LOS ANGELES--John Carmack is known to PC gamers as the lead programmer behind classics such as Wolfenstein 3D, Doom, and Quake. His latest project is Rage, a post-apocalyptic action game coming to PC, Xbox 360, and PS3 later this year. An iOS prequel, called Rage HD, was released last year, redefining what the iPad and iPhone were graphically capable of.
I spoke to Carmack during E3, and here's what he had to say about some of the most pressing current issues for game makers, including the state of PC hardware, the growth of mobile, casual, and social games, and the future of streaming games through services such as OnLive.
On making the iOS version of Rage
John Carmack: The only way Rage HD [for iOS] could have come out the way it was was by leveraging years of work on the high-end stuff. It was a profitable game for the iPhone because it was able to basically be a parasite onto this much larger project. It couldn't have been developed from scratch and look like that.
If you want to do something that's that media-rich, there are a few cases where you can justify doing that, but you wouldn't be able to do that on one game after another, because so much of the iOS market is scattershot. You throw a lot of things out there and see what sticks.
On technology versus gameplay
Carmack: I can recognize the knee in the curve where I can do things that make the graphics better than what they are right now, but not as much better as if we put all that engineering effort into things that make it more fun.
I could have been more exotic with the graphics [on Rage] and made a 30 frames-per-second game like our competitors out there, but I think we're going to stand out good enough with our graphics, so that we can run at twice the frame rate and get that really silky smooth subtle feeling of instant responsiveness. That actually matters more than which pixels I'm drawing on-screen.
E3 2011: Complete coverage
On Intel, AMD, and Nvidia
Carmack: The Intel story has been interesting. A year or a year and half ago, it was all up in the air. Could Intel sweep the field with Larrabee-based parts? Of course, AMD's got Fusion, and Nvidia has all the Cuda-based stuff, obviously. But it was interesting to see Intel go through this, and it turned out to be harder than they thought to make a product that could compete in there.
There's no doubt they've got some of the very best engineers in the world working on it. But the thinking that you didn't need the [GPU] specialization turned out to be incorrect. On the other hand, Intel's integrated parts are getting better, and we're working pretty closely with them for the first time, and we believe by the time Rage ships, we'll be running at 30 frames-per-second on Intel integrated Sandy Bridge graphics, and it's only going to get better from there.
In a couple of years, the integrated graphics parts are going to be good enough for the 90-percent solution, and that's going to put a serious pinch on the high-end graphics cards.
Hard-core versus casual gaming
Carmack: We've got the hard-core people who look down their noses at the casual gaming stuff and say, "That's not a real game. Real games require $500 GPUs." So much of what I've done has been a driving force for those people buying $500 video cards, and a lot of people look at it as a failure of this generation that you don't need a $500 video card to play a new game.
It seems like "serious gamer" is a lifestyle decision, and I'm not a serious gamer, I'm an engineer, and I'm not spending 20 hours over the weekend playing games. I have a lot of empathy for the people who just want to play a fun game on their phone for a little while.
On OnLive and streaming games
Carmack: I've played the OnLive stuff and a lot of people have just enough technical knowledge to count it out for the wrong reasons. When you talk about having a 50ms ping, that does not invalidate the process. One of the points that I make is that if you take a lot of the console games out there, and you're playing with your wireless controller, going through your post-process TV, the games themselves often have multiple frames of latency.
You get an event, you pipeline an animation, and it goes to the render thread and the GPU. A lot of games have over 100ms of latency in them right now. Now it's true that adding latency is always bad, and with OnLive, you're adding a compression step and two transmit steps.
But the laws of physics do not guarantee this to be a bad idea. I don't necessarily think any of the current players will live to see the pot of gold at the end of this rainbow, but I'd say it's almost a foregone conclusion that five or ten years from now, that's going to be a significant marketplace.
From a raw technical standpoint, it has too many positives going for it. There are negatives, but a lot of times, people will accept a big negative for a much bigger win. And the win for convenience and managing your library is huge. And the win for publishers and developers--zero piracy, instant patching, all that data gathering--are strong advantages. I don't think it's the big thing next year, but I think it's coming.