sat on one side of.
02:30:40 All the creative people in the world
02:30:42 could not go and do a masterpiece
02:30:43 just with the technology that Wolfenstein had.
02:30:46 Wolfenstein could do Wolfenstein,
02:30:48 but you really couldn’t do something crazy and different.
02:30:50 But it didn’t take that much more capability
02:30:53 to get to Wolfenstein with the freeform lines
02:30:56 and a little bit more artistic freedom
02:30:58 to get to the point where people still announce
02:31:00 new Doom levels today, all these years after
02:31:03 without having completely tapped out the creativity.
02:31:06 How did you put it?
02:31:07 Turing complete?
02:31:08 Yeah, Turing complete design space.
02:31:10 Design space.
02:31:11 Where it’s like, you know,
02:31:11 we have the kind of computational universality
02:31:14 on a lot of things and how different
02:31:16 subsurface work. For creativity.
02:31:17 But yeah, there’s things where a box can be too small,
02:31:20 but above a certain point,
02:31:22 you kind of are at the point where you really have
02:31:26 almost unbounded creative ability there.
02:31:28 And Doom was the first time you crossed that line.
02:31:31 Yeah, where there were thousands of Doom levels created,
02:31:35 and some of them still have something new and interesting
02:31:37 to say to the world about it.
02:31:38 Is that line, can you introspect what that line was?
02:31:43 Is it in the design space?
02:31:44 Is it something about the programming capabilities
02:31:48 that you were able to add to the game?
02:31:50 So the graphics fidelity was a necessary part
02:31:54 because the block limitations in Wolfenstein,
02:31:57 what we had right there was not enough.
02:32:00 The full scale blocks, although Minecraft really did show
02:32:04 that perhaps blocks stacked in 3D
02:32:07 and at one quarter of the scale of that,
02:32:09 or one eighth in volume,
02:32:11 is then sufficient to have all of that.
02:32:13 But the wall sized blocks that we had in Wolfenstein
02:32:16 was too much of a creative limitation.
02:32:18 We licensed the technology to a few other teams.
02:32:21 None of them made too much of a dent with that.
02:32:24 It just wasn’t enough creative ability.
02:32:27 But a little bit more,
02:32:28 whether it was the variable floors and ceilings
02:32:31 and arbitrary angles in Doom,
02:32:33 or the smaller voxel blocks in Minecraft,
02:32:37 is then enough to open it up
02:32:38 to just worlds and worlds of new capabilities.
02:32:41 What is binary space partitioning?
02:32:45 So the. Which is one of the technologies.
02:32:47 Yeah, so jump around a little bit on the story path there.
02:32:51 So while the team was working on Spirit Destiny
02:32:53 for Wolfenstein, we had met another development team,
02:32:57 Raven Software, while we were in Wisconsin.
02:32:59 And they were doing, they had RPG background
02:33:03 and I still kind of loved that.
02:33:04 And I offered to do a game engine for them
02:33:07 to let them do a 3D rendered RPG
02:33:10 instead of the, like most RPG games were kind of hand drawn.
02:33:14 They made it look kind of 3D,
02:33:15 but it was done just all with artist work
02:33:17 rather than a real engine.
02:33:19 And after Wolfenstein, this was still a tile based world,
02:33:23 but I added floors and ceilings and some lighting
02:33:25 and the ability to have some sloped floors
02:33:27 in different areas.
02:33:28 And that was my intermediate step
02:33:30 for a game called Shadowcaster.
02:33:32 And it had slowed down enough.
02:33:34 It was not fast enough to do our type of action things.
02:33:37 So they had the screen crop down a little bit.
02:33:39 So you couldn’t go the full screen width
02:33:42 like we would try to do in Wolfenstein, but I learned a lot.
02:33:46 I got the floors and ceilings and lightings
02:33:47 and it looked great.
02:33:48 They were great artists up there.
02:33:50 And it was an inspiration for us
02:33:52 to look at some of that stuff.
02:33:54 But I had learned enough from that,
02:33:57 that I had the plan for,
02:33:58 I knew faster ways to do the lighting and shadowing.
02:34:01 And I wanted to do this free form geometry.
02:34:03 I wanted to break out of this tile based
02:34:06 90 degree world limitations.
02:34:09 So we had, that was when we got our next stations
02:34:12 and we were working with these higher powered systems.
02:34:14 And we built an editor that let us draw
02:34:18 kind of arbitrary line segments.
02:34:20 And I was working hard to try to make something
02:34:22 that could render this fast enough.
02:34:26 I was pushing myself pretty hard.
02:34:28 And we were at a point where we could see some things
02:34:32 that looked amazingly cool,
02:34:33 but it wasn’t really fast enough for the way I was doing it.
02:34:37 For this flexibility, it was no longer,
02:34:39 I couldn’t just ray cast into it.
02:34:41 And I had these very complex sets of lines
02:34:43 and simple little worlds were okay.
02:34:45 But the cool things that we wanted to do
02:34:47 just weren’t quite fast enough.
02:34:49 And I wound up taking a break at that point.
02:34:52 And I did the port.
02:34:54 I did two ports of our games,
02:34:58 Wolfenstein to the Super Nintendo.
02:35:01 It was a crazy difficult thing to do,
02:35:04 which was an even slower processors,
02:35:06 like a couple of megahertz processor.
02:35:10 And it had been this whole thing
02:35:11 where we had farmed out the work and it wasn’t going well.
02:35:17 And I took it back over
02:35:18 and trying to make it go fast on there
02:35:21 where it really did not have much processing power.
02:35:25 The pixels were stretched up hugely
02:35:26 and it was pretty ugly when you looked at it.
02:35:29 But in the end, it did come out fast enough to play
02:35:31 and still be kind of fun from that.
02:35:33 But that was where I started using BSP trees
02:35:36 or binary space partitioning trees.
02:35:38 It was one of those things I had to make it faster there.
02:35:41 It was a stepping stone where it was reasonably easy
02:35:44 to understand in the grid world of Wolfenstein
02:35:46 where it was all still 90 degree angles.
02:35:49 BSP trees were, I eased myself into it with that.
02:35:53 And it was a big success.
02:35:56 Then when I came back to working on Doom,
02:35:58 I had this new tool in my toolbox.
02:36:00 It was gonna be a lot harder
02:36:02 with the arbitrary angles of Doom.
02:36:03 This was where I really started grappling
02:36:06 with epsilon problems.
02:36:07 And just up until that point,
02:36:09 I hadn’t really had to deal with the fact
02:36:11 that I am so many numeric things.
02:36:13 This almost felt like a betrayal to me
02:36:15 where people had told me that I had mathematicians
02:36:17 up on a bit of a pedestal where I was,
02:36:20 people think I’m a math wizard and I’m not.
02:36:22 I really, everything that I did was really done
02:36:25 with a solid high school math understanding.
02:36:29 Algebra two trigonometry and that was what got me
02:36:33 all the way through Doom and Quake and all of that,
02:36:35 of just understanding basics of matrices
02:36:38 and knowing it well enough to do something with it.
02:36:41 What’s the epsilon problems you ran into?
02:36:42 So when you wind up taking like a sloped line
02:36:46 and you say, I’m going to intersect it
02:36:48 with another sloped line,
02:36:50 then you wind up with something
02:36:52 that’s not going to be on these nice grid boundaries.
02:36:54 With the Wolfenstein tile maps,
02:36:57 all you’ve got is horizontal and vertical lines
02:36:58 looking at it from above.
02:37:00 And if you cut one of them, it’s just obvious
02:37:02 the other one gets cut exactly at that point.
02:37:04 But when you have angled lines,
02:37:06 you’re doing a kind of a slope intercept problem
02:37:08 and you wind up with rational numbers there
02:37:11 where things that are not going to evenly land on an integer
02:37:14 or even on any fixed point value that you’ve got.
02:37:17 So everything winds up having to snap
02:37:19 to some fixed point value.
02:37:20 So the lines slightly change their angle.
02:37:23 You wind up, if you cut something here,
02:37:25 this one’s going to bend a little this way
02:37:26 and it’s not going to be completely straight.
02:37:28 And then you come down to all these questions of,
02:37:30 well, this one is a point on an angled line.
02:37:35 You can’t answer that in finite precision
02:37:38 unless you’re doing something with actual rational numbers.
02:37:41 And later on, I did waste far too much time
02:37:43 chasing things like that.
02:37:44 How do you do precise arithmetic with rational numbers?
02:37:46 And it always blows up eventually, exponentially
02:37:49 as you do it.
02:37:50 So these kinds of things are impossible with computers.
02:37:53 So they’re possible.
02:37:55 Again, there are paths to doing it,
02:37:57 but you can’t fit them conveniently
02:37:59 in any of the numbers.
02:38:00 You need to start using big nums
02:38:01 and different factor trackings of different things.
02:38:04 So you have to, if you have any elements of OCD
02:38:08 and you want to do something perfectly,
02:38:09 you’re screwed if you’re working with floating point.
02:38:12 Yeah.
02:38:13 So you had to deal with this for the first time.
02:38:15 And there were lots of challenges there about like,
02:38:18 okay, they build this cool thing.
02:38:20 And the way the BSP trees work is it basically
02:38:22 takes the walls and it carves other walls by those walls
02:38:26 in this clever way that you can then
02:38:28 take all of these fragments.
02:38:30 And then you can for sure, from any given point,
02:38:33 get an ordering of everything in the world.
02:38:35 And you can say, this goes in front of this,
02:38:36 goes in front of this, all the way back to the last thing.
02:38:40 And that’s super valuable for graphics
02:38:42 where kind of a classic graphics algorithm
02:38:45 would be painter’s algorithm.
02:38:46 You paint the furthest thing first,
02:38:48 and then the next thing, and then the next thing,
02:38:49 and then it comes up and it’s all perfect for you.
02:38:52 That’s slow because you don’t want to have to have drawn
02:38:54 everything like that,
02:38:56 but you can also flip it around
02:38:57 and draw the closest thing to you.
02:38:59 And then if you’re clever about it,
02:39:01 you can figure out what you need to draw
02:39:03 that’s visible beyond that.
02:39:05 And that’s what BSP trees allow you to do.
02:39:07 Yeah, so it’s combined with a bunch of other things,
02:39:10 but it gives you that ordering.
02:39:12 It’s a clever way of doing things.
02:39:13 And I remember I had learned this
02:39:15 from one of my graphics Bible at the time,
02:39:18 a book called Foley and Van Damme.
02:39:20 And again, it was a different world back there.
02:39:22 There was a small integer number of books.
02:39:24 And this book that was big fat college textbook
02:39:29 that I had read through many times.
02:39:32 I didn’t understand everything in it.
02:39:34 Some of it wasn’t useful to me,
02:39:35 but they had the little thing about finite orderings
02:39:39 of you draw a little T shaped thing
02:39:41 and you can say you can make a fixed ahead of time order
02:39:44 from this and you can generalize this with the BSP trees.
02:39:48 And I got a little bit more information about that.
02:39:50 And it was kind of fun later while I was working on Quake,
02:39:53 I got to meet Bruce Naylor,
02:39:54 who is one of the original researchers
02:39:56 that developed those technologies for academic literature.
02:40:00 And that was kind of fun,
02:40:01 but I was very much just finding a tool
02:40:03 that can help me solve what I was doing.
02:40:05 And I was using it in this very crude way
02:40:07 in a two dimensional fashion, rather than the general 3D.
02:40:10 The Epsilon problems got much worse
02:40:12 in Quake and three dimensionals
02:40:13 when things angle in every way.
02:40:15 But eventually I did sort out
02:40:18 how to do it reliably on Doom.
02:40:20 There were still a few edge cases in Doom
02:40:22 that were not absolutely perfect
02:40:24 where they even got terminologies in the communities.
02:40:27 Like when you got to something where it was messed up,
02:40:29 it was a hall of mirrors effect
02:40:30 because you’d sweep by and it wouldn’t draw something there.
02:40:33 And you would just wind up with the leftover remnants
02:40:36 as you flipped between the two pages.
02:40:39 But BSP trees were important for it.
02:40:41 But it’s again worth noting that after we did Doom,
02:40:45 our major competition came from Ken Silverman
02:40:49 and his build engine, which was used for Duke Nukem 3D
02:40:51 and some of the other games for 3D Realms.
02:40:54 And he used a completely different technology,
02:40:56 nothing to do with BSP trees.
02:40:59 So there’s not just a one true way of doing things.
02:41:03 There were critical things about
02:41:05 to make any of those games fast.
02:41:07 You had to separate your drawing into,
02:41:09 you drew vertical lines and you drew horizontal lines,
02:41:12 just kind of changing exactly
02:41:14 what you would draw with them.
02:41:15 That was critical for the technologies at that time.
02:41:19 And like all the games that were kind of like that
02:41:21 wound up doing something similar,
02:41:23 but there were still a bunch of other decisions
02:41:25 that could be made.
02:41:26 And we made good enough decisions on everything on Doom.
02:41:30 We brought in multiplayer significantly
02:41:33 and it was our first game that was designed
02:41:35 to be modified by the user community
02:41:37 where we had this whole setup of our WAD files and PWADs
02:41:41 and things that people could build with tools
02:41:43 that we released to them.
02:41:44 And they eventually rewrote to be better
02:41:46 than what we released, but they could build things
02:41:48 and you could add it to your game
02:41:50 without destructively modifying it,
02:41:52 which is what you had to do in all the early games.
02:41:54 You literally hacked the data files or the executable
02:41:57 before while Doom was set up in this flexible way
02:42:00 so that you could just say,
02:42:02 run the normal game with this added on on top
02:42:04 and it would overlay just the things
02:42:06 that you wanted to there.
02:42:09 Would you say that Doom was kind of the first
02:42:11 true 3D game that you created?
02:42:14 So no, it’s still, Doom would usually be called
02:42:16 a two and a half D game
02:42:17 where it had three dimensional points on it.
02:42:20 And this is another one of these kind of pedantic things
02:42:22 that people love to argue about,
02:42:23 about what was the first 3D game.
02:42:25 I still, like every month probably I hear from somebody
02:42:29 about, well, was Doom really a 3D game or something?
02:42:34 And I give the point where characters had three coordinates.
02:42:38 So you had like an X, Y and Z,
02:42:40 the cacodamon could be coming in very high
02:42:42 and come down towards you.
02:42:45 The walls had three coordinates on them.
02:42:47 So on some sense it’s a 3D game engine,
02:42:50 but it was not a fully general 3D game engine.
02:42:53 You could not build a pyramid in Doom
02:42:56 because you couldn’t make a sloped wall,
02:42:59 which was slightly different
02:43:00 where in that previous shadow caster game,
02:43:02 I could have vertexes and have a sloped floor there,
02:43:05 but the changes that I made for Doom to get higher speed
02:43:08 and a different set of flexibility traded away that ability,
02:43:11 but you literally couldn’t make that.
02:43:13 You could not, you could make different heights of passages,
02:43:17 but you could not make a bridge over another area.
02:43:20 You could not go over and above it.
02:43:21 So it still had some 2D limitations to it.
02:43:24 That’s more about the building
02:43:26 versus the actual experience.
02:43:27 Cause the experience is.
02:43:29 It felt like things would come at you,
02:43:30 but again, you couldn’t look up either.
02:43:32 I am, you know, you could only pitch.
02:43:34 It was four degrees of freedom
02:43:36 rather than six degrees of freedom.
02:43:38 You did not have the ability to tilt your head this way
02:43:40 or pitch up and down.
02:43:42 So that takes us to Quake.
02:43:44 What was the leap there?
02:43:47 What was some fascinating technical challenges
02:43:50 and there were a lot or not challenges,
02:43:52 but innovations that you’ve come up with.
02:43:54 So Quake was kind of the first thing
02:43:56 where I did have to kind of come face to face
02:43:59 with my limitations,
02:44:01 where it was the first thing
02:44:02 where I really did kind of give it my all
02:44:05 and still come up, you know, come up a little bit short
02:44:08 in terms of what and when I wanted to get it done.
02:44:11 And the company ran,
02:44:13 they had some serious stresses through the whole project
02:44:16 and we bit off a lot.
02:44:19 So the things that we set out to do
02:44:21 was it was going to be really a true 3D engine
02:44:24 where it could do six degree of freedom.
02:44:26 You could have all the viewpoints.
02:44:29 You could model anything.
02:44:31 It had a really remarkable new lighting model
02:44:35 with the surface caching and things.
02:44:37 That was one of those where it was starting
02:44:38 to do some things that they weren’t doing
02:44:40 even on the very high end systems.
02:44:43 And it was going to be completely programmable
02:44:46 in the modding standpoint,
02:44:48 where the thing that you couldn’t do in Doom,
02:44:49 you could replace almost all of the media,
02:44:52 but you couldn’t really change the game.
02:44:55 There were still some people that were doing
02:44:57 the hex setting of the executable,
02:44:58 the dehacked things where you could change
02:45:00 a few things about rules
02:45:01 and people made some early capture the flag type things
02:45:04 by hacking the executable,
02:45:06 but it wasn’t really set out to do that.
02:45:08 Quake was going to have its own programming language
02:45:11 that the game was going to be implemented in it.
02:45:13 And that would be able to be overwritten
02:45:14 just like any of the media.
02:45:16 Code was going to be data for that.
02:45:18 And you would be able to have expansion packs
02:45:21 that changed fundamental things and mods and so on.
02:45:24 And the multiplayer was going to be playable
02:45:27 over the internet.
02:45:28 It was going to support a client server
02:45:30 rather than peer to peer.
02:45:32 So we had the possibility of supporting
02:45:34 larger numbers of players in disparate locations
02:45:37 with this full flexibility of the programming overrides
02:45:41 with full six degree of freedom modeling and viewing.
02:45:44 And with this fancy new light mapped
02:45:47 kind of surface caching side.
02:45:49 It was a lot.
02:45:50 And this was one of those things that
02:45:52 if I could go back and tell younger me
02:45:56 to do something differently,
02:45:57 it would have been to split those innovations up
02:45:59 into two phases in two separate games.
02:46:02 It will be phase one and phase two.
02:46:03 So it probably would have been taking
02:46:05 the Doom rendering engine and bringing in
02:46:08 the TCP IP client server.
02:46:10 Focusing on the multiplayer.
02:46:12 And the Quake C or would have been Doom C
02:46:15 programming language there.
02:46:16 So I would have split that into programming language
02:46:19 and networking with the same Doom engine
02:46:21 rather than forcing everybody to go towards
02:46:24 the Quake engine, which really meant getting a Pentium.
02:46:27 While it ran on a 486, it was not a great experience there.
02:46:31 We could have made more people happier
02:46:33 and gotten two games done in 50% more time.
02:46:37 So speaking of people happier, our mutual friend,
02:46:41 Joe Rogan, it seems like the most important moment
02:46:46 of his life is centered around Quake.
02:46:49 So it was a definitive part of his life.
02:46:53 So would he agree with your thinking that they should split?
02:46:59 So he is a person who loves Quake
02:47:01 and played Quake a lot.
02:47:03 Would he agree that you should have done the Doom engine
02:47:06 and focus on the multiplayer for phase one?
02:47:09 Or in your looking back, is the 3D world
02:47:14 that Quake created was also fundamental
02:47:17 to the enriching experience?
02:47:19 You know, I would say that what would have happened
02:47:21 is you would have had a Doom looking
02:47:25 but Quake feeling game eight months earlier
02:47:29 and then maybe six months after Quake actually shipped,
02:47:33 then there would have been the full running on a Pentium,
02:47:36 six degree of freedom graphics engine type things there.
02:47:38 So it’s not that it wouldn’t have been there.
02:47:42 It would have been something amazingly cool earlier
02:47:45 and then something even cooler somewhat later
02:47:47 where I would much rather have gone
02:47:50 and done two one year development efforts.
02:47:53 I’ve cycled them through.
02:47:54 I’ve been a little more pragmatic about that
02:47:57 rather than killing us ourselves on the whole Quake
02:48:00 development.
02:48:01 But I would say it’s obviously things worked out well
02:48:04 in the end, but looking back and saying,
02:48:06 how would I optimize and do things differently?
02:48:08 That did seem to be a clear case where I going ahead
02:48:13 and we had enormous momentum on Doom.
02:48:15 We did Doom two as the kind of commercial boxed version
02:48:19 after our shareware success with the original,
02:48:22 but we could have just made another Doom game
02:48:25 adding those new features in.
02:48:27 It would have been huge.
02:48:28 We would have learned all the same lessons, but faster.
02:48:31 And it would have given six degree of freedom
02:48:34 and Pentium class systems a little bit more time
02:48:37 to get mainstream because we did cut out a lot of people
02:48:40 with the hardware requirements for Quake.
02:48:42 Was there any dark moments for you personally,
02:48:44 psychologically in having such harsh deadlines
02:48:51 and having to solve so many difficult technical challenges?
02:48:54 So I’ve never really had really dark black places.
02:49:00 I mean, I can’t necessarily put myself
02:49:02 in anyone else’s shoes,
02:49:03 but I understand a lot of people have significant challenges
02:49:09 with kind of their mental health and wellbeing.
02:49:12 And I’ve been super stressed.
02:49:15 I’ve been unhappy as a teenager in various ways,
02:49:18 but I’ve never really gone to a very dark place.
02:49:23 I just seem to be largely immune to what really wrecks people.
02:49:29 I mean, I’ve had plenty of time when I’m very unhappy
02:49:32 and miserable about something,
02:49:33 but it’s never hit me like,
02:49:36 I believe it winds up hitting some other people.
02:49:38 I’ve borne up well under whatever stresses
02:49:41 have kind of fallen on me.
