a-kon: John Carmack panel
We are at akon this weekend, so I'll post some iPhone-driven notes where I can. Foolishly, I took these notes from yesterday in the Notes app, so I'm doing some transcription.
Carmack started with a demo reel that ranged from Commander Keen to the latest tech5 stuff. It's notable that the biggest cheers were for Cmdr Keen, Doom, Quake, and Quake 3. Doom 3 and Quake 2 not so much. Excitement for tech5 seems high though.
Carmack observed that we are headed to theoretical limits -- we'll maybe get another decade of order of magnitude improvements. When he started, games were driven by tech tricks: let's build a game around side scrolling! Let's do fake 3d! Let's do real 3d!! Now we're into subtle tech tricks. Now the tech can pretty much support whatever, so the content (and big budgets) is the driver.
He told a story from Doom 3: they decided to devote a guy to "doing audio right." They did all this crazy stuff so that you could, as he put it, hear a fly flying around the room. It was all the really edge pushing ultra realistic stuff that goes beyond good enough. It turned out to be the most unstable part of the game. They rewrote it into a much simpler audio engine... and nobody noticed. Good enough is sufficient for most users. The crazy new tech made possible by current technology has substantially less return reward. So again -- content is the driver on mature platforms.
He's fascinated by the iPhone because it's at the start of it's technology curve. To him, it's much more possible for one tech innovator to compete because the multi million dollar budgets aren't justified -- content isn't driving it yet. If the iPhone starts displacing the ds or the psp though, that'll probably change.
He thinks digital distro is the future, even for very large things -- no optical discs in 10 years.
Current technology is also trending towards parallelism because we're hitting the power and size physical limits. But it's hard to soak that up power for things beyond graphics. You'd think it'd be great for simulation, but it's hard to scale simulation up and down. This is especially true if the simulation has gameplay impact, and if multiplay is involved, because now you have to worry about correctness.
Carmack started with a demo reel that ranged from Commander Keen to the latest tech5 stuff. It's notable that the biggest cheers were for Cmdr Keen, Doom, Quake, and Quake 3. Doom 3 and Quake 2 not so much. Excitement for tech5 seems high though.
Carmack observed that we are headed to theoretical limits -- we'll maybe get another decade of order of magnitude improvements. When he started, games were driven by tech tricks: let's build a game around side scrolling! Let's do fake 3d! Let's do real 3d!! Now we're into subtle tech tricks. Now the tech can pretty much support whatever, so the content (and big budgets) is the driver.
He told a story from Doom 3: they decided to devote a guy to "doing audio right." They did all this crazy stuff so that you could, as he put it, hear a fly flying around the room. It was all the really edge pushing ultra realistic stuff that goes beyond good enough. It turned out to be the most unstable part of the game. They rewrote it into a much simpler audio engine... and nobody noticed. Good enough is sufficient for most users. The crazy new tech made possible by current technology has substantially less return reward. So again -- content is the driver on mature platforms.
He's fascinated by the iPhone because it's at the start of it's technology curve. To him, it's much more possible for one tech innovator to compete because the multi million dollar budgets aren't justified -- content isn't driving it yet. If the iPhone starts displacing the ds or the psp though, that'll probably change.
He thinks digital distro is the future, even for very large things -- no optical discs in 10 years.
Current technology is also trending towards parallelism because we're hitting the power and size physical limits. But it's hard to soak that up power for things beyond graphics. You'd think it'd be great for simulation, but it's hard to scale simulation up and down. This is especially true if the simulation has gameplay impact, and if multiplay is involved, because now you have to worry about correctness.