He talks at length about the recent years of Armadillo, I think it is really interesting his gripe was a lack of oversight from his end when he wasn't active on site, while he was spending a million a year funding the project. I wonder how many of the startups around these parts could use that kind of lesson (ie, inverse of the overbearing boss).
I also love that John Carmack and Elon Musk are email buddies, and that even John Carmack isn't thinking big enough for Elon =P
A few years ago I had the same thought about Carmack. I wondered why he seemingly abandoned his vision for space travel by returning to work in the game industry on Rage. I figured Carmack probably realized that Elon's SpaceX was going to "win the race" and gave up. Or maybe Carmack is just getting old. Either way, he's still one of my heroes.
John is less than one year older than Elon. They are both hugely smart from a technology standpoint but John's knowledge goes very deep. John's focus is implementing the future of technology personally. Elon's focus is bringing sufficient capital to bring the technological future to the present.
Both came from game dev backgrounds. Carmack announced he was getting into rockets a couple of years before Musk sold paypay... and then got into rockets :)
I wouldn't be surprised if Carmack heavily inspired Musk. If not just with shareware, and online payments, then with rockets.
Also, the public .plan files he published in the 90s heavily inspired the web log/blog craze :)
These are interesting points. I think the crowd that wasn't on the web in the mid 90's may not understand just how influential Carmack was on so many of us.
Armadillo vs. SpaceX just seem like completely different scales of project. Like Armadillo had the kind of problems that hobbyists have, like they couldn't get a good enough quality of hydrogen peroxide, and they had problems sourcing tanks and such. SpaceX just throws money and people at those kind of problems.
Honestly I think it really is as simple as what he answered in this talk: he doesn't want to risk the money.
Elon on the other hand, has been willing to double down on bets involving 100's of millions of dollars or more. That speaks to a very different view of risk/reward between them (and not one I think is negative to either side).
Also, while Carmack is certainly enormously respected in his field, he hasn't worked much outside that. Elon's track record has diverse successes, and I think that means it's a lot easier for Elon to convince investors to join his endeavors.
I would say I want him to make a podcast (and would pay like $5 a month for an episode), but considering he thinks hes wasting his free time if all hes doing is learning scheme and implementing Wolfenstein in Haskell, I definitely don't want to take away of his time to educate muggles.
oh good question, I always assumed you could just download the pdf version from scribd. but you need to pay $9 for that. Who the hell gets the money? and who would ever use it ever.
His talks are unstructured brain dumps that take a lot of effort to unpack. Even so, they contain interesting experience/research/experimentation insights that make it worth the effort. I have every single keynote right from 1998 and its interesting to see how his thoughts have shaped over the years.
For people new to game dev don't miss these ..
Tim Sweeney's talk at POPL about the next mainstream programming language.
This is probably not as fun as the stuff Carmack is up to, but if anyone out there wants to see some examples of "real" stuff done in Haskell (for "typical startup" values of "real"), check out Real World Haskell. I found it to be a great read, especially after working through Learn You a Haskell.
Is anyone doing anything related to a universal memory abstraction in the Linux kernel? He speaks for around 5 minutes at the 20 minute mark about a unified memory architecture, and I just say hell yes. I think it might become one of the largest problems in the apu / dedicated card world, and just in general it is always an issue in a transition from dedicated plug in hardware cards with their own memories to integrated solutions using system memory. Having real universal reference would make all that a non problem.
Oops! No, that's my bad. I liked your comment by the way. I was just trying to give people a reference time if they were curious.
I'm still watching the VOD because I didn't catch it live. I didn't realize the timing doesn't sync. In the VOD he first talks about unified memory architecture at about 51 minutes.
This year's speech made me think what it might have been like to listen to visionaries in the past give a talk out in the middle of a town's square.
It was like the room was dead silent (not even random people coughing or opening bags of snacks) for 3 hours while one of the smartest game devs talked about various topics.
This is the automatic twitch vod they were mentioning, with the timestamp. By the way, Carmack has been going at it for more than 2 and a half hours now. Impressive!
it was more impressive last year, when he was talking for 3.5 hours while the most of the group stage games of quakelive tournament was played. there was only one quakecon stream allowed, so those games had to be democasted later.
i had a very mixed feelings: on one hand, listening carmack is very interesting, on another - i am a quake fan and wanted to see great games.
this year they have a separate stream for games, and it is great btw. absolutely top notch work from the faceit team: http://www.twitch.tv/quakecon
Could someone post timecoded highlights? I probably can't watch a 3h keynote as I'm not a gamer, but I'd like to watch the bits I'm interested in (like about rockets, Haskell etc).
It's an extremely monolog, and he tends to follow connections in his thinking between the various topic areas, so it's difficult to index.
And honestly, I'd say only 20% of the talk is directly about gaming. Most of it is about what he thinks the future will be like in terms of programming and hardware.
if you are interested in rendering, don't miss out John's talk at 5pm dallas time today: "Principles of Lightining and Rendering with John Carmack"
http://www.quakecon.org/event-schedule/
Should also be streamed on the quakecon twitch stream.
according to the keynet video, it is supposed to be a talk he has given internally at id before and was urged to redo it at quakecon.
"John will present a lecture-style presentation on the physics of light transport and rendering. He will discuss how light behaves in the real world, and the approximations and compromises that are involved in simulating the behavior with computers. Note: not for the technically faint at heart."
Lots of big hints there for developers to pick up on. Haskell for a large simulation project , GPGPU advancement, 120 hz clock with physics interpolation, benefits of functions without side effects, performance benefits from elegant design rather than complicated optimization kludges.