02:49:44 And I’ve always coped best on that
02:49:46 when all I need to do is usually
02:49:49 just kind of bear down on my work.
02:49:51 I pull myself out of whatever hole I might be slipping into
02:49:55 by actually making progress.
02:49:57 I mean, maybe if I was in a position
02:49:59 where I was never able to make that progress,
02:50:01 I could have slid down further,
02:50:03 but I’ve always been in a place where,
02:50:06 okay, a little bit more work,
02:50:07 maybe I’m in a tough spot here,
02:50:09 but I always know if I just keep pushing,
02:50:12 eventually I break through and I make progress,
02:50:14 I feel good about what I’m doing.
02:50:17 And that’s been enough for me so far in my life.
02:50:20 Have you seen it in the distance,
02:50:23 like ideas of depression or contemplating suicide?
02:50:28 Have you seen those things far?
02:50:30 So what was interesting, when I was a teenager,
02:50:33 I was probably on some level a troubled youth.
02:50:37 I was unhappy most of my teenage years.
02:50:40 I really, I wanted to be on my own
02:50:42 doing programming all the time.
02:50:44 As soon as I was 18, 19, even though I was poor,
02:50:47 I was doing exactly what I wanted and I was very happy,
02:50:50 but high school was not a great time for me.
02:50:53 And I had a conversation with like the school counselor
02:50:56 and they’re kind of running their script.
02:50:58 It’s like, okay, it’s kind of a weird kid here.
02:51:00 Let’s carefully probe around.
02:51:02 It’s like, do you ever think about ending it all?
02:51:05 I’m like, no, of course not.
02:51:07 Never, not at all.
02:51:08 This is temporary, things are going to be better.
02:51:13 And that’s always been kind of the case for me.
02:51:15 And obviously that’s not that way for everyone
02:51:18 and other people do react differently.
02:51:20 What was your escape from the troubled youth,
02:51:24 like music, video games, books?
02:51:33 How did you escape from a world
02:51:35 that’s full of cruelty and suffering and that’s absurd?
02:51:38 Yeah, I mean, I was not a victim of cruelty and suffering.
02:51:41 It’s like, I was an unhappy, somewhat petulant youth
02:51:44 in my point where I’m not putting myself up
02:51:48 with anybody else’s suffering,
02:51:49 but I was unhappy objectively.
02:51:52 And the things that I did
02:51:54 that very much characterized my childhood
02:51:57 were I had books, comic books,
02:52:00 Dungeons and Dragons, arcade games, video games.
02:52:03 Like some of my fondest childhood memories
02:52:06 are the convenience stores, the 711s and Quick Trips,
02:52:08 because they had a spinner rack of comic books
02:52:11 and they had a little side room
02:52:12 with two or three video games, arcade games in it.
02:52:15 And that was very much my happy place.
02:52:18 If I could, I get my comic books
02:52:20 and if I could go to a library
02:52:22 and go through those little zero, zero, zero section
02:52:25 where computer books were supposed to be.
02:52:27 And there were a few sad little books there,
02:52:28 but still just being able to sit down and go through that.
02:52:31 And I read a ridiculous number of books,
02:52:35 both fiction and nonfiction as a teenager.
02:52:38 And my rebelling in high school
02:52:42 was just sitting there with my nose in a book,
02:52:44 ignoring the class through lots of it.
02:52:46 And teachers had a range of reactions to that,
02:52:49 some more accepting of it than others.
02:52:53 I’m with you on that.
02:52:54 So let us return to Quake for a bit
02:52:56 with the technical challenges.
02:52:57 What everything together from the networking
02:53:03 to the graphics, what are some things you remember
02:53:07 that were innovations you had to come up with
02:53:10 in order to make it all happen?
02:53:12 Yeah, so there were a bunch of things on Quake
02:53:14 where on the one hand,
02:53:16 the idea that I built my own programming language
02:53:19 to implement the game in.
02:53:20 Looking back, and I try to tell people,
02:53:22 it’s like every high level programmer
02:53:25 sometime in their career goes through
02:53:26 and they invent their own language.
02:53:28 It just seems to be a thing that’s pretty broadly done.
02:53:30 People will be like,
02:53:31 I’m gonna go write a computer programming language.
02:53:33 And I don’t regret having done it,
02:53:37 but after that, I switched from Quake C,
02:53:40 my quirky little pseudo object oriented
02:53:43 or entity oriented language there.
02:53:45 Quake two went back to using DLLs with C
02:53:47 and then Quake three,
02:53:48 I implemented my own C interpreter or compiler,
02:53:51 which was a much smarter thing to do
02:53:53 that I should have done originally for Quake.
02:53:55 But building my own language was an experience.
02:53:57 I learned a lot from that.
02:53:59 And then there was a generation of game programmers
02:54:02 that learned programming with Quake C,
02:54:04 which I feel kind of bad about,
02:54:05 because we give JavaScript a lot of crap,
02:54:08 but Quake C was nothing to write home about there.
02:54:13 But it allowed people to do magical things.
02:54:15 You get into programming,
02:54:16 not because you love the BNF syntax of a language,
02:54:21 it’s because the language lets you do something
02:54:23 that you cared about.
02:54:24 And here’s very much,
02:54:25 you could do something
02:54:26 in a whole beautiful three dimensional world.
02:54:29 And the idea and the fact that the code
02:54:31 for the game was out there,
02:54:32 you could say, I liked the shotgun,
02:54:34 but I want it to be more bad ass.
02:54:36 You go in there and say,
02:54:37 okay, now it does 200 points damage.
02:54:39 And then you go around with a big grin on your face,
02:54:41 blowing up monsters all over the game.
02:54:43 So yeah, it is not what I would do today
02:54:48 going back with that language,
02:54:49 but that was a big part of it.
02:54:51 Learning about the networking stuff,
02:54:54 because it’s interesting where I learned these things
02:54:56 by reading books.
02:54:57 So I would get a book on networking and find something,
02:55:00 I read all about it and learn, okay, packets,
02:55:02 they can be out of order or lost or duplicated.
02:55:06 These are all the things
02:55:07 that can theoretically happen to packets.
02:55:09 So I wind up spending all this time thinking about
02:55:11 how do we deal about all of that?
02:55:13 And it turns out, of course, in the real world,
02:55:15 those are things that yes,
02:55:16 theoretically can happen with multiple routes,
02:55:18 but they really aren’t things that your 99.999%
02:55:22 of your packets have to deal with.
02:55:25 So there was learning experiences about lots of that.
02:55:28 Like why, when TCP is appropriate versus UDP
02:55:32 and how if you do things in UDP,
02:55:34 you wind up reinventing TCP badly in almost all cases.
02:55:37 So there’s good arguments for using both
02:55:41 for different game technology,
02:55:42 different parts of the game process,
02:55:44 transitioning from level to level and all.
02:55:46 But the graphics were the showcase
02:55:49 of what Quake was all about.
02:55:51 It was this graphics technology that nobody had seen there.
02:55:55 And it was a while before
02:55:57 there were competitive things out there.
02:55:59 And it went a long time internally really not working
02:56:03 where we were even building levels
02:56:05 where the game just was not at all shippable
02:56:09 with large fractions of the world,
02:56:11 like disappearing, not being there,
02:56:14 or being really slow in various parts of it.
02:56:16 And it was this act of faith.
02:56:18 It’s like, I think I’m gonna be able to fix this.
02:56:20 I think I’m gonna be able to make this work.
02:56:23 And lots of stuff changed
02:56:25 where the level designers would build something
02:56:27 and then have to throw it away as something fundamental
02:56:29 and the kind of graphics or level technology changed.
02:56:33 And so there were two big things
02:56:37 that contributed to making it possible at that timeframe.
02:56:41 Two new things.
02:56:42 There was certainly
02:56:43 hardcore optimized low level assembly language.
02:56:46 And this was where I had hired Michael Abrash
02:56:48 away from Microsoft.
02:56:50 And he had been one of my early inspirations
02:56:52 where back in the soft disk days,
02:56:54 the library of magazines that they had,
02:56:57 some of my most treasured ones
02:56:58 were Michael Abrash’s articles in Dr. Dobbs journal.
02:57:02 And it was amazing after all of our success in Doom,
02:57:06 we were able to kind of hit him up and say,
02:57:08 hey, we’d like you to come work at id Software.
02:57:10 And he was in the senior technical role at Microsoft.
02:57:13 And he was on track for,
02:57:15 and this was right when Microsoft was starting to take off.
02:57:18 And I did eventually convince him
02:57:21 that what we were doing
02:57:22 was gonna be really amazing with Quake.
02:57:24 It was gonna be something nobody had seen before.
02:57:28 It had these aspects of what we were talking about.
02:57:31 We had metaverse talk back then.
02:57:33 We had read Snow Crash and we knew about this.
02:57:36 And Michael was big into the science fiction
02:57:40 and we would talk about all that
02:57:41 and kind of spin this tale.
02:57:42 And it was some of the same conversations
02:57:45 that we have today about the metaverse,
02:57:46 about how you could have different areas
02:57:48 linked together by portals
02:57:50 and you could have user generated content
02:57:52 and changing out all of these things.
02:57:54 So you really were creating the metaverse with Quake.
02:57:56 And we talked about things like,
02:57:58 Duke used to be advertised as a virtual reality experience.
02:58:02 That was the first wave of virtual reality
02:58:05 was in the late 80s and early 90s,
02:58:07 you had like the Lawn Mower Man movie
02:58:10 and you had time in Newsweek
02:58:12 talking about the early VPL headsets.
02:58:14 And of course that cratered so hard
02:58:16 that people didn’t wanna look at virtual reality
02:58:18 for decades afterwards,
02:58:20 where it was just, it was smoke and mirrors.
02:58:23 It was not real in the sense
02:58:24 that you could actually do something
02:58:26 real and valuable with it.
02:58:28 But still we had that kind of common set of talking points.
02:58:32 And we were talking about what these games could become
02:58:36 and how you’d like to see people
02:58:37 building all of these creative things.
02:58:39 Because we were seeing an explosion of work
02:58:41 with Doom at that time,
02:58:42 where people were doing amazingly cool things.
02:58:45 Like we saw cooler levels that we had built
02:58:48 coming out of the user community.
02:58:49 And then people finding ways
02:58:51 to change the characters in different ways.
02:58:54 And it was great.
02:58:54 And we knew what we were doing in Quake
02:58:56 was removing those last things.
02:58:58 There was some quirky things
02:59:00 with a couple of the data types
02:59:01 that didn’t work right for overriding.
02:59:03 And then the core thing about the programming model.
02:59:07 And I was definitely going to hit all of those in Quake.
02:59:10 But the graphics side of it was still,
02:59:15 I knew what I wanted to do.
02:59:17 And it was one of these hubris things
02:59:20 where it’s like, well, so far I’ve been able
02:59:21 to kind of kick everything that I set out to go do.
02:59:26 But Quake was definitely a little bit more
02:59:29 than could be comfortably chewed at that point.
02:59:32 And, but Michael was one of the strongest programmers
02:59:36 and graphics programmers that I knew.
02:59:39 And he was one of the people that I trusted
02:59:40 to write assembly code better than I could.
02:59:44 And there’s a few people that I can point to
02:59:46 about things like this where I’m a world class optimizer.
02:59:49 I mean, I make things go fast,
02:59:51 but I recognize there’s a number of people
02:59:54 that can write tighter assembly code,
02:59:56 tighter SIMD code or tighter CUDA code
02:59:59 than I can write.
03:00:02 My best strengths are a little bit more at the system level.
03:00:05 I mean, I’m good at all of that,
03:00:07 but the most leverage comes from making the decisions
03:00:10 that are a little bit higher up
03:00:11 where you figure out how to change your large scale problems
03:00:15 so that these lower level problems are easier to do
03:00:18 or it makes it possible to do them in a uniquely fast way.
03:00:23 So most of my big wins in a lot of ways
03:00:27 from all the way from the early games through VR
03:00:30 and the aerospace work that I’m doing or did,
03:00:33 and hopefully the AI work that I’m working on now
03:00:35 is finding an angle on something
03:00:37 that means you trade off something
03:00:40 that you maybe think you need,
03:00:41 but it turns out you don’t need.
03:00:43 And by making a sacrifice in one place,
03:00:45 you can get big advantages in another place.
03:00:48 Is it clear at which level of the system
03:00:51 those big advantages can be gained?
03:00:54 It’s not always clear.
03:00:55 And that’s why the thing that I try
03:00:58 to make one of my core values,
03:01:00 and I proselytize to a lot of people
03:01:02 is trying to know the entire stack,
03:01:05 trying to see through everything that happens.
03:01:08 And it’s almost impossible
03:01:09 on like the web browser level of things
03:01:11 where there’s so many levels to it,
03:01:13 but you should at least understand what they all are,
03:01:15 even if you can’t understand
03:01:16 all the performance characteristics at each level,
03:01:20 but it goes all the way down to literally the hardware.
03:01:22 So what is this chip capable of?
03:01:26 And what is this software that you’re writing capable of?
03:01:29 And then with this architecture you put on top of that,
03:01:31 then the ecosystem around it,
03:01:33 all the people that are working on it.
03:01:35 So there are all these decisions
03:01:38 and they’re never made in a globally optimal way,
03:01:41 but sometimes you can drive a thread
03:01:43 of global optimality through it.
03:01:45 You can’t look at everything, it’s too complicated,
03:01:47 but sometimes you can step back up
03:01:49 and make a different decision.
03:01:51 And we kind of went through this on the graphic side
03:01:53 on Quake, where in some ways it was kind of bad
03:01:56 where Michael would spend his time writing,
03:01:59 like I’d rough out the basic routines,
03:02:01 like, okay, here’s our span rasterizer.
03:02:03 And he would spend a month writing
03:02:05 this beautiful cycle optimized piece of assembly language
03:02:10 that does what I asked it to do.
03:02:13 And he did it faster than like my original code would do,
03:02:16 or probably what I would be able to do
03:02:17 even if I had spent that month on it.
03:02:20 But then we’d have some cases when I’d be like,
03:02:22 okay, well, I figured out at this higher level
03:02:25 instead of drawing these in a painter’s order here,
03:02:28 I do a span buffer and it cuts out 30% or 40%
03:02:32 of all of these pixels,
03:02:34 but it means you need to rewrite kind of this interface
03:02:36 of all of that.
03:02:37 And I could tell that wore on him a little bit,
03:02:39 but in the end it was the right thing to do
03:02:41 where we wound up changing that rasterization approach
03:02:45 and we wound up with a super optimized
03:02:47 assembly language core loop
03:02:49 and then a good system around it
03:02:51 which minimized how much that had to be called.
03:02:54 And so in order to be able to do
03:02:55 this kind of system level thinking,
03:02:58 whether we’re talking about game development,
03:03:02 aerospace, nuclear energy, AI, VR,
03:03:07 you have to be able to understand the hardware,
03:03:10 the low level software, the high level software,
03:03:14 the design decisions, the whole thing,
03:03:16 the full stack of it.
03:03:18 Yeah, and that’s where a lot of these things
03:03:20 become possible.
03:03:21 When you’re really, when you’re bringing the future forward,
03:03:23 I mean, there’s a pace that everything
03:03:25 just kind of glides towards
03:03:26 where we have a lot of progress
03:03:27 that’s happening at such a different,
03:03:29 so many different ways you kind of slide towards progress
03:03:32 just left to your own, programs just get faster.
03:03:34 For a while it wasn’t clear
03:03:36 if they were gonna get fatter more than they get,
03:03:38 quicker than they get faster and it cancels out,
03:03:40 but it is clear now in retrospect,
03:03:42 no, programs just get faster
03:03:44 and have gotten faster for a long time.
03:03:46 But if you wanna do something
03:03:48 like back at that original talking about scrolling games,
03:03:52 say what, this needs to be five times faster.
03:03:54 Well, we can wait six years
03:03:57 and just it’ll naturally get that much faster at that time
03:04:00 or you come up with some really clever way of doing it.
03:04:03 So there are those opportunities like that
03:04:06 in a whole bunch of different areas.
03:04:08 Now, most programmers don’t need to be thinking about that.
03:04:11 There’s not that many,
03:04:13 there’s a lot of opportunities for this,
03:04:15 but it’s not everyone’s work a day type stuff.
03:04:17 So everyone doesn’t have to know how all these things work.
03:04:20 They don’t have to know how their compiler works,
03:04:22 how the processor chip manages cache eviction
03:04:26 and all these low level things.
03:04:28 But sometimes there are powerful opportunities
03:04:31 that you can look at and say,
03:04:33 we can bring the future five years faster.
03:04:37 We can do something that,
03:04:38 wouldn’t it be great if we could do this?
03:04:40 Well, we can do it today
03:04:42 if we make a certain set of decisions.
03:04:44 And it is in some ways smoke and mirrors
03:04:47 where you say it’s like,
03:04:48 Doom was a lot of smoke and mirrors
03:04:50 where people thought it was more capable
03:04:52 than it actually was,
03:04:53 but we picked the right smoke and mirrors
03:04:56 to deploy in the game where by doing this,
03:04:58 people will think that it’s more general.
03:05:01 We are gonna amaze them with what they’ve got here
03:05:03 and they won’t notice
03:05:05 that it doesn’t do these other things.
03:05:07 So smart decision making at that point,
03:05:09 that’s where that kind of global holistic top down view
03:05:15 can work.
03:05:16 And I’m really a strong believer
03:05:20 that technology should be sitting at that table
03:05:24 having those discussions
03:05:25 because you do have cases where you say,
03:05:26 well, you wanna be the Jonathan Ivy or whatever,
03:05:28 where it’s a pure design solution.
03:05:32 And that’s in some cases now
03:05:35 where you truly have almost infinite resources.
03:05:37 Like if you’re trying to do a scrolling game on the PC now,
03:05:41 you don’t even have to talk to a technology person.
03:05:43 You can just have,
03:05:45 any intern can make that go run as fast as it needs to there
03:05:48 and it can be completely design based.
03:05:50 But if you’re trying to do something that’s hard,
03:05:53 either that can’t be done for resources
03:05:55 like VR on a mobile chip set,
03:05:58 or that we don’t even know how to do yet,
03:05:59 like artificial general intelligence,
03:06:02 it’s probably going to be a matter
03:06:03 of coming at it from an angle.
03:06:05 Like, I mean, for AGI,
03:06:06 we have some of like what are some of the hudder principles
03:06:09 about how you can AXI or some of the,
03:06:11 there are theoretical ways that you can say,
03:06:13 this is the optimal learning algorithm
03:06:15 that can solve everything, but it’s completely impractical.
03:06:18 You just can’t do that.
03:06:20 So clearly you have to make some concessions
03:06:23 for general intelligence
03:06:25 and nobody knows what the right ones are yet.
03:06:27 So people are taking different angles of attack.
03:06:29 I hope I’ve got something clever
03:06:31 to come up with in that space.
03:06:34 It’s been surprising to me.
03:06:35 And I think perhaps it is a principle of progress
03:06:38 that smoke and mirrors somehow
03:06:40 is the way you build the future.
03:06:42 You kind of fake it till you make it
03:06:46 and you almost always make it.
03:06:47 And I think that’s going to be the way we achieve AGI.
03:06:50 That’s going to be the way we build consciousness
03:06:53 into our machines.
03:06:55 There’s philosophers debate about the Turing test
03:06:59 is essentially about faking it till you make it.
03:07:02 You start by faking it.
03:07:04 And I think that always leads to making it
03:07:09 because if we look at history.
03:07:10 Arguments when, as soon as people start talking about
03:07:13 qualia and consciousness and Chinese rooms and things,
03:07:16 it’s like, I just check out.
03:07:17 I just don’t think there’s any value in those conversations.
03:07:20 It’s just like, go ahead, tell me it’s not going to work.
03:07:22 I’m going to do my best to try to make it work anyways.
03:07:25 I don’t know if you work with legged robots.
03:07:26 There’s a bunch of these.
03:07:29 They sure as heck make me feel like they’re cautious
03:07:33 in a certain way that’s not here today,
03:07:37 but is you could see the kernel.
03:07:41 It’s like the flame, the beginnings of a flame.
03:07:45 We don’t have line of sight,
03:07:47 but there’s glimmerings of light in the distance
03:07:50 for all of these things.
03:07:51 Yeah, I’m hearing murmuring in a distant room.
03:07:54 Well, let me ask you a human question here.
03:07:56 You’ve in the game design space,
03:08:00 you’ve done a lot of incredible work throughout,
03:08:01 but in terms of game design, you have changed the world.
03:08:05 And there’s a few people around you that did the same.
03:08:08 So famously there’s some animosity, there’s much love,
03:08:13 but there’s some animosity between you and John Romero.
03:08:16 What is at the core of that animosity and human tension?
03:08:19 So there really hasn’t been for a long time.
03:08:24 And even at the beginning, it’s like,
03:08:25 yes, I did push Romero out of the company.
03:08:29 And this is one of the things that I look back,
03:08:32 if I could go back telling my younger self
03:08:35 some advice about things,
03:08:37 the original founding kind of corporate structure
03:08:41 of id Software really led to a bunch of problems.
03:08:45 We started off with us as equal partners
03:08:48 and we had a buy sell agreement
03:08:50 because we didn’t want outsiders
03:08:51 to be telling us what to do inside the company.
03:08:54 And that did lead to a bunch of the problems
03:08:57 where I was sitting here going,
03:08:59 it’s like, all right, I’m working harder than anyone.
03:09:02 I’m doing these technologies, nobody’s done before,
03:09:06 but we’re all equal partners.
03:09:08 And then I see somebody that’s not working as hard.
03:09:11 And I mean, I can’t say I was the most mature about that.
