Spector: Carmack Should Stop Working On Improving Graphics

Epic Mickey director Warren Spector believes that gaming would’ve been much better today had the greatest minds in the industry spent as much time on improving A.I. and gameplay as they spend on polishing graphics.

"I've been actively trying to shame some of my fellow developers, specifically John Carmack and Tim Sweeney,"he said.

"Can you imagine what games would look like if those two guys spent as much time working on non-combat AI as they do on rendering?” he added. "Can you imagine what games we would have if John Carmack decided he wanted to create a believable character as opposed to a believable gun?"

"Those guys are way smarter than I am. I don't know how to solve the problem, but they could figure it out."

Add new comment

This question is for testing whether you are a human visitor and to prevent automated spam submissions.


WTF drug is he on?

First off... ID has NOT been even close to the forefront of graphic improvements for games since Doom 3, every single title including RAGE has been graphically lower grade since compared to the hardware available at launch. If you want to complain about lack of AI, well hardly a surprise when you look at console horsepower, you'd have to scale the game back enormously, not to mention with so many pushing at PvP games AI becomes only for single player games which is a dying breed sadly. That said I can't really think of a game with great AI, most of them tend to be heavily scripted not AI especially "epic mickey" seriously wheres the Ai in that game?


you really that stupid ??? if you are comparing graphics dont forget that rage runs at 60fps so the detail cant be that high rage is has good has anthing that came out at the time .both 60 and ps3 are only capable of so much they cant perform magic! since launch people say that graphics have got much better but i say they are a little better cos effects and tricks they use are easily spotted by pc owners resolutions have dropped and got much worse .project gotham racing 3 still has far more effects than 4 and forza 2 was simply really bad programing


Today's games suck, plain and simple, they have gotten prettier, shorter, and boring, if i see another fps i'm going to puke, they keep making the same shitty game over and over but with better graphics, the gaming industry is in decadence.


As a big gamer since the 1970's I've seen the same been said about every generation of game on every format for over 30 years now... it's systemic of the industry... 100 clone games before new gimmick changes it and moves the genres forward a notch in evolution. It's gonna be the same for decades to come so just learn to spot the lame stuff from the ground breakers, something most doting fanboy review sites are unable to do.

AI is important

I think what Mr Spector is missing is that there is no multibillion dollar industry pushing (gaming) AI the way AMD, Intel and NVidia push graphics chips. You going to spend $500 on a killer AI processing card? (I might if there were some available...). Good looks sell. Clearly an attempt to get more marketing for his game by Warren, but I like Carmack doing what he's doing -- making awesome graphics engines and not making craptastic games (Rage). Additionally, a clever AI system doesn't necessarily port between games the way a graphics engine does. A zombie AI is not the same as a human AI or a dog AI or an Alien Super Nazi Demon AI. Lets let vanGogh paint and let Motzart compose. If Mr Spector wants AI, he should STFU and get on it himself.


Thing that shoots his arguement down in flames is there is examples of older games that had to dumb down the Ai because humans simply couldn't beat the games, i seem to recall FEAR1 was an example where even the game on max difficulty setting is actually only about 10% of what it was during development of the AI routines because the testers couldn't hit the AI bots and would get gunned down.

Could you imagine...all this

Could you imagine...all this great AI, oh yeah...So how the fuck can you have nice looking AI and BELIEVABLE characters without graphics? There has to be a balance, which is what is going on today. One thing at a time douche bag.


"without graphics"????? He didn't say games shouldn't have graphics. He said that AI should be a stronger focus. The last 10 years of game development has been centered around rendering engines. The graphics have improved 100 fold whereas the AI has improved very little. I'd rather play a game with amazing AI but rendered on the Unreal 1.0 engine than play any of today's dumbed down but pretty looking consolified crap.

Gfx are pretty good now.

I kind of agree, but there's still a lot of optimization left to do, so that the games run faster and more stable, but at the same time, Graphics are really advanced now, just look at crytek's latest demo, I'm kind of amazed (but not really, I've seen another demo 2yrs ago that looked almost as good) I'm excited for the days that i play multiplayer, and I can't tell if I'm chatting with a bot or a human.

Everyone has their own field

Everyone has their own field of specialty, I'm sure there are great minds working on A.I. as well. However, graphics has always been the main selling point so obviously people like John Carmack gets more attention. A.I. is also very much an unsolved problem still, while real time graphics are just trying to catch up to hollywood.