03:09:16 I was 20 something years old
03:09:18 and it did bother me when I’m like, everybody,
03:09:22 okay, we need to all pull together
03:09:23 and we’ve done it before everybody.
03:09:25 We know we can do this if we get together
03:09:27 and we grind it all out,
03:09:29 but not everybody wanted to do that for all time.
03:09:33 And I was the youngest one of the crowd there.
03:09:35 I had different sets of kind of backgrounds and motivations
03:09:40 and left at that point where it was, all right,
03:09:44 either everybody has to be contributing like up to this level
03:09:48 or they need to get pushed out was not,
03:09:52 that was not a great situation.
03:09:54 And I look back on it and know that we pushed people
03:09:57 out of the company that could have contributed
03:10:00 if there was a different framework for them.
03:10:02 And the modern kind of Silicon Valley,
03:10:04 like let your stock vest over a time period
03:10:07 and maybe it’s non voting stock
03:10:08 and all those different things.
03:10:09 We knew nothing about any of that.
03:10:11 I mean, we didn’t know what we were doing
03:10:13 in terms of corporate structure or anything.
03:10:16 So if you think the framework was different,
03:10:18 some of the human tension could have been a little bit.
03:10:20 It almost certainly would have.
03:10:22 I mean, I look back at that
03:10:24 and it’s like even trying to summon up in my mind,
03:10:27 it’s like, I know I was really, really angry about,
03:10:32 like Romero not working as hard as I wanted him to work
03:10:35 or not carrying his load on the design for Quake
03:10:39 and coming up with things there.
03:10:40 But he was definitely doing things.
03:10:43 He made some of the best levels there.
03:10:44 He was working with some of our external teams
03:10:47 like Raven on the licensing side of things,
03:10:50 but there were differences of opinion about it,
03:10:55 but he landed right on his feet.
03:10:57 He went and he got $20 million from Eidos
03:10:59 to go do Ion Storm and he got to do things his way
03:11:02 and spun up three teams simultaneously
03:11:05 because that was always one of the challenging things
03:11:08 in it where we were doing these single string,
03:11:11 one project after another.
03:11:13 And I think some of them wanted to grow the company more
03:11:16 and I didn’t because I knew people that were saying that,
03:11:19 oh, companies turn to shit when you got 50 employees.
03:11:22 It’s just a different world there.
03:11:24 And I loved our little dozen people working on the projects,
03:11:28 but you can look at it and say,
03:11:29 well, business realities matter.
03:11:31 It’s like, you’re super successful here
03:11:33 and we could take a swing and a miss on something,
03:11:36 but you do it a couple of times and you’re out of luck.
03:11:39 There’s a reason companies try to have multiple teams
03:11:43 running at one time.
03:11:45 And so that was, again,
03:11:47 something I didn’t really appreciate back then.
03:11:49 So if you look past all that,
03:11:51 you did create some amazing things together.
03:11:53 What did you love about John Romero?
03:11:55 What did you respect and appreciate about him?
03:11:57 What did you admire about him?
03:11:59 What did you learn from him?
03:12:00 When I met him, he was the coolest programmer
03:12:02 I had ever met.
03:12:04 He had done all of this stuff.
03:12:05 He had made all of these games.
03:12:07 He had worked at one of the companies
03:12:10 that I thought was the coolest at Origin Systems.
03:12:12 And he knew all this stuff.
03:12:14 He made things happen fast.
03:12:16 And he was also kind of a polymath about this
03:12:18 where he could do, he drew his own art.
03:12:21 He made his own levels,
03:12:22 as well as we worked on sound design systems
03:12:26 on top of actually being a really good programmer.
03:12:29 And we went through a little,
03:12:32 it was kind of fun where one of the early things
03:12:34 that we did where there was kind of the young buck bit
03:12:36 going in where I was the new guy.
03:12:39 And he was the top man programmer at the soft disk area.
03:12:44 And eventually we had sort of a challenge over the weekend
03:12:47 that we were gonna like race to implement this game,
03:12:49 to port one of our PC games back down to the Apple II.
03:12:52 And that was where we finally kind of became clear.
03:12:55 It’s like, okay, Carmack stands a little bit apart
03:12:57 on the programming side of things.
03:13:00 But Romero then very gracefully moved into,
03:13:03 well, he’ll work on the tools, he’ll work on the systems,
03:13:06 do some of the game design stuff,
03:13:07 as well as contributing on,
03:13:09 starting to lead the design aspects of a lot of things.
03:13:12 So he was enormously valuable in the early stuff.
03:13:16 And so much of Doom and even Quake have his stamp on it
03:13:20 in a lot of ways.
03:13:21 But he wasn’t at the same level of focus
03:13:25 that I brought to the work that we were doing there.
03:13:29 And he really did, we hit such a degree of success
03:13:33 that it was all in the press about that.
03:13:35 The rockstar game programmers.
03:13:38 I mean, it’s the Beatles problem.
03:13:40 Yeah, I mean, he ate it up and he did personify.
03:13:43 There was the whole game developers with Ferraris
03:13:46 that we had there.
03:13:49 And I thought that led to some challenges there.
03:13:53 But so much of the stuff that was great in the games
03:13:58 did come from him.
03:13:59 And I would certainly not take that away from him.
03:14:01 And even after we parted ways and he took his swing
03:14:05 with Eidos, in some ways, he was ahead of the curve
03:14:09 with mobile gaming as well, where
03:14:11 one of his companies after Eidos was working on feature phone
03:14:15 game development.
03:14:16 And I wound up doing some of that
03:14:19 just before the iPhone crossing over into the iPhone phase
03:14:22 there.
03:14:23 And that was something that clearly
03:14:24 did turn out to be a huge thing, although he
03:14:27 was too early for what he was working on at that time.
03:14:31 We’ve had pretty cordial relationships
03:14:34 where I was happy to talk with him any time I’d
03:14:35 run into him at a conference.
03:14:38 I have actually had some other people just say,
03:14:40 it’s like, oh, you shouldn’t go over there
03:14:42 and give him the time of day.
03:14:44 Or felt that Masters of Doom played things up in a way
03:14:50 that I shouldn’t be too happy with.
03:14:52 But I’m OK with all of that.
03:14:54 So you’ve still got love in your heart.
03:14:56 Yeah, I mean, I just talked with him like last year,
03:14:59 or I guess it was even this year,
03:15:01 about mentioning that I’m going off doing this AI stuff.
03:15:03 I’m going big into artificial intelligence.
03:15:06 And he had a bunch of ideas for how
03:15:09 AI is going to play into gaming.
03:15:10 And asked if I was interested in collaborating.
03:15:13 And it’s not in line with what I’m doing.
03:15:16 But I do wish almost everyone the best.
03:15:19 I mean, I know I may not have partnered
03:15:21 on the best of terms with some people.
03:15:24 But I was thrilled to see Tom Hall writing VR games now.
03:15:29 He wrote, I’m working on a game called Demio, which
03:15:31 is really an awesome VR game.
03:15:33 It’s like Dungeons and Dragons.
03:15:35 We all used to play Dungeons and Dragons together.
03:15:36 That was one of the things.
03:15:37 That was what we did on Sundays in the early days.
03:15:40 I would Dungeon Master, and they’d all play.
03:15:42 And so it really made me smile seeing
03:15:44 Tom involved with an RPG game in virtual reality.
03:15:49 You were the CTO of Oculus VR since 2013,
03:15:54 and maybe less than a year involved in a bit in 2019.
03:16:00 Oculus was acquired by Facebook Now Meta in 2014.
03:16:04 You’ve spoken brilliantly about both the low level details,
03:16:07 the experimental design, and the big picture
03:16:09 vision of virtual reality.
03:16:12 Let me ask you about the metaverse, the big question
03:16:14 here, both philosophically and technically.
03:16:18 How hard is it to build the metaverse?
03:16:20 What is the metaverse in your view?
03:16:22 You started with discussing and thinking about Quake
03:16:24 as a kind of a metaverse.
03:16:27 As you think about it today, what is the metaverse,
03:16:31 the thing that could create this compelling user value,
03:16:34 this experience that will change the world,
03:16:36 and how hard is it to build it?
03:16:39 So the term comes from Neal Stephenson’s book Snow Crash,
03:16:42 which many of us had read back in the 90s.
03:16:44 It was one of those kind of formative books.
03:16:47 And there was this sense that the possibilities
03:16:53 and kind of the freedom and unlimited capabilities
03:16:56 to build a virtual world that does whatever you want,
03:16:59 whatever you ask of it, has been a powerful draw
03:17:02 for generations of developers, game developers specifically,
03:17:05 and people that are thinking about more general purpose
03:17:08 applications.
03:17:10 So we were talking about that back in the Doom and Quake days
03:17:13 about how do you wind up with an interconnected set of worlds
03:17:17 that you kind of visit from one to another.
03:17:19 And as web pages were becoming a thing,
03:17:21 you start thinking about what is the interactive kind
03:17:25 of 3D based equivalent of this.
03:17:27 And there were a lot of really bad takes.
03:17:29 You had like Vermol and virtual reality markup languages.
03:17:34 And there’s aspects like that that came from people saying,
03:17:38 well, what kind of capabilities should we
03:17:40 develop to enable this?
03:17:43 And that kind of capability first work
03:17:45 has usually not panned out very well.
03:17:48 On the other hand, we have successful games
03:17:51 that started with things like Doom and Quake and communities
03:17:54 that formed around those, whether it
03:17:56 was server lists in the early days
03:17:58 or literal portaling between different games,
03:18:01 and then modern things that are on completely different order
03:18:04 of magnitude like Minecraft and Fortnite that
03:18:07 have 100 million plus users.
03:18:11 I still think that that’s the right way
03:18:13 to go to build the metaverse is you build something that’s
03:18:16 amazing that people love and people wind up
03:18:18 spending all their time in because it’s awesome.
03:18:21 And you expand the capabilities of that.
03:18:24 So even if it’s a very basic experience, if it’s awesome.
03:18:28 Minecraft is an amazing case study in so many things.
03:18:31 That’s basic as it gets.
03:18:32 What’s been able to be done with that is really enlightening.
03:18:36 And there are other cases where, like right now,
03:18:39 Roblox is basically a game construction kit aimed at kids.
03:18:43 And that was a capability first play.
03:18:45 And it’s achieving scale that’s on the same order
03:18:48 of those things.
03:18:49 So it’s not impossible, but my preferred bet
03:18:52 would be you make something amazing that people love
03:18:55 and you make it better and better.
03:18:56 And that’s where I could say we could have gone back
03:18:59 and followed a path like that in the early days
03:19:02 if you just take the same game, whether it’s
03:19:05 when Activision demonstrated that you could make
03:19:07 Call of Duty every year.
03:19:09 And not only is it not bad, people love it.
03:19:12 And it’s very profitable.
03:19:14 The idea that you could have taken something like that,
03:19:17 take a great game, release a new version every year
03:19:20 that lets the capabilities grow and expand
03:19:23 to start saying it’s like, OK, it’s
03:19:24 a game about running around and shooting things,
03:19:26 but now you can bring your media into it.
03:19:30 You can add persistence of social signs of life
03:19:35 or whatever you want to add to it.
03:19:37 I still think that’s quite a good position to take.
03:19:41 And I think that while Meta is doing a bottoms up capability
03:19:45 approach with Horizon Worlds where it’s
03:19:48 a fairly general purpose, creators
03:19:51 can build whatever they want in their sort of thing,
03:19:55 it’s hard to compare and compete with something
03:19:57 like Fortnite, which also has enormous amounts of creativity
03:20:01 even though it was not designed originally
03:20:03 as a general purpose sort of thing.
03:20:05 So we have examples on both sides.
03:20:08 Me personally, I would have bet on trying
03:20:11 to do entertainment, valuable destination first,
03:20:15 and expanding from there.
03:20:16 So can you imagine the thing that will be kind of,
03:20:22 if we look back a couple of centuries from now
03:20:25 and you think about the experiences that
03:20:29 marked the singularity, the transition where most
03:20:34 of our world moved into virtual reality,
03:20:37 what do you think those experiences will look like?
03:20:40 So I do think it’s going to be kind of like the way the web
03:20:43 slowly took over, where you’re the frog
03:20:46 in the pot of water that’s slowly heating up,
03:20:48 where having lived through all of that,
03:20:51 I remember when it was shocking to start
03:20:53 seeing the first website address on a billboard when you’re
03:20:57 like, hey, my computer world is infecting the real world.
03:21:00 This is spreading out in some way.
03:21:03 But when you look back and say, well,
03:21:05 what made the web take off?
03:21:08 And it wasn’t a big bang sort of moment there.
03:21:12 It was a bunch of little things that turned out
03:21:14 not to even be the things that are relevant now
03:21:17 that brought them into it.
03:21:18 So I wonder if, I mean, like you said, you’re not a historian.
03:21:23 So maybe there is a historian out there
03:21:26 that could really identify that moment, data driven way.
03:21:30 It could be like MySpace or something like that.
03:21:33 Maybe the first major social network that really reached
03:21:37 into non geek world or something like that.
03:21:42 I think that’s kind of the fallacy of historians, though,
03:21:45 looking for some of those kind of primary dominant causes
03:21:48 where so many of these things are like we
03:21:51 see an exponential curve.
03:21:52 But it’s not because like one thing is going exponential.
03:21:55 It’s because we have hundreds of little sigmoid curves
03:21:59 overlapped on top of each other.
03:22:00 And they just happen to keep adding up
03:22:02 so that you’ve got something kind of going exponential
03:22:05 at any given point.
03:22:06 But no single one of them was the critical thing.
03:22:09 There were dozens and dozens of things.
03:22:11 I mean, seeing the transitions of stuff
03:22:13 like as obviously MySpace giving way to other things,
03:22:16 but even like blogging giving way to social media
03:22:20 and getting resurrected in other guises.
03:22:22 And the memes with the dancing baby GIF
03:22:26 or whatever the all your base now belong to us.
03:22:29 Whatever those early memes that led to the modern memes
03:22:32 and the different evolution of humor on the internet
03:22:37 that I’m sure the historians will also write books about
03:22:40 from the different website that support
03:22:42 to create the infrastructure for that humor like Reddit
03:22:45 and all that kind of stuff.
03:22:46 So people will go back, and they will name
03:22:48 firsts and critical moments.
03:22:50 But it’s probably going to be a poor approximation of what
03:22:53 actually happens.
03:22:54 And we’ve already seen in the VR space
03:22:57 where it didn’t play out the way we thought it would in terms
03:23:01 of what was going to be like when the modern era of VR
03:23:04 basically started with my E3 demo of Doom 3
03:23:07 on the Rift prototype.
03:23:08 So we’re like first person shooters in VR,
03:23:10 match made in heaven, right?
03:23:12 And that didn’t work out that way at all.
03:23:15 They have the most comfort problems with it.
03:23:18 And then the most popular virtual reality app
03:23:21 is Beat Saber, which nobody predicted back then.
03:23:24 What’s that make you like from first principles
03:23:28 if you were to reverse engineer that?
03:23:31 Why are these silly fun games the most?
03:23:35 It actually makes very clear sense
03:23:37 when you analyze it from hindsight
03:23:40 and look at the engineering reasons
03:23:41 where it’s not just that it was a magical quirky idea.
03:23:45 It was something that played almost perfectly
03:23:47 to what turned out to be the real strengths of VR
03:23:50 where the one thing that I really underestimated
03:23:52 importance in VR was the importance of the controllers.
03:23:55 I was still thinking we could do a lot more
03:23:57 with the game pad and just the amazingness
03:24:00 of taking any existing game,
03:24:01 being able to move your head around and look around,
03:24:04 that that was really amazing.
03:24:06 But the controllers were super important.
03:24:09 But the problem is so many things
03:24:10 that you do with the controllers just suck.
03:24:13 It feels like it breaks the illusion
03:24:14 like trying to pick up glasses with the controllers
03:24:16 where you’re like, oh, use the grip button
03:24:18 when you’re kind of close and it’ll snap into your hand.
03:24:21 All of those things are unnatural actions
03:24:24 that you do them and it’s still part of the VR experience.
03:24:27 But Beat Saber winds up playing only to the strengths.
03:24:32 It completely hides all the weaknesses of it
03:24:34 because you are holding something in your hand.
03:24:37 You keep a solid grip on it the whole time.
03:24:39 It slices through things without ever bumping into things.
03:24:42 You never get into the point where,
03:24:44 I’m knocking on this table,
03:24:45 but in VR, my hand just goes right through it.
03:24:48 So you’ve got something that slices through.
03:24:51 So it’s never your brain telling you,
03:24:53 oh, I should have hit something.
03:24:54 You’ve got a lightsaber here.
03:24:55 It’s just, you expect it to slice through everything.
03:24:58 Audio and music turned out to be a really powerful aspect
03:25:02 of virtual reality where you’re blocking the world off
03:25:05 and constructing the world around you.
03:25:07 And being something that can run efficiently
03:25:10 on even this relatively low powered hardware
03:25:13 and can have a valuable loop in a small amount of time
03:25:17 where a lot of modern games,
03:25:19 you’re supposed to sit down and play it for an hour
03:25:21 just to get anywhere.
03:25:22 Sometimes a new game takes an hour
03:25:23 to get through the tutorial level.
03:25:25 And that’s not good for VR for a couple reasons.
03:25:27 You do still have the comfort issues
03:25:29 if you’re moving around at all,
03:25:31 but you’ve also got just discomfort from the headset,
03:25:34 battery lifespan on the mobile versions.
03:25:37 So having things that do break down
03:25:39 into three and four minute windows of play,
03:25:42 that turns out to be very valuable
03:25:43 from a gameplay standpoint.
03:25:45 So it winds up being kind of a perfect storm
03:25:48 of all of these things that are really good.
03:25:50 It doesn’t have any of the comfort problems.
03:25:52 You’re not navigating around.
03:25:53 You’re standing still.
03:25:54 All the stuff flies at you.
03:25:56 It has placed audio strengths.
03:25:59 It adds the whole fitness in VR.
03:26:01 Nobody was thinking about that back at the beginning.
03:26:04 And it turns out that that is an excellent
03:26:07 daily fitness thing to be doing.
03:26:09 If you go play an hour of Beat Saber
03:26:11 or Supernatural or something,
03:26:13 that is legit solid exercise.
03:26:16 And it’s more fun than doing it
03:26:17 just about any other way there.
03:26:19 So that’s kind of the arcade stage of things.
03:26:23 If I were to say with my experience with VR,
03:26:27 the thing that I think is powerful is the,
03:26:30 maybe it’s not here yet,
03:26:32 but the degree to which it is immersive
03:26:35 in the way that Quake is immersive.
03:26:38 It takes you to another world.
03:26:40 For me, because I’m a fan of role playing games,
03:26:44 the Elder Scrolls series,
03:26:47 like Skyrim or even Daggerfall,
03:26:50 it just takes you to another world.
03:26:52 And when you’re not in that world,
03:26:53 you miss not being there.
03:26:55 And then you just, you kind of want to stay there forever
03:26:58 because life is shitty.
03:27:00 And you just want to go to this place.
03:27:04 Is that there was a time when we were kind of asked
03:27:09 to come up with like, what’s your view about VR?
03:27:12 And my pitch was that it should be better
03:27:15 inside the headset than outside.
03:27:17 It’s the world as you want it.
03:27:19 And everybody thought that was dystopian.
03:27:21 And like, that’s like, oh,
03:27:23 you’re just going to forget about the world outside.
03:27:25 And I don’t get that mindset where the idea
03:27:29 that if you can make the world better inside the headset
03:27:32 than outside, you’ve just improved the person’s life
03:27:36 that has a headset that can wear it.
03:27:38 And there are plenty of things that we just can’t do
03:27:41 for everyone in the real world.
03:27:42 Everybody can’t have Richard Branson’s Private Island,
03:27:44 but everyone can have a private VR Island
03:27:47 and it can have the things that they want on it.
03:27:49 And there’s a lot of these kind of rivalrous goods
03:27:51 in the real world that VR can just be better at.
03:27:55 We can do a lot of things like that
03:27:57 that can be very, very rich.
03:27:59 So yeah, I want the,
03:28:01 I think it’s going to be a positive thing,
03:28:02 this world where people want to go back into their headset,
03:28:05 where it can be better than somebody that’s living
03:28:07 in a tiny apartment can have a palatial estate
03:28:10 in virtual reality.
03:28:11 They can have all their friends from all over the world
03:28:13 come over and visit them without everybody getting
03:28:16 on a plane and meeting in some place
03:28:19 and dealing with all the other logistics hassles.
03:28:21 There is real value in the presence that you can get
03:28:25 for remote meetings.
03:28:26 It’s all the little things that we need to sort out,
03:28:30 but those are things that we have line of sight on.
03:28:32 People that have been in like a good VR meeting
03:28:35 using work rooms where you can say,
03:28:38 oh, that was better than a Zoom meeting.
03:28:40 But of course it’s more of a hassle to get into it.
03:28:42 Not everyone has the headset.
03:28:44 Interoperability is worse.
03:28:46 You can’t have, you cap out at a certain number.
03:28:48 There’s all these things that need to be fixed,
03:28:50 but that’s one of those things you can look at and say,
03:28:52 we know there’s value there.
03:28:54 We just need to really grind hard,
03:28:56 file off all the rough edges and make that possible.
03:28:59 So you do think we have line of sight
03:29:02 because there’s a reason like,
03:29:06 I do this podcast in person, for example.
03:29:09 So doing it remotely, it’s not the same.
03:29:14 And if somebody were to ask me why it’s not the same,
03:29:16 I wouldn’t be able to write down exactly why.
03:29:20 But you’re saying that it’s possible,
03:29:23 whatever the magic is for in person interaction,
03:29:26 that immersiveness of the experience,
03:29:30 we are almost there.
03:29:32 Yes, so the idea of like,
03:29:34 I’m doing a VR interview with someone.
03:29:37 I’m not saying it’s here right now,
03:29:39 but you can see glimmers of what it should be.
03:29:42 And we largely know what would need to be fixed
03:29:46 and improved to, like you say,
03:29:48 there’s a difference between at remote interview
03:29:50 doing a podcast over Zoom or something and face to face.
03:29:53 There’s that sense of presence, that immediacy,
03:29:56 the super low latency responsiveness,
03:29:59 being able to see all the subtle things there,
03:30:01 just occupying the same field of view.
03:30:03 And all of those are things
03:30:04 that we absolutely can do in VR.
03:30:07 And that simple case of a small meeting
03:30:10 with a couple of people, that’s the much easier case
03:30:13 than everybody thinks the Ready Player One multiverse
03:30:15 with a thousand people going across a huge bridge
03:30:18 to amazing places,
03:30:20 that’s harder in a lot of other technical ways.
03:30:22 Not to say we can’t also do that,
03:30:24 but that’s further away and has more challenges.
03:30:26 But this small thing about being able to have a meeting
03:30:29 with one or a few people and have it feel real,
03:30:33 feel like you’re there,
03:30:35 like you have the same interactions and talking with them,
03:30:37 you get subtle cues as we start getting eye
03:30:40 and face tracking and some of the other things
03:30:42 on high end headsets.
03:30:43 A lot of that is going to come over
03:30:46 and it doesn’t have to be as good.
03:30:49 This is an important thing that people miss
03:30:51 where there was a lot of people that,
03:30:54 especially rich people that would look at VR and say,
03:30:57 it’s like, oh, this just isn’t that good.
03:30:59 And I’d say, it’s like, well,
03:31:00 you’ve already been courtside backstage and on pit row
03:31:05 and you’ve done all of these experiences
03:31:06 because you get to do them in real life.
03:31:08 But most people don’t get to.
03:31:10 And even if the experience is only half as good,
03:31:13 if it’s something that they never would have gotten
03:31:15 to do before, it’s still a very good thing.
03:31:17 And as we can just, we can push that number up over time.
03:31:20 It has a minimum viable value level
03:31:24 when it does something that is valuable enough to people,
03:31:27 as long as it’s better inside the headset on any metric
03:31:29 than it is outside and people choose to go there,
03:31:32 we’re on the right path.
03:31:33 And we have a value gradient
03:31:35 that I’m just always hammering on.
03:31:37 We can just follow this value gradient,
03:31:39 just keep making things better
03:31:41 rather than going for that one, close your eyes,
03:31:44 swing for the fences, kind of silver bullet approach.
03:31:48 Well, I wonder if there’s a value gradient
03:31:50 for in person meetings, because if you get that right,
03:31:53 I mean, that would change the world.
03:31:54 Yeah. That it doesn’t need to,
03:31:57 I mean, you don’t need a ready player one.
03:31:59 But I wonder if there’s that value gradient
03:32:02 you can follow along, because if there is and you follow it,
03:32:08 then there’ll be a certain like phase shift
03:32:11 at a certain point where people will shift
03:32:14 from Zoom to this.
03:32:18 I wonder, what are the bottlenecks?
03:32:23 Is it software, is it hardware?
03:32:25 Is it like, is it all about latency?
03:32:27 So I have big arguments internally
03:32:30 about strategic things like that,
03:32:32 where like the next headset that’s coming out
03:32:36 that we’ve made various announcements about
03:32:38 is gonna be a higher end headset,
03:32:40 more expensive, more features.
03:32:41 Lots of people wanna make those trade offs.
03:32:45 We’ll see what the market has to say
03:32:47 about the exact trade offs we’ve made here.
03:32:49 But if you wanna replace Zoom,
03:32:51 you need to have something that everybody has.
03:32:55 So you like cheaper.
03:32:56 I like cheaper because also lighter and cheaper
03:33:00 wind up being a virtuous cycle there
03:33:03 where expensive and more features
03:33:05 tends to also lead towards heavier.
03:33:07 And it just kind of goes, it’s like, let’s add more features.
03:33:09 The features are not, they have physical presence
03:33:13 and weight and draw from batteries and all of those things.
03:33:16 So I’ve always favored a lower end, cheaper, faster approach.
03:33:21 That’s why I was always behind the mobile side of VR
03:33:24 rather than the higher end PC headsets.
03:33:26 And I think that’s proven out well.
03:33:29 But there’s, ideally we have a whole range of things,
03:33:32 but if you’ve only got one or two things,
03:33:35 it’s important that those two things cover
03:33:37 the scope that you think is most important.
03:33:40 When we’re in a world when it’s like cell phones
03:33:42 and there’s 50 of them on the market
03:33:44 covering every conceivable ecological niche you want,
03:33:46 that’s gonna be great,
03:33:47 but we’re not gonna be there for a while.
03:33:49 Where are the bottlenecks?
03:33:51 Is it the hardware or the software?
03:33:53 Yeah, so right now you can play,
03:33:56 you can get work rooms on Quest
03:33:58 and you can set up these things
03:34:00 and it’s a pretty good experience.
03:34:01 It’s surprisingly good.
03:34:02 I haven’t tried it.
03:34:03 It’s surprisingly good.
03:34:05 Yeah, the voice latency is better on that
03:34:08 than a lot better than a Zoom meeting.
03:34:10 So you’ve got a more, a better sense of immediacy there.
03:34:13 The expressions that you get from the current hardware
03:34:17 with just kind of your controllers and your head
03:34:20 is pretty realistic feeling.
03:34:21 You’ve got a pretty good sense of being there
03:34:23 with someone with it.
03:34:24 Are these like avatars of people?
03:34:27 Yeah.
03:34:28 Do you get to see their body?
03:34:29 Yeah.
03:34:30 And they’re sitting around a table?
03:34:32 Yeah.
03:34:33 And it feels better than Zoom?
03:34:35 It feels better than you, yeah,
03:34:36 better than you’d expect for that.
03:34:37 It is definitely, yeah, I’d say it’s quite a bit better
03:34:42 than Zoom when everything’s working right,
03:34:43 but there’s still all the rough edges of,
03:34:46 the reason Zoom became so successful
03:34:48 is because they just nailed the usability of everything.
03:34:51 It’s high quality with a absolutely first rate experience.
03:34:54 And we are not there yet with any of the VR stuff.
03:34:58 I’m trying to push hard to get,
03:35:00 I keep talking about it’s like,
03:35:02 it needs to just be one click to make everything happen.
03:35:05 And we’re getting there in our home environment,
03:35:07 not the whole work rooms application,
03:35:09 but the main home where you can now kind of go over
03:35:11 and click and invite,
03:35:12 and it still winds up taking five times longer
03:35:15 than it should,
03:35:16 but we’re getting close to that where you click there,
03:35:19 they click on their button and then they’re sitting there
03:35:22 in this good presence with you,
03:35:23 but latencies need to get a lot better.
03:35:25 User interface needs to get a lot better.
03:35:28 Ubiquity of the headsets needs to get better.
03:35:30 We need to have a hundred million of them out there
03:35:33 just so that everybody knows somebody
03:35:35 that uses this all the time.
03:35:37 Well, I think it’s a virtuous cycle
03:35:38 because I do think the interface
03:35:44 is the thing that makes or breaks this kind of revolution.
03:35:48 It’s so interesting how like you said one click,
03:35:50 but it’s also like how you achieve that one click.
03:35:53 I don’t know.
03:35:54 What is, can I ask a dark question?
03:35:58 Maybe let’s keep it outside of meta,
03:36:00 but this is about meta, but also Google and big company.
03:36:05 Are they able to do this kind of thing?
03:36:07 It seems like, let me put on my cranky old man hat,
03:36:12 is they seem to not do a good job
03:36:15 of creating these user friendly interfaces
03:36:20 as they get bigger and bigger as a company.
03:36:22 Like Google has created some of the greatest interfaces ever
03:36:27 early on in its, I mean, creating Gmail,
03:36:30 just so many brilliant interfaces
03:36:33 and it just seems to be getting crappier
03:36:35 and crappier at that.
03:36:37 Same with meta, same with Microsoft.
03:36:42 It’s just, it seems to get worse and worse at that.
03:36:45 I don’t know what it is,
03:36:46 because you’ve become more conservative,
03:36:47 careful, risk averse.
03:36:49 Is that why?
03:36:51 Can you speak to that?
03:36:51 It’s been really eye opening to me
03:36:53 working inside a tech titan,
03:36:55 where I had my small companies
03:36:59 and then we’re acquired by a midsize game publisher
03:37:03 and then Oculus getting acquired by meta
03:37:06 and meta has grown by a factor of many
03:37:08 just in the eight years since the acquisition.
03:37:12 So I did not have experience with this
03:37:16 and it was interesting because I remember like previously
03:37:19 my benchmark for kind of use of resources
03:37:23 was some of the government programs I interacted with
03:37:25 on the aerospace side.
03:37:26 And I remember thinking there was,
03:37:28 okay, there’s an air force program
03:37:30 and they spent $50 million and they didn’t launch anything.
03:37:34 They didn’t even build anything.
03:37:35 It was just kind of like they made a bunch of papers
03:37:39 and had some parts in a warehouse and nothing came of it.
03:37:42 It’s like $50 million and I’ve had to radically recalibrate
03:37:47 my sense of like how much money can be spent
03:37:50 with mediocre resources.
03:37:54 Where on the plus side, VR has turned out,
03:37:58 we’ve built pretty much exactly what,
03:38:02 we just passed the 10 year mark then
03:38:03 from like my first demo of the Rift.
03:38:06 And if I could have said what I wanted to have,
03:38:09 it would have been a standalone inside out tracked
03:38:12 4K resolution headset that I,
03:38:15 that could still plug into a PC for high end rendering.
03:38:18 And that’s exactly what we’ve got on Quest 2 right now.
03:38:20 Yes, first of all, let’s pause on that
03:38:22 with me being cranky and everything.
03:38:24 It’s what Meta achieved with Oculus and so on is incredible.
03:38:30 I mean, this is, when I thought about the future of VR,
03:38:34 this is what I imagined in terms of hardware, I would say.
03:38:36 And maybe in terms of the experience as well,
03:38:39 but it’s still not there somehow.
03:38:42 On the one hand, we did kind of achieve it and win.
03:38:44 And we’ve got, we’ve sold, we’re a success right now,
03:38:48 but the amount of resources that have gone into it,
03:38:51 it winds up getting cluttered up in accounting
03:38:53 where Mark did announce that they spent $10 billion a year,
03:38:58 like on Reality Labs.
03:38:59 Now Reality Labs covers a lot.
03:39:01 It was, VR was not the large part of it.
03:39:04 It also had Portal and Spark
03:39:05 and the big AR research efforts.
03:39:08 And it’s been expanding out to include AI
03:39:11 and other things there where there’s a lot going on there.
03:39:15 But $10 billion was just a number
03:39:18 that I had trouble processing.
03:39:20 It’s just, I feel sick to my stomach
03:39:22 thinking about that much money being spent.
03:39:24 But that’s how they demonstrate commitment to this,
03:39:27 where it’s not more so than like, yeah,
03:39:31 Google goes and cancels all of these projects,
03:39:34 different things like that,
03:39:36 while Meta is really sticking with the funding of VR
03:39:39 and AR is still further out with it.
03:39:41 So there’s something to be said for that.
03:39:44 It’s not just gonna vanish, the work’s going in.
03:39:46 I just wish it could be,
03:39:48 all those resources could be applied more effectively
03:39:51 because I see all these cases.
03:39:53 I point out these examples of how a third party
03:39:56 that we’re kind of competing with in various ways.
03:39:58 There’s a number of these examples
03:40:00 and they do work with a 10th of the people
03:40:03 that we do internally and a lot of it comes from,
03:40:06 yes, the small company can just go do it
03:40:10 while in a big company, you do have to worry about,
03:40:13 is there some SDK internally that you should be using
03:40:16 because another team’s making it,
03:40:18 you have to have your cross functional group meetups
03:40:21 for different things.
03:40:22 You do have more concerns about privacy
03:40:25 or diversity and equity and safety of different things,
03:40:28 parental issues and things that a small startup company
03:40:31 can just kind of cowboy off and do something interesting.
03:40:36 And there’s a lot more that is a problem
03:40:39 that you have to pay attention to in the big companies,
03:40:41 but I’m not willing to believe that we are within
03:40:43 even a factor of two or four
03:40:46 of what the efficiency could be.
03:40:48 I am constantly kind of crying out for it’s like,
03:40:52 we can do better than this.
03:40:53 And you wonder what the mechanisms
03:40:55 to unlock that efficiency are.
03:40:57 I don’t, there is some sense in a large company
03:41:02 that like an individual engineer might not believe
03:41:05 that they can change the world.
03:41:07 Maybe you delegate a little bit of the responsibility
03:41:11 to be the one who changes the world in a big company,
03:41:14 I think, but the reality is like the world will get changed
03:41:19 by a single engineer anyway.
03:41:21 So if whether inside Google or inside a startup,
03:41:24 it doesn’t matter, it’s just like Google
03:41:26 and Meta needs to help those engineers believe
03:41:29 they’re the ones that are gonna decrease that latency.
03:41:32 It’ll take one John Carmack, the 20 year old Carmack
03:41:37 that’s inside Meta right now to change everything.
03:41:40 And I try to point that out and push people.
03:41:43 It’s like, try to go ahead.
03:41:45 And when you see some, because there is,
03:41:46 you get the silo mentality where you’re like,
03:41:48 okay, I know something’s not right over there,
03:41:50 but I’m staying in my lane here.
03:41:53 And there’s a couple of people that I can think about
03:41:57 that are willing to just like hop all over the place
03:41:59 and man, I treasure them.
03:42:01 The people that are just willing to, they’re fearless.
03:42:03 They will go over and they will go rebuild the kernel
03:42:06 and change this distribution and go in
03:42:08 and hack a firmware over here to get something done right.
03:42:11 And that is relatively rare.
03:42:14 There’s thousands of developers
03:42:15 and you’ve got a small handful that are willing
03:42:17 to operate at that level.
03:42:19 And it’s potentially risky for them.
03:42:22 The politics are real in a lot of that.
03:42:24 And I’m in the very much the privileged position
03:42:27 of I’m more or less untouchable there
03:42:31 where I’ve been dinged like twice
03:42:33 for it’s like you said something insensitive in that post
03:42:35 and you should probably not say that.
03:42:38 But for the most part, yes, I get away with,
03:42:42 every week I’m posting something pretty loud
03:42:45 and opinionated internally.
03:42:47 And I think that’s useful for the company,
03:42:50 but yeah, it’s rare to have a position like that.
03:42:54 And I can’t necessarily offer advice
03:42:56 for how someone can do that.
03:42:59 Well, you could offer advice to a company in general
03:43:01 to give a little bit of freedom for the young wild,
03:43:06 like the wildest ideas come from the young minds.
03:43:10 And so you need to give the young minds freedom
03:43:13 to think big and wild and crazy.
03:43:16 And for that, they have to be opinionated.
03:43:18 They have to think crazy ideas and thoughts
03:43:24 and pursue them with a full passion
03:43:26 without being slowed down by bureaucracy or managers
03:43:29 and all that kind of stuff.
03:43:31 Obviously startups really empower that,
03:43:33 but big companies could too.
03:43:34 And that’s a design challenge for big companies
03:43:37 to see how can you enable that?
03:43:40 How can you empower that?
03:43:41 Yeah, because the big company,
03:43:41 there are so many resources there.
03:43:43 And they do, amazing things do get accomplished,
03:43:46 but there’s so much more that could come out of that.
03:43:49 And I’m always hopeful.
03:43:51 I’m an optimist in almost everything.
03:43:53 I think things can get better.
03:43:54 I think that they can improve things
03:43:56 that you go through a path
03:43:58 and you’re learning kind of what does and doesn’t work.
03:44:01 And I’m not ready to be fatalistic
03:44:03 about the kind of the outcome of any of that.
03:44:06 Me neither, I know too many good people
03:44:09 inside of those large companies that are incredible.
03:44:12 You have a friendship with Elon Musk.
03:44:16 Often when I talk to him,
03:44:17 he’ll bring up how incredible of an engineer
03:44:19 and just a big picture thinker you are.
03:44:22 His huge amount of respect for you.
03:44:26 I just, I’ve never been a fly in the wall
03:44:28 between a discussion between the two of you.
03:44:30 I just wonder, is there something you guys debate,
03:44:34 argue about, discuss?
03:44:36 Is there some interesting problems
03:44:38 that the two of you think about?
03:44:41 You come from different worlds.
03:44:42 Maybe there’s some intersection in aerospace.
03:44:45 Maybe there’s some intersection
03:44:48 in your new efforts in artificial intelligence
03:44:52 in terms of thinking.
03:44:53 Is there something interesting you could say
03:44:55 about sort of the debates the two of you have?
03:44:57 So I think in some ways,
03:44:58 we do have a kind of similar background
03:45:01 where we’re almost exactly the same age.
03:45:03 And we had kind of similar programming backgrounds
03:45:06 on the personal computers
03:45:07 and even some of the books that we would read
03:45:10 and things that would kind of turn us into the people
03:45:13 that we are today.
03:45:14 And I think there is a degree of sensibility similarities
03:45:19 where we kind of call bullshit on the same things
03:45:22 and kind of see the same opportunities
03:45:25 in different technology.
03:45:27 And there’s that sense of,
03:45:29 I always talk about the speed of light solutions for things.
03:45:31 And he’s thinking about kind of minimum manufacturing
03:45:34 and engineering and operational standpoints for things.
03:45:39 And so, I mean, I first met Elon
03:45:42 right at the start of the aerospace era
03:45:44 where I wasn’t familiar with,
03:45:46 I was still in my game dev bubble.
03:45:47 I really wasn’t familiar with all the startups
03:45:49 that were going and being successful
03:45:52 and what went on with PayPal
03:45:53 and all of his different companies.
03:45:54 But I met him as I was starting to do armadillo aerospace.
03:45:58 And he came down with kind of his right hand propulsion guy.
03:46:03 And we talked about rockets.
03:46:05 What can we do with this?
03:46:07 And it was kind of specific things
03:46:09 about like how are our flight computers set up?
03:46:12 What are different propellant options?
03:46:14 You know, what can happen with different ways
03:46:17 of putting things together?
03:46:19 And then in some ways,
03:46:21 he was certainly the biggest player
03:46:22 in the sort of alt space community
03:46:24 that was going on in the early 2000s.
03:46:28 He was the most well funded,
03:46:30 although his funding in the larger scheme of things
03:46:32 compared to like a NASA or something like that
03:46:36 was really tiny.
03:46:38 It was a lot more than I had at the time.
03:46:40 But it was interesting.
03:46:42 I had a point years later when I realized,
03:46:45 okay, my financial resources at this point
03:46:48 are basically what Elon’s was
03:46:50 when he went all in on SpaceX and Tesla.
03:46:54 And there’s, I think in many corners,
03:46:58 he does not get the respect that he should
03:47:01 about being a wealthy person that could just retire.
03:47:04 And he went all in where he was really going to,
03:47:08 he could have gone bust.
03:47:11 And there’s plenty of people,
03:47:12 you’d look at the sad athletes or entertainers
03:47:16 that had all the money in the world and blew it.
03:47:18 And he could have been the business case example of that.
03:47:22 But the things that he was doing,
03:47:25 space exploration, electrification of transportation,
03:47:29 solar city type things,
03:47:31 these are big world level things.
03:47:34 And I have a great deal of admiration
03:47:36 that he was willing to throw himself so completely
03:47:40 into that because in contrast with myself,
03:47:43 I was doing armadillo aerospace with this tightly bounded,
03:47:47 it was John’s crazy money at the time
03:47:49 that had a finite limit on it.
03:47:51 It was never going to impact me or my family
03:47:54 if it completely failed.
03:47:56 And I was still hedging my bets,
03:47:58 working at id Software at the time
03:48:01 when he had been really all in there.
03:48:04 And I have a huge amount of respect for that.
03:48:08 And people do not,
03:48:09 the other thing I get irritated with is people that say,
03:48:11 it’s like, oh, Elon’s just a business guy.
03:48:13 He just got like, he was gifted the money
03:48:16 and he’s just kind of investing in all of this
03:48:19 when he was really deeply involved
03:48:22 in a lot of the decisions.
03:48:24 Not all of them were perfect,
03:48:25 but he cared very much about engine material selection,
03:48:30 propellant selection.
03:48:31 And for years he’d be kind of telling me,
03:48:34 it’s like, get off that hydrogen peroxide stuff.
03:48:37 It’s like, liquid oxygen is the only proper oxidizer
03:48:40 for this.
03:48:41 And the times that I’ve gone through the factories
03:48:46 with him, we’re talking very detailed things
03:48:49 about like how this weld is made,
03:48:51 how this sub assembly goes together,
03:48:54 what are like startup shut down behaviors
03:48:57 of the different things.
03:48:58 So he is really in there at a very detailed level.
03:49:03 And I think that he is the best modern example now
03:49:06 of someone that tries to,
03:49:08 that can effectively micromanage some decisions on things
03:49:11 on both Tesla and SpaceX to some degree
03:49:15 where he cares enough about it.
03:49:16 I worry a lot that he stretched too thin
03:49:19 that you get Boring Company and Neuralink and Twitter
03:49:23 and all the other possible things there
03:49:25 where I know I’ve got limits on how much I can pay attention
03:49:30 to that I have to kind of box off different amounts of time.
03:49:34 And I look back at like at my aerospace side of things,
03:49:36 it’s like, I did not go all in on that.
03:49:38 I did not commit myself at a level
03:49:40 that it would have taken to be successful there.
03:49:43 And I, yeah, and it’s kind of a weird thing
03:49:46 just like having a discussion with him.
03:49:49 He’s the richest man in the world right now,
03:49:50 but he operates on a level that is still very much
03:49:55 in my wheelhouse on a technical side of things.
03:50:00 So.
03:50:01 That doing that systems level type of thinking
03:50:02 where you can go to the low level details
03:50:04 and go up high to the big picture.
03:50:07 Do you think in aerospace arena
03:50:12 in the next five, 10 years,
03:50:14 do you think we’re gonna put a human on Mars?
03:50:16 Like, what do you think is the interesting?
03:50:20 No, I do, in fact, I made a bet with someone
03:50:23 with a group of people kind of this
03:50:24 about whether boots on Mars by 2030.
03:50:28 And this was kind of a fun story
03:50:31 because I was at an Intel sponsored event
03:50:34 and we had a bunch of just world class, brilliant people.
03:50:37 And we were talking about computing stuff,
03:50:39 but the after dinner conversation was like,
03:50:41 what are some other things?
03:50:42 How are they gonna go in the future?
03:50:43 And one of the ones tossed up on the whiteboard
03:50:46 was like boots on Mars by 2030.
03:50:49 And most of the people in the room thought, yes,
03:50:52 I thought that like, SpaceX is kicking ass.
03:50:54 We’ve got all this possible stuff.
03:50:57 Seems likely that it’s gonna go that way.
03:50:59 And I said, no, I think less than 50% chance
03:51:03 that it’s going to make it there.
03:51:05 And people were kind of like,
03:51:07 oh, why the pessimism or whatever?
03:51:09 And of course I’m an optimist at almost everything,
03:51:12 but for me to be the one kind of outlier saying,
03:51:14 no, I don’t think so.
03:51:17 Then I started saying some of the things I said,
03:51:19 well, let’s be concrete about it.
03:51:21 Let’s bet $10,000 that it’s not gonna happen.
03:51:24 And this was really a startling thing to see that I,
03:51:29 again, room full of brilliant people,
03:51:31 but as soon as like money came on the line
03:51:34 and they were like, do I wanna put 10,000?
03:51:36 I was not the richest person in the room.
03:51:38 There were people much better off than I was.
03:51:41 There’s a spectrum, but as soon as they started thinking,
03:51:45 it’s like, oh, I could lose money
03:51:46 by keeping my position right now.
03:51:50 And all these engineers, they engaged their brain.
03:51:53 They started thinking, it’s like, okay,
03:51:55 launch windows, launch delays,
03:51:57 like how many times would it take to get this right?
03:52:00 What historical precedents do we have?
03:52:02 And then it mostly came down to, it’s like,
03:52:05 well, what about in transit by 2030?
03:52:08 And then what about different things
03:52:11 or would you go for 2032?
03:52:13 But one of the people did go ahead
03:52:15 and was optimistic enough to make a bet with me.
03:52:18 So I have a $10,000 bet that by 2030,
03:52:21 I think it’s gonna happen shortly thereafter.
03:52:24 I think there will probably be infrastructure on Mars
03:52:26 by 2030, but I don’t think that we’ll have humans
03:52:29 on Mars on 2030.
03:52:31 I think it’s possible, but I think it’s less
03:52:33 than a 50% chance, so I felt safe making that bet.
03:52:36 Well, I think you had an interesting point.
03:52:39 Correct me if I’m wrong.
03:52:41 That’s a dark one.
03:52:42 That should perhaps help people appreciate Elon Musk,
03:52:48 which is, in this particular effort,
03:52:53 Elon is critical to the success.
03:52:56 SpaceX seems to be critical to, you know,
03:53:03 humans on Mars by 2030 or thereabouts.
03:53:07 So if something happens to Elon,
03:53:09 then all of this collapses.
03:53:12 And this is in contrast to the other $10,000 bet
03:53:16 I made kind of recently, and that was self driving cars
03:53:19 at like a level five running around cities.
03:53:22 And people have kind of nitpicked that
03:53:24 that we probably don’t mean exactly level five,
03:53:26 but the guy I’m having the bet with is,
03:53:29 we’re gonna be, we know what we mean about this.
03:53:31 Jeff Atwood.
03:53:32 Yeah, coding horror and stack overflow and all.
03:53:36 But yeah, I mean, he doesn’t think that people
03:53:40 are gonna be riding around in robo taxis in 2030
03:53:43 in major cities, just like you take an Uber now.
03:53:47 And I think it will.
03:53:48 You think it will.
03:53:49 And the difference is, everybody looks at this,
03:53:51 it’s like, oh, but Tesla’s been wrong for years.
03:53:53 They’ve been promising it for years and it’s not here yet.
03:53:56 And the reason this is different than the bet with Mars
03:54:00 is Mars really is more than is comfortable
03:54:04 a bet on Elon Musk.
03:54:06 That is his thing and he is really going to move
03:54:11 heaven and earth to try to make that happen.
03:54:13 And perhaps not even SpaceX, perhaps just Elon Musk.
03:54:17 Yeah, because if Elon went away and SpaceX went public
03:54:21 and got a board of directors,
03:54:23 there are more profitable things they could be doing
03:54:26 than focusing on human presence on Mars.
03:54:29 So this really is a sort of personal thing there.
03:54:32 And in contrast with that, self driving cars
03:54:36 have a dozen credible companies working really hard.
03:54:40 And while, yes, it’s going slower
03:54:43 than most people thought it would,
03:54:45 betting against that is a bet against almost the entire world
03:54:49 in terms of all of these companies
03:54:51 that have all of these incentives.
03:54:53 It’s not just one guy’s passion project.
03:54:56 And I do think that it is solvable.
03:54:59 Although I recognize it’s not 100% chance
03:55:02 because it’s possible the long tail of self driving problems
03:55:05 winds up being an AGI complete problem.
03:55:08 I think there’s plenty of value to mine out of it with narrow AI.
03:55:11 And I think that it’s going to happen probably more so
03:55:15 than people expect.
03:55:16 But it’s that whole sigmoid curve
03:55:18 where you overestimate the near term progress
03:55:21 and you underestimate the long term progress.
03:55:23 And I think self driving is going to be like that.
03:55:26 And I think 2030 is still a pretty good bet.
03:55:29 Yeah, unfortunately, self driving is a problem
03:55:35 that is safety critical, meaning that if you don’t do it well,
03:55:41 people get hurt.
03:55:43 But the other side of that is people are terrible drivers.
03:55:46 So it is not going to be,
03:55:48 that’s probably going to be the argument that gets it through
03:55:50 is like we can save 10,000 lives a year
03:55:54 by taking imperfect self driving cars
03:55:57 and letting them take over a lot of driving responsibilities.
03:56:00 It’s like 30,000 people a year die in auto accidents right now in America.
03:56:05 And a lot of those are preventable.
03:56:07 And the problem is you’ll have people that
03:56:09 every time a Tesla crashes into something,
03:56:11 you’ve got a bunch of people that literally have vested interests
03:56:14 shorting Tesla to come out and make it the worst thing in the world.
03:56:17 And people will be fighting against that.
03:56:19 But optimist in me again,
03:56:21 I think that we will have systems that are statistically safer than human drivers
03:56:26 and we will be saving thousands and thousands of lives every year
03:56:30 when we can hand over more of those responsibilities to it.
03:56:34 I do still think as a person who studied this problem very deeply
03:56:38 from a human side as well,
03:56:40 it’s still an open problem how good slash bad humans are driving.
03:56:46 It’s a kind of funny thing we say about each other.
03:56:49 Oh, humans suck at driving.
03:56:51 Everybody except you, of course.
03:56:54 Like we think we’re good at driving.
03:56:56 But I, after really studying it,
03:56:59 I think you start to notice, you know,
03:57:03 because I watched hundreds of hours of humans driving
03:57:06 with the projects of this kind of thing.
03:57:08 You’ve noticed that even with the distraction,
03:57:11 even with everything else,
03:57:13 humans are able to do some incredible things
03:57:17 with the attention,
03:57:19 even when you’re just looking at the smartphone,
03:57:21 just to get cues from the environment
03:57:23 to make last second decisions,
03:57:26 to use instinctual type of decisions that actually save your ass
03:57:30 time and time and time again.
03:57:32 And are able to do that with so much uncertainty around you
03:57:37 in such tricky dynamic environments.
03:57:40 I don’t know.
03:57:42 I don’t know exactly how hard is it to beat that kind of skill
03:57:49 of common sense reasoning.
03:57:51 This is one of those interesting things that there have been
03:57:53 a lot of studies about how experts in their field
03:57:56 usually underestimate the progress that’s going to happen
03:57:59 because an expert thinks about all the problems they deal with
03:58:02 and they’re like, damn, I’m going to have a hard time
03:58:05 solving all of this.
03:58:06 And they filter out the fact that they are one expert
03:58:08 in a field of thousands.
03:58:10 And you think about, yeah, I can’t do all of that.
03:58:13 And you sometimes forget about the scope of the ecosystem
03:58:16 that you’re embedded in.
03:58:18 And if you think back eight years, very specifically,
03:58:20 the state of AI and machine learning, where was that?
03:58:23 We had just gotten Resnets probably at that point.
03:58:26 And you look at all the amazing magical things
03:58:29 that have happened in eight years,
03:58:31 and they do kind of seem to be happening a little faster
03:58:34 in recent years also.
03:58:36 And you project that eight more years into the future,
03:58:39 where, again, I think there’s a 50% chance
03:58:41 we’re going to have signs of life of AGI,
03:58:43 which we can put through driver’s ed if we need to
03:58:46 to actually build self driving cars.
03:58:48 And I think that the narrow systems are going to have
03:58:51 real value demonstrated well before then.
03:58:54 So signs of life in AGI, you’ve mentioned that,
03:59:00 okay, first of all, you’re one of the most brilliant people
03:59:04 on this earth.
03:59:05 You could be solving a number of different problems,
03:59:08 as you’ve mentioned.
03:59:09 Your mind was attracted to nuclear energy.
03:59:12 Obviously, virtual reality with the metaverse
03:59:14 is something you could have a tremendous impact on.
03:59:16 So I do want to say a quick thing about nuclear energy,
03:59:19 where this is something that so precisely feels like
03:59:25 aerospace before SpaceX, where from everything that I know
03:59:29 about all of these, the physics of this stuff hasn’t changed.
03:59:33 And the reasons why things are expensive now
03:59:36 are not fundamental.
03:59:38 Somebody should be going into a really hard Elon Musk style
03:59:44 at fission, economical fission, not fusion,
03:59:48 where the fusion is the kind of the darling of people
03:59:52 that want to go and do nuclear because it doesn’t have
03:59:55 the taint that fission has in a lot of people’s minds.
03:59:58 But it’s an almost absurdly complex thing where nuclear fusion,
04:00:03 as you look at the tokamaks or any of the things that people
04:00:06 are building, and it’s doing all of this infrastructure
04:00:09 just at the end of the day to make something hot
04:00:12 that you can then turn into energy
04:00:14 through a conventional power plant.
04:00:16 And all of that work, which we think
04:00:18 we’ve got line of sight on, but even if it comes out,
04:00:21 then you have to do all of that immensely complex expensive
04:00:25 stuff just to make something hot, where nuclear fission
04:00:28 is basically you put these two rocks together
04:00:30 and they get hot all by themselves.
04:00:32 That is just that much simpler.
04:00:35 It’s just orders of magnitude simpler.
04:00:37 And the actual rocks, the refined uranium,
04:00:40 is not very expensive.
04:00:41 It’s a couple percent of the cost of electricity.
04:00:44 That’s why I made that point where you could have something
04:00:47 which was five times less efficient than current systems.
04:00:51 And if the rest of the plant was a whole bunch cheaper,
04:00:54 you could still be super, super valuable.
04:00:56 So how much of the pie do you think could be solved
04:01:02 by nuclear energy by fission?
04:01:04 So how much could it become the primary source
04:01:07 of energy on Earth?
04:01:08 It could be most of it.
04:01:10 Like the reserves of uranium, as it stands now,
04:01:12 could not power the whole Earth.
04:01:13 But you get into breeder reactors and thorium
04:01:16 and things like that that you do for conventional fission.
04:01:19 There is enough for everything.
04:01:22 Now, I mean, solar, photovoltaic has been amazing.
04:01:25 One of my current projects is working on an off grid system.
04:01:29 And it’s been fun just kind of, again, putting my hands
04:01:31 on all this, stripping the wires and wiring things together
04:01:34 and doing all of that.
04:01:35 And just having followed that a little bit
04:01:37 from the outside over the last couple decades,
04:01:40 there’s been semiconductor like magical progress
04:01:43 in what’s going on there.
04:01:45 So I’m all for all of that.
04:01:47 But it doesn’t solve everything.
04:01:49 And nuclear really still does seem like the smart money
04:01:52 bet for what you should be getting for baseband
04:01:55 on a lot of things.
04:01:56 And solar may be cheaper for peaking over air conditioning
04:02:00 loads during the summer and things
04:02:02 that you can push around in different ways.
04:02:05 But it’s one of those things that’s
04:02:07 it’s just strange how we’ve had the technology sitting there.
04:02:10 But these non technical reasons on the social optics of it
04:02:14 has been this major forcing function for something
04:02:18 that really should be at the cornerstone of all
04:02:21 of the world’s concerns with energy.
04:02:23 It’s interesting how the non technical factors have really
04:02:27 dominated something that is so fundamental to the existence
04:02:31 of the human race as we know it today.
04:02:34 And much of the troubles of the world,
04:02:36 including wars in different parts of the world like Ukraine
04:02:40 is energy based.
04:02:41 And yeah, it’s just sitting right there to be solved.
04:02:47 That said, I mean, to me personally,
04:02:50 I think it’s clear that if AGI were to be achieved,
04:02:53 that would change the course of human history.
04:02:55 So AGI wise, I was making this decision about what
04:03:00 do I want to focus on after VR.
04:03:03 And I’m still working on VR regularly.
04:03:06 I spend a day a week kind of consulting with Meta.
04:03:09 And Boz styles me the consulting CTO.
04:03:13 It’s kind of like the Sherlock Holmes that
04:03:15 comes in and consults on some of the specific tough issues.
04:03:18 And I’m still pretty passionate about all of that.
04:03:21 But I have been figuring out how to compartmentalize and force
04:03:25 that into a smaller box to work on some other things.
04:03:29 And I did come down to this decision
04:03:30 between working on economical nuclear fission
04:03:34 or artificial general intelligence.
04:03:36 And the fission side of things, I’ve
04:03:39 got a bunch of interesting things going that way.
04:03:42 But that would be a fairly big project thing to do.
04:03:46 I don’t think it needs to be as big as people expect.
04:03:49 I do think something original SpaceX sized, you build it,
04:03:53 power your building off of it, and then the government,
04:03:56 I think, will come around to what you need to.
04:03:59 Everybody loves an existence proof.
04:04:01 I think it’s possible somebody should be doing this.
04:04:03 But it’s going to involve some politics.
04:04:05 It’s going to involve decent sized teams
04:04:07 and a bunch of this cross functional stuff
04:04:09 that I don’t love.
04:04:11 While the artificial general intelligence side of things,
04:04:15 it seems to me like this is the highest leverage
04:04:21 moment for potentially a single individual,
04:04:24 potentially in the history of the world,
04:04:26 where the things that we know about the brain, about what we
04:04:30 can do with artificial intelligence,
04:04:33 nobody can say absolutely on any of these things.
04:04:36 But I am not a madman for saying that it is likely
04:04:40 that the code for artificial general intelligence
04:04:44 is going to be tens of thousands of lines of code,
04:04:48 not millions of lines of code.
04:04:50 This is code that conceivably one individual could write,
04:04:53 unlike writing a new web browser or operating system.
04:04:57 And based on the progress that AI, machine learning,
04:05:01 has made in the recent decade, it’s
04:05:04 likely that the important things that we don’t know
04:05:07 are relatively simple.
04:05:09 There’s probably a handful of things.
04:05:11 My bet is that I think there’s less than six key insights that
04:05:16 need to be made.
04:05:17 Each one of them can probably be written
04:05:19 on the back of an envelope.
04:05:20 We don’t know what they are.
04:05:22 But when they’re put together in concert with GPUs at scale
04:05:26 and the data that we all have access to,
04:05:28 that we can make something that behaves like a human being
04:05:32 or like a living creature and that can then
04:05:35 be educated in whatever ways that we need to get to the point
04:05:39 where we can have universal remote workers where anything
04:05:43 that somebody does mediated by a computer
04:05:45 and doesn’t require physical interaction that an AGI will
04:05:49 be able to do.
04:05:50 We can already simulate the equivalent of the Zoom meetings
04:05:54 with avatars and synthetic deep fakes and whatnot.
04:05:58 We can definitely do that.
04:06:00 We have superhuman capabilities on any narrow thing
04:06:03 that we can formalize and make a loss function for.
04:06:07 But there’s things we don’t know how to do now.
04:06:10 But I don’t think they are unapproachably hard.
04:06:13 Now, that’s incredibly hubristic to say that it’s like,
04:06:16 but I think that what I said a couple years ago
04:06:19 is a 50% chance that somewhere there will be signs of life
04:06:23 of AGI in 2030.
04:06:25 And I’ve probably increased that slightly.
04:06:27 I may be at 55%, 60% now because I
04:06:30 do think there’s a little sense of acceleration there.
04:06:34 So I wonder what the, and by the way,
04:06:36 you also written that I bet with hindsight
04:06:39 we will find that clear antecedents of all
04:06:42 the critical remaining steps for AGI
04:06:44 are already buried somewhere in the vast literature of today.
04:06:48 So the ideas are already there.
04:06:50 I think that’s likely the case.
04:06:52 One of the things that appeals to so many people,
04:06:54 including me, about the promise of AGI
04:06:56 is we know that we’re only drinking
04:06:59 from a straw from the fire hose of all the information out there.
04:07:03 I mean, you look at just in a very narrowly bounded field
04:07:07 like machine learning, like you can’t read all the papers that
04:07:10 come out all the time.
04:07:11 You can’t go back and read all the clever things
04:07:14 that people did in the 90s or earlier
04:07:16 that people have forgotten about because they didn’t pan out
04:07:18 at the time when they were trying to do them with 12 neurons.
04:07:22 So this idea that, yeah, I think there
04:07:25 are gems buried in some of the older literature that was not
04:07:29 the path taken by everything.
04:07:31 And you can see a herd mentality on the things that
04:07:34 happen right now.
04:07:35 It’s almost funny to see.
04:07:36 It’s like, oh, Google does something,
04:07:38 and OpenAI does something, Meta does something.
04:07:41 They’re the same people that all talk to each other,
04:07:43 and they’re all one upping each other,
04:07:45 and they’re all capable of implementing
04:07:47 each other’s work given a month or two
04:07:49 after somebody has an announcement of that.
04:07:52 But there’s a whole world of possible approaches
04:07:55 to machine learning.
04:07:57 And I think that we probably will, in hindsight,
04:08:00 go back and see it’s like, yeah, that was kind of clearly
04:08:03 predicted by this early paper here.
04:08:06 And this turns out that if you do this and this
04:08:08 and take this result from animal training
04:08:11 and this thing from neuroscience over here
04:08:13 and put it together and set up this curriculum for them
04:08:16 to learn in, that that’s kind of what it took.
04:08:19 You don’t have too many people now
04:08:21 that are still saying it’s not possible
04:08:23 or it’s going to take hundreds of years.
04:08:25 And 10 years ago, you would get a collection of experts,
04:08:29 and you would have a decent chunk on the margin that either
04:08:32 say not possible or a couple hundred years,
04:08:35 might be centuries.
04:08:36 And the median estimate would be like 50, 70 years.
04:08:40 And it’s been coming down.
04:08:41 And I know with me saying eight years for something,
04:08:44 that still puts me on the optimistic side,
04:08:46 but it’s not crazy out in the fringes.
04:08:49 And just being able to look at that at a meta level
04:08:52 about the trend of the predictions going down there,
04:08:57 the idea that something could be happening relatively soon.
04:09:01 Now, I do not believe in fast takeoffs.
04:09:04 That’s one of the safety issues that people say,
04:09:06 it’s like, oh, it’s going to go, boom,
04:09:08 and the AI is going to take over the world.
04:09:10 There’s a lot of reasons I don’t think
04:09:12 that’s a credible position.
04:09:14 And I think that we will go from a point
04:09:16 where we start seeing things that credibly look
04:09:20 like animals behaviors.
04:09:22 And I have a human voice box wired into them.
04:09:25 It’s like I tried to get Elon to say,
04:09:27 it’s like your pig in Neuralink, give it a human voice box
04:09:30 and let it start learning human words.
04:09:33 I think animal intelligence is closer to human intelligence
04:09:36 than a lot of people like to think.
04:09:38 And I think that culture and modalities of IO
04:09:42 make the Gulf seem a lot bigger than it actually is.
04:09:45 There’s just that smooth spectrum
04:09:47 of how the brain developed and cortexes and scaling
04:09:51 of different things going on there.
04:09:53 Culture modalities of IO, yes.
04:09:55 Language is sort of lost in translation,
04:10:00 conceals a lot of intelligence.
04:10:03 So when you think about signs of life for AGI,
04:10:06 you’re thinking about human interpretable signs.
04:10:10 So the example I give, if we get to the point
04:10:12 where you’ve got a learning disabled toddler,
04:10:15 some kind of real special needs child
04:10:18 that can still interact with their favorite TV show
04:10:20 and video game and can be trained and learn
04:10:24 in some appreciably human like way.
04:10:26 At that point, you can deploy an army of engineers,
04:10:30 cognitive scientists, developmental education people.
04:10:35 And you’ve got so many advantages there
04:10:37 unlike real education where you can do rollbacks
04:10:39 and AB testing and you can find a golden path
04:10:42 through a curriculum of different things.
04:10:44 If you get to that point, learning disabled toddler,
04:10:47 I think that it’s gonna be a done deal.
04:10:50 But do you think we’ll know when we see it?
04:10:53 So there’s been a lot of really interesting
04:10:56 general learning progress from DeepMind,
04:11:00 OpenAI a little bit too.
04:11:02 I tend to believe that Tesla Autopilot
04:11:06 deserves a lot more credit than it’s getting
04:11:08 for making progress on the general,
04:11:12 on doing the multitask learning thing
04:11:15 and increasing the number of tasks
04:11:17 and automating that process of sort of learning
04:11:23 from the edge, discovering the edge cases
04:11:25 and learning from the edge cases.
04:11:26 That is, it’s really approaching from a different angle,
04:11:30 the general learning problem of AGI.
04:11:33 But the more clear approach comes from DeepMind
04:11:36 where you have these kind of game situations
04:11:38 and you build systems there.
04:11:41 But I don’t know, people seem to be quite…
04:11:47 Yeah, there will always be people that just won’t believe it
04:11:50 and I fundamentally don’t care.
04:11:52 I mean, I don’t care if they don’t believe it.
04:11:54 You know, when it starts doing people’s jobs
04:11:57 and I don’t care about the philosophical zombie argument at all.
04:12:01 Absolutely.
04:12:02 But will you, do you think you will notice
04:12:04 that something special has happened here?
04:12:06 Because to me, I’ve been noticing a lot of special things.
04:12:12 I think a lot of credit should go to DeepMind for AlphaZero.
04:12:18 That was truly special.
04:12:20 The self play mechanisms achieve sort of solve problems
04:12:24 that used to be thought unsolvable like the game of Go.
04:12:28 Also, I mean, protein folding, starting to get into that space
04:12:32 where learning is doing…
04:12:34 At first there’s not, it wasn’t end to end learning
04:12:37 and now it’s end to end learning of a very difficult
04:12:41 previously thought unsolvable problem of protein folding.
04:12:45 And so, yeah, where do you think
04:12:50 would be a really magical moment for you?
04:12:54 There have been incredible things happening in recent years.
04:12:57 Like you say, all of the things from DeepMind and OpenAI
04:13:00 that have been huge showpiece things.
04:13:03 But when you really get down to it and you read the papers
04:13:05 and you look at the way the models are going,
04:13:08 it’s still like a feed forward.
04:13:10 You push something in, something comes out on the end.
04:13:13 I mean, maybe there’s diffusion models
04:13:15 or Monte Carlo tree rollouts and different things going on,
04:13:18 but it’s not a being.
04:13:20 It’s not close to a being.
04:13:22 That’s going through a lifelong learning process.
04:13:27 Do you want something that kind of gives signs of a being?
04:13:31 What’s the difference between a neural network,
04:13:36 a feed forward neural network and a being?
04:13:39 Where’s the loop?
04:13:40 Fundamentally, the brain is a recurrent neural network
04:13:43 generating an action policy.
04:13:45 I mean, it’s implemented on a biological substrate.
04:13:47 And it’s interesting thinking about things like that
04:13:49 where we know fundamentally the brain is not
04:13:52 a convolutional neural network or a transformer.
04:13:55 Those are specialized things that are very valuable
04:13:58 for what we’re doing, but it’s not the way the brain’s doing.
04:14:00 Now, I do think consciousness and AI in general
04:14:03 is a substrate independent mechanism
04:14:06 where it doesn’t have to be implemented the way the brain is,
04:14:09 but if you’ve only got one existence proof,
04:14:11 there’s certainly some value in caring about what it says and does.
04:14:17 And so the idea that anything that can be done with a narrow AI
04:14:21 that you can quantify up a loss function for or reward mechanism,
04:14:25 you’re almost certainly going to be able to produce something
04:14:28 that’s more resource effective to train and deploy
04:14:31 and use in an inference mode.
04:14:33 You know, train a whole lot using an inference.
04:14:35 But a living being is going to be something that’s a continuous,
04:14:39 lifelong learned task agnostic thing.
04:14:42 So the lifelong learning is really important too.
04:14:46 And the long term memory.
04:14:48 So memory is a big weird part of that puzzle.
04:14:51 And we’ve got, you know, again, I have all the respect in the world
04:14:55 for the amazing things that are being done now,
04:14:57 but sometimes they can be taken a little bit out of context
04:15:00 with things like there’s some smoke and mirrors going on,
04:15:04 like the Gato, the recent work, the multitask learning stuff.
04:15:07 You know, it’s amazing that it’s one model that plays all the Atari games
04:15:12 as well as doing all of these other things.
04:15:14 But of course, it didn’t learn to do all of those.
04:15:17 It was instructed in doing that by other reinforcement learners
04:15:21 going through and doing that.
04:15:23 And even in the case of all the games,
04:15:25 it’s still going with a specific hand coded reward function
04:15:29 in each of those Atari games where it’s not that, you know,
04:15:33 it just wants to spend its summer afternoon playing Atari
04:15:35 because that’s the most interesting thing for it.
04:15:38 So it’s, again, not a general, it’s not learning the way humans learn.
04:15:42 And there’s, I believe, a lot of things that are challenging
04:15:45 to make a loss function for that you can train
04:15:48 through these existing conventional things.
04:15:51 We’re going to chip away at all the things that people do
04:15:54 that we can turn into narrow AI problems
04:15:59 and billions of, like trillions of dollars of value
04:16:02 are going to be created by that.
04:16:04 But there’s still going to be a set of things
04:16:06 and we’ve got questionable cases like the self driving car
04:16:09 where it’s possible, it’s not my bet, but it’s plausible
04:16:13 that the long tail could be problematic enough
04:16:15 that that really does require a full on artificial general intelligence.
04:16:20 The counter argument is that data solves almost everything.
04:16:23 There’s an interpolation problem if you have enough data
04:16:26 and Tesla may be able to get enough data
04:16:29 from all of their deployed stuff to be able to work like that,
04:16:31 but maybe not.
04:16:33 And there are all the other problems about, like,
04:16:35 say you want to have a strategy meeting
04:16:37 and you want to go ahead and bring in all of your remote workers
04:16:40 and your consultants, and you want a world where
04:16:43 some of those could be AIs that are talking and interacting with you
04:16:48 in an area that is too murky to have a crisp loss function,
04:16:52 but they still have things that on some level,
04:16:54 they’re rewarded on some internal level
04:16:57 for building a valuable to humans kind of life
04:17:01 and ability to interact with things.
04:17:03 See, I still think that self driving cars,
04:17:07 solving that problem will take us very far towards AGI.
04:17:10 You might not need AGI, but I am really inspired
04:17:13 by what Autopilot is doing.
04:17:16 Waymo, so some of the other companies,
04:17:19 I think Waymo leads the way there is also really interesting,
04:17:23 but they don’t have quite as ambitious of an effort
04:17:26 in terms of learning based sort of data hungry approach to driving,
04:17:32 which I think is very close to the kind of thing
04:17:34 that would take us far towards AGI.
04:17:37 Yeah, and it’s a funny thing because as far as I can tell,
04:17:40 Elon is completely serious about all of his concerns about AGI
04:17:44 being an existential threat.
04:17:46 And I tried to draw him out to talk about AGI
04:17:49 and he just didn’t want to.
04:17:50 And I think that I get that little fatalistic sense from him.
04:17:54 It’s weird because his company could very well be the leading company
04:17:58 leading towards a lot of that where Tesla being a super pragmatic company
04:18:03 that’s doing things because they really want to solve this actual problem.
04:18:06 It’s a different vibe than the research oriented companies
04:18:10 where it’s a great time to be an AI researcher.
04:18:12 You’ve got your pick of trillion dollar companies
04:18:14 that will pay you to kind of work on the problems you’re interested in,
04:18:18 but that’s not necessarily driving hard towards the core problem
04:18:22 of AGI as something that’s going to produce a lot of value
04:18:25 by doing things that people currently do or would like to do.
04:18:30 I mean, I have a million questions to you about your ideas about AGI,
04:18:35 but do you think it needs to be embodied?
04:18:39 Do you think it needs to have a body to start to notice the signs of life
04:18:44 and to develop the kind of system that’s able to reason,
04:18:49 perceive the world in the way that an AGI should and act in the world?
04:18:53 So should we be thinking about robots
04:18:55 or can this be achieved in a purely digital system?
04:18:58 So I have a clear opinion on that and that’s that no,
04:19:01 it does not need to be embodied in the physical world
04:19:04 where you could say most of my career is about making simulated virtual worlds
04:19:09 in games or virtual reality.
04:19:11 And so on a fundamental level,
04:19:13 I believe that you can make a simulated environment
04:19:16 that provides much of the value of what the real environment does
04:19:19 and restricting yourself to operating at real time
04:19:23 in the physical world with physical objects,
04:19:25 I think is an enormous handicap.
04:19:27 I mean, that’s one of the real lessons driven home by all my aerospace work
04:19:31 is that, you know, reality is a bitch in so many ways there
04:19:36 where dealing with all the mechanical components,
04:19:38 like everything fails, Murphy’s law,
04:19:40 even if you’ve done it right before on your fifth one,
04:19:42 it might come out differently.
04:19:44 So yeah, I think that anybody that is all in on the embodied aspect of it,
04:19:50 they are tying a huge weight to their ankles
04:19:53 and I think that I would almost count them out,
04:19:57 anybody that’s making that a cornerstone of their belief about it,
04:20:00 I would almost write them off as being worried about them getting to AGI first.
04:20:04 I was very surprised that Elon’s big on the humanoid robots.
04:20:09 I mean, like the NASA Robonaut stuff was almost a gag line,
04:20:12 like what are you doing people?
04:20:14 Well, that’s very interesting because he has a very pragmatic view of that.
04:20:18 That’s just a way to solve a particular problem in a factory.
04:20:23 Now, I do think that once you have an AGI,
04:20:26 robotic bodies, humanoid bodies are going to be enormously valuable.
04:20:29 I just don’t think they’re helpful getting to AGI.
04:20:32 Well, he has a very sort of practical view, which I disagree with and argue with him,
04:20:37 but it’s a practical view that there’s, you know,
04:20:40 you could transfer the problem of driving to the problem of robotic manipulation
04:20:46 because so much of it is perception.
04:20:49 It’s perception and action and it’s just a different context
04:20:53 and so you can apply all the same kind of data engine learning processes
04:20:58 to a different environment.
04:20:59 And so why not apply it to the humanoid robot environment?
04:21:03 But I think, I do think that there’s a certain magic to the embodied robot.
04:21:12 That may be the thing that finally convinces people.
04:21:15 But again, I don’t really care that much about convincing people.
04:21:18 You know, the world that I’m looking towards is, you know,
04:21:21 you go to the website and say,
04:21:24 I want five Frank 1As to work on my team today and they all spin up
04:21:29 and they start showing up in your Zoom meetings.
04:21:31 To push back, but also to agree with you.
04:21:33 But first to push back, I do think you need to convince people
04:21:36 for them to welcome that thing into their life.
04:21:40 I think there’s enough businesses that operate on an objective kind of
04:21:44 profit loss sort of basis that, I mean, if you look at how many things,
04:21:48 again, talking about the world as an evolutionary space there,
04:21:52 when you do have free markets and you have entrepreneurs,
04:21:56 you are going to have people that are going to be willing to go out
04:21:58 and try whatever crazy things.
04:22:00 And when it proves to be beneficial, you know,
04:22:03 there’s fast followers in all sorts of places.
04:22:05 Yeah, and you’re saying that, I mean, you know, Quake and VR
04:22:10 is a kind of embodiment, but just in a digital world.
04:22:13 And if you’re able to demonstrate, if you’re able to do something
04:22:17 productive in that kind of digital reality,
04:22:22 then AGI doesn’t need to have a body.
04:22:25 Yeah, it’s like one of the really practical technical questions
04:22:28 that I kind of keep arguing with myself over.
04:22:31 If you’re doing a training and learning and you’ve got like,
04:22:34 you can watch Sesame Street, you can play Master System games or something,
04:22:38 is it enough to have just a video feed that is that video coming in?
04:22:42 Or should it literally be on a virtual TV set in a virtual room,
04:22:47 even if it’s, you know, a simple room, just to have that sense of
04:22:50 you’re looking at a 2D projection on a screen versus having the screen
04:22:54 beam directly into your retinas.
04:22:56 And, you know, I think it’s possible to maybe get past some of these
04:23:00 signs of life of things with the just kind of projected directly
04:23:04 into the receptor fields, but eventually for more kind of human
04:23:09 emotional connection for things, probably having some VR room
04:23:14 with a lot of screens in it for the AI to be learning in is likely helpful.
04:23:18 It may be a world of different AIs interacting with each other.
04:23:22 That self play I do think is one of the critical things where
04:23:24 socialization wise, one of the other limitations I set for myself
04:23:28 thinking about these is I need something that is at least potentially
04:23:33 real time because I want, it’s nice you can always slow down time,
04:23:37 you can run on a subscale system and test an algorithm at some lower level.
04:23:42 And if you’ve got extra horsepower, running it faster than real time
04:23:45 is a great thing.
04:23:46 But I want to be able to have the AIs either socially interact with
04:23:52 each other or critically with actual people, your sort of child
04:23:56 development psychiatrist that comes in and interacts and does the
04:24:00 good boy, bad boy sort of thing as they’re going through and
04:24:04 exploring different things.
04:24:05 And it’s nice to, I come back to the value of constraints in a lot of ways.
04:24:10 And if I say, well, one of my constraints is real time operation.
04:24:13 I mean, it might still be a huge data center full of computers,
04:24:17 but it should be able to interact on a Zoom meeting with people.
04:24:21 And that’s how you also do start convincing people, even if it’s not
04:24:24 a robot body moving around, which eventually gets to irrefutable levels.
04:24:28 But if you can go ahead and not just type back and forth to a GPT bot
04:24:33 on something, but you’re literally talking to them in an embodied
04:24:37 over Zoom form and working through problems with them or exploring
04:24:42 situations, having conversations that are fully stateful and learned.
04:24:46 I think that that’s a valuable thing.
04:24:49 So I do keep all of my eyes on things that can be implemented within
04:24:54 sort of that 30 frames per second kind of work.
04:24:58 And I think that’s feasible.
04:24:59 Do you think the most compelling experiences that first will be
04:25:03 for pleasure or for business as they ask in airports?
04:25:07 So meaning is if it’s interacting with AI agents, will it be sort of
04:25:17 like friends, entertainment, almost like a therapist or whatever,
04:25:25 that kind of interaction?
04:25:26 Or is it in the business setting, something like you said, brainstorming
04:25:30 different ideas, sort of this is all a different formulation of kind of
04:25:34 a Turing test or the spirit of the original Turing test.
04:25:37 Where do you think the biggest benefit will first come?
04:25:40 So it’s going to start off hugely expensive.
04:25:42 I mean, you’re going to if we’re still all guessing about what compute
04:25:47 is going to be necessary.
04:25:48 I fall on the side of I don’t think you run the numbers and you’re like
04:25:51 86 billion neurons, 100 trillion synapses.
04:25:53 I don’t think those all need to be weights.
04:25:55 I don’t think we need models that are quite that big evaluated quite
04:25:58 that often.
04:25:59 I based that on we’ve got reasonable estimates of what some parts of
04:26:03 the brain do.
04:26:04 We don’t have the neocortex formula, but we kind of get some of the
04:26:08 other sensory processing and it doesn’t feel like we need to.
04:26:11 We can simulate that in computers for less weights.
04:26:14 But still, it’s probably going to be thousands of GPUs to be running
04:26:20 a human level AGI, depending on how it’s implemented.
04:26:23 That might give you sort of a clan of 128 kind of run in batch people,
04:26:28 depending on whether there’s sparsity in the way that the weights are
04:26:31 and things are set up.
04:26:32 If it is a reasonably dense thing, then just the memory bandwidth
04:26:36 tradeoffs means you get 128 of them at the same time.
04:26:39 And either it’s all feeding together, learning in parallel or kind of
04:26:43 all running together, kind of talking to a bunch of people.
04:26:46 But still, if you’ve got thousands of GPUs necessary to run these
04:26:50 things, it’s going to be kind of expensive where it might start off
04:26:55 $1,000 an hour for even post development or something for that, which
04:27:00 would be something that you would only use for a business, something
04:27:04 where you think they’re going to help you make a strategic decision
04:27:07 or point out something super important.
04:27:09 But I also am completely confident that we will have another factor
04:27:14 of 1,000 in cost performance increase in AGI type calculations.
04:27:19 Not in general computing necessarily, but there’s so much more that
04:27:23 we can do with packaging, making those right tradeoffs, all those
04:27:26 same types of things that in the next couple of decades, 1,000 X easy.
04:27:30 And then you’re down to $1 an hour.
04:27:32 And then you’re kind of like, well, I should have an entourage of
04:27:36 AIs that are following me around, helping me out on anything that
04:27:40 I want them to do.
04:27:41 That’s one interesting trajectory.
04:27:43 But I’ll push back because I have, so in that case, if you want to
04:27:50 pay thousands of dollars, it should actually provide some value.
04:27:54 I think it’s easier for cheaper to provide value via a dumb AI that
04:28:04 will take us towards AGI to just have a friend.
04:28:08 I think there’s an ocean of loneliness in the world.
04:28:11 And I think an effective friend that doesn’t have to be perfect,
04:28:16 that doesn’t have to be intelligent, that has to be empathic,
04:28:20 having emotional intelligence, having ability to remember things,
04:28:24 having ability to listen.
04:28:26 Most of us don’t listen to each other.
04:28:28 One of the things that love and when you care about somebody,
04:28:31 when you love somebody is when you listen.
04:28:34 And that is something we treasure about each other.
04:28:37 And if an AI can do that kind of thing.
04:28:40 I think that provides a huge amount of value and very importantly,
04:28:45 provides value in its ability to listen and understand
04:28:51 versus provide really good advice.
04:28:53 I think providing really good advice is another next level step
04:28:59 that I think is just easier to do companionship.
04:29:05 Yeah, I wouldn’t disagree.
04:29:06 I mean, I think that there’s very few things that I would argue
04:29:10 can’t be reduced to some kind of a narrow AI.
04:29:14 I think we can do a trillion dollars of value easily
04:29:17 and all the things that can be done there.
04:29:19 And a lot of it can be done with smoke and mirrors
04:29:21 without having to go the whole thing.
04:29:23 I mean, there’s going to be the equivalent of the doom version
04:29:27 for the AGI that’s not really AGI, it’s all smoke and mirrors,
04:29:31 but it happens to do enough valuable things that it’s enormously
04:29:34 useful and valuable to people.
04:29:36 But at some point, you do want to get to the point
04:29:38 where you have the fully general thing
04:29:40 and you stop making bespoke specialized systems for each thing
04:29:44 and you wind up start using the higher level language
04:29:47 instead of writing everything in assembly language.
04:29:50 What about consciousness?
04:29:52 The C word, do you think that’s fundamental to solving AGI
04:29:58 or is it a quirk of human cognition?
04:30:02 So I think most of the arguments about consciousness
04:30:06 don’t have a whole lot of merit.
04:30:08 I think that consciousness is kind of the way the brain feels
04:30:12 when it’s operating.
04:30:14 And this idea that, you know, I do generally subscribe
04:30:19 to sort of the pandemonium theories of consciousness
04:30:21 where there’s all these things bubbling around
04:30:23 and I think of them as kind of slightly randomized sparse
04:30:27 distributed memory bit strings of things that are kind of happening,
04:30:30 recalling different associative memories,
04:30:32 and eventually you get some level of consensus
04:30:35 and it bubbles up to the point of being a conscious thought there.
04:30:38 And the little bits of stochasticity that are sitting on in this
04:30:42 as it cycles between different things and recalls different memory,
04:30:45 that’s largely our imagination and creativity.
04:30:49 So I don’t think there’s anything deeply magical about it,
04:30:52 certainly not symbolic.
04:30:54 I think it is generally the flow of these associations
04:30:58 drawn up with stochastic noise overlaid on top of them.
04:31:02 And I think so much of that is like it depends on
04:31:05 what you happen to have in your field of view
04:31:07 as some other thought was occurring to you
04:31:09 that overlay and blend into the next key that queries your memory for things.
04:31:13 And that kind of determines how your chain of consciousness goes.
04:31:17 So that’s kind of the qualia, the subjective experience of it
04:31:22 is not essential for intelligence.
04:31:25 So I don’t think there’s anything really important there.
04:31:28 What about some other human qualities,
04:31:30 like fear of mortality and stuff like that?
04:31:32 Like the fact that this ride ends, is that important?
04:31:37 Like you talked so much about this conversation
04:31:40 about the value of deadlines and constraints.
04:31:43 Do you think that’s important for intelligence?
04:31:46 That’s actually a super interesting angle
04:31:48 that I don’t usually take on that about
04:31:50 has death being a deadline that forces you to make better decisions.
04:31:53 Because I have heard people talk about how if you have immortality,
04:31:56 people are going to stop trying and working on things
04:31:59 because they’ve got all the time in the world.
04:32:02 But I would say that I don’t expect it to be a super critical thing
04:32:08 that a sense of mortality and impending death is necessary there
04:32:13 because those are things that they do wind up providing reward signals to us
04:32:17 and we will be in control of the reward signals.
04:32:19 And there will have to be something fundamental
04:32:22 that causes, that engenders curiosity and goal setting and all of that.
04:32:27 Something is going to play in there at the reward level,
04:32:32 whether it’s positive or negative or both.
04:32:35 I don’t have any strong opinions on exactly what it’s going to be,
04:32:40 but that’s that type of thing where I doubt
04:32:43 that might be one of those half dozen key things
04:32:45 that has to be sorted out on exactly what the master reward,
04:32:49 the meta reward over all of the local task specific rewards have to be.
04:32:54 That could be that big negative reward of death.
04:32:57 Maybe not death, but ability to walk away from an interaction.
04:33:01 So it bothers me when people treat AI systems like servants.
04:33:06 So it doesn’t bother me, but I mean it really is drawing the line
04:33:12 between what an AI system could be.
04:33:15 It’s limiting the possibility of what an AI system could be.
04:33:17 It’s treating them as justice tools.
04:33:19 Now that’s, of course, from a narrow AI perspective,
04:33:23 there’s so many problems that narrow AI could solve,
04:33:27 just like you said, in its form of a tool.
04:33:33 But it could also be a being, which is much more than a tool.
04:33:38 And to become a being, you have to respect that thing for being a being.
04:33:43 And for that, it has to be able to make its own decisions,
04:33:49 to walk away, to say, I had enough of you.
04:33:51 I would like to break up with you now.
04:33:53 You’ve not treated me well, and I would like to move on.
04:33:57 So I think that actually that choice to end things.
04:34:04 So a couple of things on that.
04:34:07 So on the one hand, it is kind of disturbing
04:34:10 when you see people being like people that are mean to robots
04:34:13 and mean to Alexa or whatever.
04:34:15 And that seems to speak badly about humanity.
04:34:18 But there’s also the exact opposite side of that
04:34:21 where you have so many people that imbue humanity in inanimate objects
04:34:25 or things that are toys or that are relatively limited.
04:34:28 So I think there may even be more danger
04:34:31 about people putting more emotional investment
04:34:34 into a lot of these proto AIs in different ways.
04:34:37 And then the AI would manipulate that.
04:34:41 But as far as like the AI ethnic sides of things,
04:34:44 I really stay away from any of those discussions
04:34:48 or even really thinking about it.
04:34:50 It’s similar with the safety things where I think it’s just premature.
04:34:54 And there’s a certain class of people
04:34:56 that enjoy thinking about impractical things,
04:34:59 things that are not in the world and of pragmatic effect around you.
04:35:03 And I think that, again, because I don’t think there’s going to be a fast takeoff,
04:35:08 I think we actually will have time to have these debates
04:35:11 when we know the shape of what we’re debating.
04:35:13 And some people do take a principled approach
04:35:15 that they think it’s going to go too fast,
04:35:17 that you really do need to get ahead of it,
04:35:19 that you need to be thinking about this
04:35:21 because we have slow processes of coming to any kind of consensus
04:35:24 or even coming up with ideas about this.
04:35:26 And maybe that’s true.
04:35:30 I wouldn’t put any of my money or funding into something like that
04:35:34 because I don’t think it’s a problem yet.
04:35:36 And I think that we will have these signs of life
04:35:39 when we’ve got our learning disabled toddler,
04:35:42 we should really start talking about some of the safety and ethics issues,
04:35:45 but probably not before then.
04:35:47 Can you elaborate briefly about why you don’t think there’ll be a fast takeoff?
04:35:52 Is there some deep intuition you have about it?
04:35:55 Is it because it’s grounded in the physical world or why?
04:35:58 Yeah, so it is my belief that we’re going to start off
04:36:01 with something that requires thousands of GPUs.
04:36:04 And I don’t know if you’ve tried to go get a thousand GPU instance
04:36:08 on a cloud anytime recently,
04:36:10 but these are not things that you can just go spin up hundreds of.
04:36:14 There are real challenges to, I mean,
04:36:17 these things are going to take data centers
04:36:19 and data centers take years to build.
04:36:22 The last few years, we’ve seen a few of them kind of coming up,
04:36:25 going in different places.
04:36:26 They’re big engineering efforts.
04:36:27 You can hear people bemoan about the fact that I know
04:36:31 that the network was wired all wrong
04:36:33 and it took them a month to go unwire it and rewire it the right way.
04:36:37 These aren’t things that you can just magic into existence.
04:36:40 And the ideas of like the old tropes about it’s going to escape
04:36:44 onto the internet and take over other systems,
04:36:47 the fast takeoff ones are clearly nonsense
04:36:49 because you just can’t open TCP connections above a certain rate,
04:36:52 no matter how smart you are, even if you have perfect hacking ability,
04:36:55 that take over the world in an instant sort of thing
04:36:58 just isn’t plausible at all.
04:37:00 And even if you had access to all of the resources,
04:37:03 these are going to be specialized systems
04:37:06 where you’re going to wind up with something
04:37:08 that is architected around exactly this chip with this interconnect,
04:37:12 and it’s not just going to be able to be plopped somewhere else.
04:37:15 Now, interestingly, it is going to be something
04:37:18 that the entire code for all of it will easily fit on a thumb drive.
04:37:23 That’s total spy movie, thriller sorts of things
04:37:26 where you could have, hey, we cracked the secret to AGI
04:37:29 and it fits on this thumb drive and anyone could steal it.
04:37:32 Now, they’re still going to have to build the right data center to deploy it
04:37:35 and have the right kind of life experience curriculum
04:37:38 to take it up to the point where it’s valuable.
04:37:40 But the real core of it, the magic that’s going to happen there
04:37:43 is going to be very small.
04:37:45 It’s, again, tens of thousands of lines of code,
04:37:47 not millions of lines of code.
04:37:49 It is possible to imagine a world, as you mentioned in the spy thriller view,
04:37:54 if it’s just a few lines of code,
04:37:58 we can imagine a world where the surface of computation is growing,
04:38:02 maybe growing exponentially,
04:38:04 meaning the refrigerators start getting a GPU.
04:38:12 First of all, the smartphones, the billions of smartphones.
04:38:15 But maybe if they become highways
04:38:20 through which code can spread across the entirety of the computation surface,
04:38:25 then you don’t any longer have to book AWS GPUs.
04:38:32 There were real fundamental issues there.
04:38:34 When you start getting down to taking an actual problem
04:38:36 and putting it on an abstract machine like that,
04:38:39 that has not worked out well in practice.
04:38:42 And the idea that there was always,
04:38:45 like it’s always been easy to come up with ways to compute faster,
04:38:49 say more flops or more giga ops or whatever there.
04:38:52 That’s usually the easy part.
04:38:54 But you then have interconnect and then memory for what goes into it.
04:38:59 And when you talk about saying, well, cell phones,
04:39:01 well, you’re limited to like a 5G connection or something on that.
04:39:04 And if you take your calculation
04:39:08 and you factor it across a million cell phones
04:39:11 instead of a thousand GPUs in a warehouse,
04:39:14 you might be able to have some kind of a substrate like that,
04:39:17 but it could be operating then at one one thousandth the speed.
04:39:21 And so, yes, you could have an AGI working there,
04:39:24 but it wouldn’t be a real time AGI.
04:39:26 It would be something that is operating at really a snail’s pace,
04:39:30 much, much slower than kind of human level thought for things.
04:39:34 I’m not worried about that problem.
04:39:36 You’re transferring the problem into the interconnect,
04:39:39 the communication, the shared memory,
04:39:41 the collective intelligence aspect of it,
04:39:44 which is extremely difficult as well.
04:39:46 I mean, it’s back to the very earliest days of supercomputers.
04:39:49 You still have the balance between bandwidth, storage and computation.
04:39:53 And sometimes they’re easier to get one or the other,
04:39:56 but it’s been remarkably constant across all those years
04:40:00 that you still need all three.
04:40:03 What do your efforts now, you mentioned to me
04:40:07 that you’re really committing to AI at this stage.
04:40:11 What do you see your life in the next few months, years look like?
04:40:15 What do you hope to achieve here?
04:40:18 So I literally just this week signed a term sheet
04:40:22 to take some investment money for my company
04:40:25 where the last two years I had backed off from Meta
04:40:29 and I was still doing my consulting CTO role there,
04:40:32 but I had styled it as I was going to take
04:40:35 the Victorian gentleman scientist route
04:40:37 where I was going to be the wealthy person
04:40:40 that was going to go pursue science
04:40:42 and learn about this and do experiments.
04:40:44 And honestly, I’m surprised there aren’t more people like that
04:40:48 that are like me, technical people that made a bunch of money
04:40:52 and are interested in some of these,
04:40:54 possibly the biggest leverage point in human history.
04:40:57 I mean, I know of, I’ve heard of a couple organizations
04:41:00 that are basically led by one rich techie guy
04:41:03 that gets a few people around him to try to work on this,
04:41:06 but I’m surprised that there’s not more,
04:41:08 that there aren’t like a dozen of them.
04:41:10 I mean, maybe people still think that it’s an unapproachable problem,
04:41:14 that it’s kind of beyond their ability to get a wrench on
04:41:17 and have some effect on like whatever startups they’ve run before.
04:41:21 But that was my kind of, like with all the stuff I’ve learned,
04:41:25 whether it’s gaming, aerospace, whatever,
04:41:28 I go through a larval phase where I’m like,
04:41:30 okay, I’m sucking up all of this information trying to see,
04:41:33 is this something that I can actually do?
04:41:36 Is this something that’s practical to devote a large chunk of my life to?
04:41:41 And I’ve gone through that with the AI machine learning space of things.
04:41:46 And I think I’ve got my arms around it.
04:41:49 I’ve got the measure of it where some of the most brilliant people
04:41:52 in the world are working on this problem,
04:41:54 but nobody knows exactly the path that it’s going on.
04:41:58 We’re throwing a lot of things at the wall and seeing what sticks.
04:42:02 But I have another interesting thing just learning about all of this,
04:42:06 the contingency of your path to knowledge
04:42:08 and talking about the associations and the context that you have with them
04:42:12 where people that learn in the same path will have similar thought processes.
04:42:17 And I think it’s useful that I come at this from a different background,
04:42:21 a different history than the people that have had
04:42:23 the largely academic backgrounds for this
04:42:25 where I have huge blind spots that they could easily point out,
04:42:29 but I have a different set of experiences in history
04:42:33 and approaches to problems in systems engineering
04:42:35 that might turn out to be useful.
04:42:39 And I can afford to take that bet where I’m not going to be destitute.
04:42:44 I have enough money to fund myself working on this for the rest of my life.
04:42:49 But what I was finding is that I was still not committing
04:42:55 where I had a foot firmly in the VR and meta side of things
04:42:58 where in theory, I’ve got a very nice position there.
04:43:02 I only have to work one day a week for my consulting role,
04:43:06 but I was engaging every day.
04:43:08 I’d still be like my computer’s there.
04:43:10 I’d be going and checking the workplace and notes
04:43:12 and testing different things and communicating with people.
04:43:15 But I did make the decision recently that I’m going to get serious.
04:43:21 I’m still going to keep my ties with meta,
04:43:24 but I am seriously going for the AGI side of things.
04:43:28 And it’s actually a really interesting point
04:43:30 because a lot of the machine learning, the AI community is quite large,
04:43:34 but really, basically almost everybody has taken the same trajectory
04:43:39 through life in that community.
04:43:42 And it’s so interesting to have somebody like you
04:43:44 with a fundamentally different trajectory,
04:43:47 and that’s where the big solutions can come
04:43:49 because there is a kind of silo,
04:43:51 and it is a bunch of people kind of following the same kind of set of ideas.
04:43:56 And I was really worried that I didn’t want to come off
04:43:59 as an arrogant outsider for things
04:44:02 where I have all the respect in the world for the work.
04:44:05 It’s been a miracle decade.
04:44:07 We’re in the midst of a scientific revolution happening now,
04:44:10 and everybody doing this is, you know, these are the Einsteins and Boers
04:44:14 and whatevers of our modern era.
04:44:17 And I was really happy to see that the people that I’ve sat down and talked with,
04:44:21 everybody does seem to really be quite great about,
04:44:24 just happy to talk about things, willing to acknowledge that
04:44:27 we don’t know what we’re doing, we’re figuring it out as we go along.
04:44:30 And I mean, I’ve got a huge debt on this where this all really started for me
04:44:36 because Sam Altman basically tried to recruit me to open AI.
04:44:39 And it was at a point when I didn’t know anything about
04:44:43 what was really going on in machine learning.
04:44:46 And in fact, it’s funny how the first time you reached out to me,
04:44:49 it’s like four years ago for your AI podcast.
04:44:52 Yeah, for people who are listening to this should know that,
04:44:59 first of all, obviously I’ve been a huge fan of yours for the longest time,
04:45:03 but we’ve agreed to talk like, yeah, like four years ago,
04:45:06 back when this was called the artificial intelligence podcast,
04:45:09 we wanted to do a thing and you said yes.
04:45:13 And I said, it’s like, I don’t know anything about modern AI.
04:45:15 That’s right.
04:45:16 I said, I could kind of take an angle on machine perception
04:45:18 because I’m doing a lot of that with the sensors and the virtual reality,
04:45:22 but we could probably find something to talk about.
04:45:24 And that’s where, when did Sam talk to you about open AI, around the same time?
04:45:30 It was a little bit, it was a bit after that.
04:45:32 So I had done the most basic work, I had kind of done the neural networks
04:45:37 from scratch where I had gone and written it all in C just to make sure
04:45:40 I understood backpropagation at the lowest level and my nuts and bolts approach.
04:45:45 But after Sam approached me, it was flattering to think that he thought
04:45:51 that I could be useful at open AI largely for kind of like systems optimization
04:45:56 sorts of things without being an expert.
04:46:00 But I asked Ilya Sutskever to give me a reading list and he gave me a binder
04:46:06 full of all the papers that like, okay, these are the important things.
04:46:10 If you really read and understand all of these, you’ll know like 80% of what most
04:46:14 of the machine language researchers work on.
04:46:17 And I went through and I read all those papers multiple times
04:46:20 and highlighted them and went through and kind of figured the things out there
04:46:24 and then started branching out into my own sets of research on things
04:46:28 and eventually started writing my own experiments and doing kind of figuring out,
04:46:33 you know, finding out what I don’t know, what the limits of my knowledge are
04:46:36 and starting to get some of my angles of attack on things,
04:46:39 the things that I think are a little bit different from what people are doing.
04:46:44 And I’ve had a couple of years now, like two years since I kind of left
04:46:49 the full time position at Meta and now I’ve kind of pulled the trigger and said,
04:46:54 I’m going to get serious about it but some of my lessons all the way back
04:46:58 to Armadillo Aerospace about how I know I need to be more committed to this
04:47:02 where there is that, you know, it’s both a freedom and a cost in some ways
04:47:07 when you know that you’re wealthy enough to say it’s like this doesn’t really mean anything.
04:47:11 I can spend, you know, I can spend a million dollars a year for the rest of my life
04:47:15 and it doesn’t mean anything, it’s fine.
04:47:19 But that is an opportunity to just kind of meander
04:47:22 and I could see that in myself when I’m doing some things.
04:47:25 It’s like, oh, this is a kind of interesting, curious thing.
04:47:28 Let’s look at this for a little while, let’s look at that.
04:47:31 It’s not really bearing down on the problem.
04:47:33 So there’s a few things that I’ve done that are kind of tactics for myself
04:47:39 to make me more effective.
04:47:40 Like one thing I noticed I was not doing well is I had a Google Cloud account
04:47:45 to get GPUs there and I was finding I was very rarely doing that
04:47:49 for no good psychological reasons where I’m like, oh, I can always think of something to do
04:47:54 other than to spin up instances and run an experiment.
04:47:56 I can keep working on my local Titans or something.
04:48:00 But it was really stupid.
04:48:01 I mean, it was not a lot of money.
04:48:03 I should have been running more experiments there.
04:48:05 So I thought to myself, well, I’m going to go buy a quarter million dollar DGX station.
04:48:10 I’m going to just like sit it right there and it’s going to mock me if I’m not using it.
04:48:14 If the fans aren’t running on that thing, I’m not properly utilizing it.
04:48:18 And that’s been helpful.
04:48:19 You know, I’ve done a lot more experiments since then.
04:48:22 It’s been it’s been interesting where I thought I’d be doing all this low level
04:48:25 envy link optimized stuff.
04:48:27 But 90 percent of what I do is just spin up four instances of an experiment
04:48:30 with different hyper parameters on it.
04:48:33 You’re doing like really sort of building up intuition by doing
04:48:37 ML experiments of different kinds.
04:48:40 But so the next big thing, though, is I am you know, I decided that I was going
04:48:45 to take some take some investor money because I have an overactive sense
04:48:50 of responsibility about other people’s money.
04:48:54 And it’s like I I don’t want I mean, a lot of my my push and my passionate
04:48:59 entreaties for things that matter.
04:49:01 It’s like I don’t want to have wasted his money investing in Oculus.
04:49:04 I want it to work out.
04:49:06 I want it to change the world.
04:49:07 I want it to be worth all of this time, money and effort going into it.
04:49:11 And I expect that it’s going to be that like that with my with my company
04:49:16 where it’s a huge forcing function.
04:49:18 This investment investors that are going to expect something of me now.
04:49:23 We’ve all had the conversation that this is a low probability long term bet.
04:49:27 It’s not something that there’s a million things I could do that I would have line
04:49:31 of sight on the value proposition for.
04:49:33 This isn’t that I think there are there are unknown unknowns in the way.
04:49:37 But it’s one of these things that it’s, you know, it’s hyperbole.
04:49:42 But it’s potentially one of the most important things humans ever do.
04:49:45 And it’s something that I think is within our lifetimes, if not within a decade
04:49:49 to happen.
04:49:51 So, yeah, this is just now happening.
04:49:54 Like term sheet, like the ink is barely virtually barely dry.
04:49:58 I mean, as I mentioned to you offline, like somebody I admire and somebody
04:50:03 you know, Andre Karpathy, I think the two of you different trajectories in life,
04:50:07 but approach problems similarly in that he calls stuff from scratch up all the time.
04:50:12 And I he’s created a bunch of little things outside of even outside the course
04:50:19 at Stanford that have been tremendously useful to build up intuition about stuff,
04:50:25 but also to help people.
04:50:27 And they’re all in the realm of A.I.
04:50:29 Do you see yourself potentially doing things like this?
04:50:34 You know, not necessarily solving a gigantic problem, but on the journey,
04:50:38 on the path to that building up intuitions and sharing code or ideas or systems
04:50:48 that give inklings of AGI, but also kind of are useful to people in some way.
04:50:55 So, yeah, first of all, Andre is awesome.
04:50:57 I learned a lot when I was going through my larval phase from his blog posts
04:51:01 and his Stanford course and, you know, super valuable.
04:51:04 I got to meet him first a couple of years ago when I was first kind of starting off
04:51:08 on my gentleman scientist bit.
04:51:11 And just just a couple of months ago when he went out on his sabbatical,
04:51:15 he stopped by in Dallas and we talked for a while and I had a great time with him.
04:51:19 And then when I heard he actually left Tesla, I did, of course,
04:51:22 along with a hundred other people say, hey, if you ever want to work with me,
04:51:26 it would be an honor.
04:51:28 So, he thinks that he’s going to be doing this educational work,
04:51:31 but I think someone’s going to make him an offer he can’t refuse
04:51:34 before he gets too far along on it.
04:51:36 Oh, his current interest is educational.
04:51:39 So, yeah, he’s a special mind.
04:51:42 Is there something you could speak to what makes him so special?
04:51:46 So, you know, like he did, he was very much a programmer’s programmer
04:51:50 that was doing machine learning work rather than,
04:51:53 it’s a different feel than an academic where you can see it in paper sometimes
04:51:57 where somebody that’s really a mathematician or a statistician at heart
04:52:01 and they’re doing something with machine learning.
04:52:04 But, you know, Andre is about getting something done.
04:52:07 And you could see it in like all of his earliest approaches to it’s like,
04:52:10 okay, here’s how reinforcement learning works.
04:52:12 Here’s how recurrent neural networks work.
04:52:14 Here’s how transformers work.
04:52:16 Here’s how crypto works.
04:52:19 And, yeah, it’s just, he’s a hacker’s, you know,
04:52:23 one of his old posts was like a hacker’s guide to machine learning.
04:52:26 And, you know, he deprecated that and said,
04:52:28 don’t really pay attention to what’s in here,
04:52:30 but it’s that thought that carries through in a lot of it
04:52:33 where it is that back again to that hacker mentality
04:52:36 and the hacker ethic with what he’s doing and sharing all of it.
04:52:40 Yeah, and a lot of his approach to a new thing, like you said,
04:52:44 larva stage is let me code up the simplest possible thing to build up intuition about it.
04:52:50 Yeah, like I say, I sketch with structs and things.
04:52:52 When I’m just thinking about a problem, I’m thinking in some degree of code.
04:52:57 You are also, among many things, a martial artist, both Judo and Jiu Jitsu.
04:53:03 How has this helped make you the person you are?
04:53:07 So, I mean, I was a competent club player in Judo and grappling.
04:53:10 I mean, I was, you know, by no means any kind of a superstar, but it was,
04:53:15 I went through a few phases with it where I did some when I was quite young,
04:53:20 a little bit more when I was 17.
04:53:22 And then I got into it kind of seriously in my mid 30s.
04:53:26 And, you know, I went pretty far with it.
04:53:28 And I was pretty good at some of the things that I was doing.
04:53:32 And I did appreciate it quite a bit where, I mean, on the one hand,
04:53:36 it’s always if you’re going to do exercise or something,
04:53:39 it’s a more motivating form of exercise.
04:53:41 If someone is crushing you, you are motivated to do something about that,
04:53:46 to up your attributes and be better about getting out of that.
04:53:49 Up your attributes, yes.
04:53:51 But there’s also that sense that I was not a sports guy.
04:53:57 I did do wrestling in junior high.
04:53:59 And I often wish that, I think I would have been good for me
04:54:03 if I’d carried that on into high school and had a little bit more of that.
04:54:07 I mean, it’s like I, you know, filch a little bit of wrestling vibe
04:54:10 with always going on about embracing the grind and like that push
04:54:14 that I associate with the wrestling team that in hindsight,
04:54:18 I wish I had gone through that and pushed myself that way.
04:54:21 But even getting back into Judo and Jiu Jitsu in my mid 30s,
04:54:25 as usually the old man on the mat with that, there was still the, you know,
04:54:30 the sense that, you know, working out with the group
04:54:34 and having the guys that you’re beating each other up with it,
04:54:38 but you just feel good coming out of it.
04:54:41 And I can remember those driving home aching in various ways
04:54:46 and just thinking, it’s like, oh, that was really great.
04:54:49 And, you know, it’s mixing with a bunch of people that had nothing to do
04:54:53 with any of the things that I worked with.
04:54:55 You know, every once in a while, someone would be like,
04:54:57 oh, you’re the Doom guy.
04:54:59 But for the most part, it was just different slice of life.
04:55:02 You know, a good thing.
04:55:04 And I made the call when I was 40.
04:55:07 That’s like, maybe I’m getting a little old for this.
04:55:09 I had separated a rib and tweaked a few things.
04:55:12 And I got out of it without any really bad injuries.
04:55:15 And it was like, have I dodged enough bullets?
04:55:18 Should I, you know, should I hang it up?
04:55:20 I went back.
04:55:21 I’ve gone a couple of times in the last decade
04:55:24 trying to get my kids into it a little bit.
04:55:26 I didn’t really stick with any of them,
04:55:28 but it was fun to get back on the mats.
04:55:30 It really hurts for a while when you haven’t gone for a while.
04:55:34 But I still debate this pretty constantly.
04:55:37 My brother’s only a year younger than me,
04:55:39 and he’s going kind of hard in jiu jitsu right now.
04:55:42 You know, he won a few medals at the last tournament he was at.
04:55:46 So he’s competing, too.
04:55:47 Yeah, and I was thinking, yeah, I guess we’re in the executive division
04:55:50 if you’re over 50 or over 45 or something.
04:55:53 And it’s not out of the question that I go back at some point to do some of this.
04:55:58 But again, I’m just reorganizing my life around more focus.
04:56:02 Probably not going to happen.
04:56:04 I’m pushing my exercise around to give me a longer,
04:56:07 uninterrupted intellectual focus time,
04:56:09 pushing it to the beginning or the end of the game.
04:56:11 Like running and stuff like that, walking.
04:56:13 Yeah, I got running and calisthenics and some things like that.
04:56:17 It allows you to still think about a problem.
04:56:19 But if you’re going to a judo club or something, you’ve got it fixed.
04:56:23 It’s going to be 7 o clock or whatever, 10 o clock on Saturday.
04:56:26 Although, I talked about this a little bit when I was on Rogan.
04:56:30 And shortly after that, Carlos Machado did reach out.
04:56:33 And I had trained with him for years back in the day.
04:56:36 And he was like, hey, we’ve got kind of a small private club
04:56:39 with a bunch of executive type people.
04:56:42 And it does tempt me.
04:56:45 Yeah, I don’t know if you know him, but John Donahart moved here to Austin
04:56:50 with Gordon Ryan and a few other folks.
04:56:53 He has a very interesting way, very deep systematic way of thinking about jiu jitsu
04:56:58 that reveals the chest of it, the science of it.
04:57:06 And I do think about that more as kind of an older person
04:57:09 considering the martial arts where I can remember the very earliest days
04:57:13 getting back into judo and I’m like, teach me submissions right now.
04:57:17 It’s like, learn the arm bar, learn the choke.
04:57:19 But as you get older, you start thinking more about,
04:57:22 it’s like, okay, I really do want to like learn the entire canon of judo.
04:57:26 It’s like all the different things there and like all the different approaches for it.
04:57:30 Not just the, you know, if you want to compete,
04:57:32 there’s just a handful of things you learn really, really well.
04:57:34 But sometimes there’s interest in learning a little bit more of the scope there
04:57:38 and figuring some things out from, you know, at one point I had wasn’t exactly a spreadsheet,
04:57:43 but I did have a big long text file with like, here’s the things that I learned
04:57:47 and here are like ways you chain this together.
04:57:50 And while, when I went back a few years ago,
04:57:53 it was good to see that I whipped myself back into reasonable shape
04:57:57 about doing the basic grappling.
04:57:58 But I know there was a ton of the subtleties that were just, that were gone,
04:58:02 but could probably be brought back reasonably quickly.
04:58:05 And there’s also the benefit.
04:58:06 I mean, you’re exceptionally successful now.
04:58:11 You’re brilliant.
04:58:13 And the problem, the old problem of the ego.
04:58:16 Yeah.
04:58:17 I still pushed kind of harder than I should.
04:58:20 I mean, that was, I was one of those people that I, you know,
04:58:23 I’m on the smaller side for a lot of the people competing and I would, you know,
04:58:28 I’d go with all the big guys and I’d go hard and I’d push myself a lot.
04:58:32 And that would be one of those where I would, you know,
04:58:36 I’d be dangerous to anyone for the first five minutes,
04:58:38 but then sometimes after that I’m already dead.
04:58:40 And I knew it was terrible for me because it made the, you know,
04:58:44 it meant I got less training time with all of that when you go and you just gas
04:58:48 out relatively quickly there.
04:58:50 And I like to think that I would be better about that where after I gave up judo,
04:58:55 I started doing the half marathons and tough butters and things like that.
04:58:58 And so when I did go back to the local judokai club, I thought it’s like, oh,
04:59:03 I should have better cardio for this because I’m a runner now and I do all of
04:59:06 this and didn’t work out that way.
04:59:08 It was the same old thing where just push really hard, strain really hard.
04:59:12 And of course, when I worked with good guys like Carlos, it’s like just the
04:59:17 whole flow like water thing is real and he’s just like.
04:59:20 That’s true with judo too.
04:59:21 Some of the best people like I’ve trained with Olympic gold medalists and for
04:59:26 some reason with them, everything’s easier.
04:59:29 Everything is you actually start to feel the science of it, the music of it,
04:59:35 the dance of it.
04:59:36 Everything is effortless.
04:59:39 You understand that there’s an art to it.
04:59:42 It’s not just an exercise.
04:59:43 It was interesting where I did go to the Kodokan in Japan.
04:59:46 I kind of the birthplace of judo and everything.
04:59:49 And I remember I rolled with one old guy.
04:59:51 I didn’t didn’t start standing, just started on groundwork and it was and it
04:59:56 was striking how different it was from Carlos.
04:59:58 He was still he was better than me and he got my arm and I had to tap there.
05:00:02 But it was a completely different style where I just felt like I could do
05:00:06 nothing.
05:00:06 He was just enveloping me and just like slowly grounded down with my arm and
05:00:10 bent it while with Carlos, you know, he’s just loose and free.
05:00:14 And you always thought like, oh, you’re just going to go grab something.
05:00:16 But you never had any chance to do it.
05:00:18 But it was very different feeling.
05:00:19 That’s a good summary of the difference between jiu jitsu and judo and jiu
05:00:23 jitsu.
05:00:23 There’s it is a dance and you feel like there’s a freedom and actually anybody
05:00:29 like the Gordon Ryan, one of the best the best grappler in the world.
05:00:33 Nogi grappler in the world. There’s a feeling like you can do anything.
05:00:37 But when you actually try to do something, you can’t just magically doesn’t
05:00:42 work.
05:00:42 But with the best judo players in the world, it does feel like there’s a
05:00:47 blanket that weighs a thousand pounds on top of you.
05:00:49 And there’s not a feeling like you can do anything.
05:00:52 You just you’re trapped.
05:00:54 And that’s a that’s a style that’s a difference in the style of martial
05:00:58 arts. But it’s also once you start to study, you understand it all has to
05:01:03 do with human movement and the physics of it and the leverage and all that
05:01:06 kind of stuff.
05:01:07 And that’s like that’s super fascinating.
05:01:09 At the end of the day, for me, the biggest benefits and the humbling
05:01:13 aspect when another human being kind of tells you that, you know, there’s
05:01:19 a hierarchy or there’s a you’re not that special.
05:01:25 And in the most extreme case, when you tap to a choke, you are basically
05:01:29 living because somebody let you live.
05:01:31 And that’s that is one of those.
05:01:33 If you think about it, that is a closer brush with mortality than than most
05:01:37 people consider.
05:01:40 And that kind of humbling act is good to take to your work then where it’s
05:01:46 harder to get humbled.
05:01:48 Yeah.
05:01:49 Nobody’s nobody that does any martial art is coming out thinking I’m the
05:01:52 best in the world at anything because everybody loses.
05:01:57 Let me ask you for advice.
05:01:59 What advice would you give to young people today about life, about career,
05:02:05 how they can have a job, how they can have an impact, how they can have a
05:02:10 life they could be proud of.
05:02:12 So it was kind of fun.
05:02:14 I got invited to give the commencement speech back at the I went to college
05:02:18 for two semesters and dropped out and went on to do my tech stuff.
05:02:23 But they still wanted me to come back and give a commencement speech.
05:02:26 And I’ve got that pinned on my Twitter account.
05:02:29 I still feel good about everything that I said there.
05:02:32 And my biggest point was that the path for me might not be the path for
05:02:37 everyone.
05:02:38 And in fact, the advice, the path that I took and even the advice that I
05:02:42 would give based on my experience and learnings probably isn’t the best
05:02:47 advice for everyone.
05:02:48 Because what I did was all about this knowledge in depth.
05:02:52 It was about not just having this surface level ability to make things do
05:02:56 what I want, but to really understand them through and through.
05:02:59 To let me do the systems engineering work and to sometimes find these
05:03:04 inefficiencies that can be bypassed.
05:03:06 And the whole world doesn’t need that.
05:03:09 Most programmers or engineers of any kind don’t necessarily need to do that.
05:03:14 They need to do a little job that’s been parceled out to them.
05:03:17 Be reliable.
05:03:18 Let people depend on you.
05:03:20 Do quality work with all of that.
05:03:22 But people that do have an inclination for wanting to know things deeper and
05:03:28 learn things deeper, there are just layers and layers of things out there.
05:03:34 And it’s amazing.
05:03:36 If you’re the right person that is excited about that, the world’s never
05:03:41 been like this before.
05:03:42 It’s better than ever.
05:03:43 I mean, everything that was wonderful for me is still there, and there’s
05:03:47 whole new worlds to explore on the different things that you can do.
05:03:51 And that, you know, it’s hard work.
05:03:54 Embrace the grind with it.
05:03:56 And understand as much as you can.
05:03:59 And then be prepared for opportunities to present themselves.
05:04:03 Where you can’t just say this is my goal in life and just push at that.
05:04:07 I mean, you might be able to do that, but you’re going to make more total
05:04:10 progress if you say I’m preparing myself with this broad set of tools.
05:04:15 And then I’m being aware of all the way things are changing as I move through
05:04:19 the world and as the whole world changes around me.
05:04:22 And then looking for opportunities to deploy the tools that you’ve built.
05:04:26 And there’s going to be more and more of those types of things there where an
05:04:31 awareness of what’s happening, where the inefficiencies are, what things can
05:04:35 be done, what’s possible versus what’s current practice.
05:04:38 And then finding those areas where you can go and make an adjustment and make
05:04:43 something that may affect millions or billions of people in the world.
05:04:47 Make it better.
05:04:49 When, maybe from your own example, how were you able to recognize this about
05:04:53 yourself that you saw the layers in a particular thing and you were drawn to
05:04:58 discovering deeper and deeper truths about it?
05:05:01 Is that something that was obvious to you that you couldn’t help or is there some
05:05:05 actions you had to take to actually allow yourself to dig deep?
05:05:08 So in the earliest days of personal computers, I remember the reference
05:05:12 manuals and the very early ones even had schematics of computers in the
05:05:16 background, in the back of the books, as well as firmware listings and things.
05:05:21 And I could look at that.
05:05:23 And at that time, when I was a younger teenager, I didn’t understand a lot of
05:05:27 that stuff, how the different things worked.
05:05:30 I was pulling out the information that I could get, but I always wanted to know
05:05:34 all of that.
05:05:35 There was kind of magical information sitting down there.
05:05:38 It’s like the elder lore that some gray beard wizard is the keeper of.
05:05:43 And so I always felt that pull for wanting to know more, wanting to explore
05:05:48 the mysterious areas there.
05:05:51 And that followed right in through all the things that got the value,
05:05:55 exploring the video cards leading to the scrolling advantages,
05:06:00 exploring some of the academic papers and things, learning about BSP trees
05:06:04 and the different things that I could do with those systems and just the huge
05:06:10 larval phases going through aerospace, just reading bookshelves full of books.
05:06:15 Again, that point where I have enough money, I can buy all the books I want.
05:06:18 It was so valuable there where I was terrible with my money when I was a kid.
05:06:23 My mom thought I would always be broke because I’d buy my comic books
05:06:27 and just be out of money.
05:06:28 But it was like all the pizza I want, all the Diet Coke I want,
05:06:32 video games and then books.
05:06:34 And it didn’t take that much.
05:06:36 As soon as I was making 27K a year, I felt rich and I was just getting
05:06:40 all the things that I wanted.
05:06:42 But that sense of books have always been magical to me.
05:06:46 And that was one of the things that really made me smile is Andre had said
05:06:50 when he came over to my house, he said he found my library inspiring.
05:06:54 And it was great to see.
05:06:55 I used to look at him.
05:06:56 He’s kind of a younger guy.
05:06:57 I sometimes wonder if younger people these days have the same relationship
05:07:01 with books that I do where they were such a cornerstone for me in so many ways.
05:07:06 But that sense that, yeah, I always wanted to know it all.
05:07:08 I know I can’t.
05:07:09 And that was like one of the last things I said.
05:07:11 You can’t know everything, but you should convince yourself
05:07:14 that you can know anything.
05:07:16 Any one particular thing, it was created and discovered by humans.
05:07:20 You can learn it.
05:07:21 You can find out what you need on there.
05:07:23 And you can learn it deeply.
05:07:24 Yeah.
05:07:25 And you can drive a nail down through whatever layer cake problem space
05:07:29 you’ve got and learn a cross section there.
05:07:32 And not only can you have an impact doing that,
05:07:34 you can attain happiness doing that.
05:07:37 There’s something so fulfilling about becoming a craftsman of a thing.
05:07:41 Yeah.
05:07:42 And I don’t want to tell people that, look, this is a good career move.
05:07:45 Just grit your teeth and bear it.
05:07:48 You want people.
05:07:50 And I do think it is possible sometimes to find the joy in something.
05:07:54 Like it might not immediately appeal to you.
05:07:56 But I had told people early on, like in software times,
05:08:00 that a lot of game developers are in it just because they
05:08:04 are so passionate about games.
05:08:06 But I was always really more flexible in what appealed to me,
05:08:10 where I said, I think I could be quite engaged doing
05:08:14 operating system work or even database work.
05:08:17 I would find the interest in that.
05:08:19 Because I think most things that are significant in the world
05:08:22 have a lot of layers and complexity to them
05:08:25 and a lot of opportunities hidden within them.
05:08:28 So that would probably be the most important thing to encourage to people
05:08:31 is that you can weaponize curiosity.
05:08:35 You can deploy your curiosity to find,
05:08:38 to kind of like make things useful and valuable to you,
05:08:41 even if they don’t immediately appear that way.
05:08:44 Deploy your curiosity.
05:08:45 Yeah, that’s very true.
05:08:47 We’ve mentioned this debate point,
05:08:49 whether mortality or fear of mortality is fundamental to creating an AGI.
05:08:55 But let’s talk about whether it’s fundamental to human beings.
05:08:58 Do you think about your own mortality?
05:09:01 I really don’t.
05:09:03 And you probably always have to like take with a grain of salt
05:09:07 anything somebody says about fundamental things like that.
05:09:10 But I don’t think about really aging, impending death,
05:09:16 having a legacy with my children, things like that.
05:09:20 And clearly it seems most of the world does a lot more than I do.
05:09:25 So I mean I think I’m an outlier in that where it doesn’t wind up
05:09:32 being a real part of my thinking and motivation about things.
05:09:36 So daily existence is about sort of the people you love
05:09:41 and the problems before you.
05:09:44 And I’m focused on what I’m working on right now.
05:09:47 I do take that back.
05:09:49 There’s one aspect where the kind of finiteness of the life does impact me,
05:09:54 and that is about thinking about the scope of the problems that I’m working on.
05:09:58 When I decided to work on it, when I was like nuclear fission or AGI,
05:10:03 these are big ticket things that impact large fractions of the world.
05:10:09 And I was thinking to myself at some level that, okay,
05:10:12 I may have a couple more swings at bat with me at full capability,
05:10:17 but yes, my mental abilities will decay with age, mostly inevitably.
05:10:23 I don’t think it’s a 0% chance that we will address some of that
05:10:26 before it becomes a problem for me.
05:10:28 I think exciting medical stuff in the next couple decades.
05:10:31 But I do have this kind of vague plan that when I’m not at the top of my game
05:10:36 and I don’t feel that I’m in a position to put a dent in the world some way,
05:10:40 that I’ll probably wind up doing some kind of recreational retro programming
05:10:44 or I’ll work on something that I would not devote my life to now,
05:10:50 but I can while away my time as the old man gardening in the code worlds.
05:10:56 And then to step back even bigger, let me ask you about why we’re here,
05:11:01 we human beings.
05:11:03 What’s the meaning of it all?
05:11:05 What’s the meaning of life, John Carmack?
05:11:07 So very similar with that last question.
05:11:09 I know a lot of people fret about this question a lot,
05:11:12 and I just really don’t.
05:11:14 I really don’t give a damn.
05:11:16 We are biological creatures that happenstance of evolution.
05:11:21 We have innate drives that evolution crafted for survival
05:11:25 and passing on of genetic codes.
05:11:28 I don’t find a lot of value in trying to go much deeper than that.
05:11:34 I have my motivations, some of which are probably genetically coded
05:11:38 and many of which are contingent on my upbringing
05:11:41 and the path that I’ve had through my life.
05:11:44 I don’t run into like spades of depression or envy
05:11:48 or anything that winds up being a challenge
05:11:52 and forcing a degree of soul searching with things like that.
05:11:55 I seem to be okay kind of without that.
05:12:00 As a brilliant ant in the ant colony without looking up to the sky
05:12:04 wondering why the hell am I here again.
05:12:07 So the why of it, the incredible mystery of the fact that we started,
05:12:15 first of all, the origin of life on Earth,
05:12:18 and from that, from single cell organisms,
05:12:21 the entirety of the evolutionary process took us somehow
05:12:24 to this incredibly intelligent thing that is able to build Wolfenstein 3D
05:12:29 and Doom and Quake and take a crack at the problem of AGI
05:12:34 and things that eventually supersede human beings.
05:12:37 That doesn’t, the why of it is…
05:12:42 It’s been my experience that people that focus on,
05:12:46 that don’t focus on the here and now right in front of them
05:12:49 tend to be less effective.
05:12:51 I mean, it’s not 100%, you know, vision matters to some people,
05:12:55 but it doesn’t seem to be a necessary motivator for me
05:12:59 and I think that the process of getting there is usually done.
05:13:03 It’s like the magic of gradient descent.
05:13:05 People just don’t believe that just looking locally
05:13:08 gets you to all of these spectacular things.
05:13:11 That’s been, you know, the decades of looking at
05:13:15 really some of the smartest people in the world
05:13:17 that would just push back forever against this idea
05:13:20 that it’s not this grand, sophisticated vision of everything,
05:13:24 but little tiny steps, local information winds up
05:13:27 leading to all the best answers.
05:13:29 So the meaning of life is following locally
05:13:34 wherever the gradient descent takes you.
05:13:37 This was an incredible conversation,
05:13:39 officially the longest conversation I’ve ever done on the podcast,
05:13:43 which means a lot to me because I get to do it with one of my heroes.
05:13:46 John, I can’t tell you how much it means to me
05:13:48 that you would sit down with me.
05:13:50 You’re an incredible human being.
05:13:52 I can’t wait what you do next, but you’ve already changed the world.
05:13:55 You’re an inspiration to so many people.
05:13:57 And again, we haven’t covered like most of what I was planning to talk about,
05:14:02 so I hope we get a chance to talk someday in the future.
05:14:06 I can’t wait to see what you do next.
05:14:08 Thank you so much again for talking to me.
05:14:10 Thank you very much.
05:14:11 Thanks for listening to this conversation with John Carmack.
05:14:14 To support this podcast, please check out our sponsors in the description.
05:14:18 And now, let me leave you with some words from John Carmack himself.
05:14:22 Focused hard work is the real key to success.
05:14:26 Keep your eyes on the goal,
05:14:28 and just keep taking the next step towards completing it.
05:14:31 If you aren’t sure which way to do something,
05:14:34 do it both ways and see which works better.
05:14:37 Thank you for listening and hope to see you next time.