No, creating efficient

No, creating efficient netcode isn't a simple task. It has to be more and more efficient in order to support more players per server, most games still use 32 players and that has been the standard since th elate 90's/early 2000's. There are a few games that support 64 or 128 players but those are rare. MMO's dont count for several reasons, they're hosted in many servers (clustered), have a bunch internet connections available to them and they're not really as interactive as other games. Also nowadays some games are implementing networked physics and thats a whole new level of networking.

Yeah, it's not that it's not

Yeah, it's not that it's not a simple task, but it is in fact easier to encourage multiplayer and a scripted mainstream story rather than an alternating, procedurally advancing story with adapting AI. Furthermore, so much netcode is established for popular frameworks and middlewares of popular languages, so for a regular fps game (COD, RAGE, Bioshock, Crysis etc.), it's very much the same things that goes for synchronization of in game entities regardless of engine. As such, each platform (and/or OS) sees new netcode, but there are many more iterations for each platform, done in graphics. Mainly because it relies much more on the execution of innovative algorithms than it does depend on the capabilities (and limitations) of the platform network APIs.

Procedurally generating

Procedurally generating stories? Thats new. You'd need a computer that can actually write a book on its own to do that, not to mention script the events it generated and create beliveable synthetized voices. Sorry but the first part is a bit too much for AI, it takes real intelligence to do that, or if you mean just pick bits and pieces of story that were already made and sewing them together, then you don't need AI. So putting that aside all we have is you saying graphics get updated more often that netcode, which is true. But that doesn't mean improving multiplayer any easier than improving graphics it only shows where people are focusing their efforts. As for the middleware part, its the same both for graphics and netcode, there are plenty of scenegraph middlewares around (like Ogre) and graphics evolve from their previous iterations, RAGE was built ontop of the previous engine, same goes for cryengine and others. Plus graphics code is also limited by platform API's, directX drawcalls limit on PCs (which has been the same for about a decade) is a great example, yet graphics keep improving. The fact that people look at netcode and say 'well it'd be difficult to make it any better' or 'we can't make it better' only shows how difficult it is to actually improve it. Whats the difference then? In graphics developers give you the choice for lowering the quality so the game runs on a broad range of machines while allowing the faster ones to max it out and make things look great. We don't have that with netcode, its the same for all computers even though each has a different connection with different speeds. Networking in games still has a lot of unsolved problems like how to use more bandwidth to improve gameplay without giving players with faster connections an advantage (and not hogging the connection for anyone).

I never said procedurally

I never said procedurally generating a story. And it aint' new. In the sims and minecraft, the story advances procedurally. In Wolfenstein and COD, the story is scripted. That's what i meant. You sound like you know what you're talking about, but I want to empathize that precreated assets are not needed more for a procedurally advancing story than for a scripted one. In fact, games using scripted stories are often resemble movies more and thus they use more artwork and speech created for specific events. I see your points, but there's still so much more room for messing with graphics than there is for writing new network code. Basically, we agree on the fact that graphics are easier and more likely to be worked on.

Procedural's really the wrong

Procedural's really the wrong word, it's become so buzzy lately though. We don't want procedural stories, we want dynamic, flowing, intelligent stories. I can write a procedural story generator in half an hour if I had a couple of books cut up into sequences (even a markov chain algorithm would do), having something intelligently craft stories rather than procedurally generate them however is a much meatier task.

I agree that it's become

I agree that it's become buzzy, but I always saw procedural as when The next state can be resolved by procedure, given state[] and params[]. It doesn't need to be non-player driven, all at once or with landscapes. The items dropped in diablo 3 are created procedurally. I guess "gradually changing" would have been a better term for my two mentioned games, though. I yield.

not really

I don't think AI is really the problem most games are multiplayer now a days and the ones that are single player usally just last 10 hours or so. But I have to agree tho graphics have more or less to the point it doesn't matter anymore things move so fast your not going to see the detail. I think the whole VR thing is where its going. Right now you can play games on pretty much anything cell phone tablets consoles ect and PC needs something to stand apart from that and I think VR is where its at even using facetrack or trackir makes a big difference.

The reason most games are

The reason most games are multiplayer these days is only because they're more cost-effective and easier to make. It can take only a few months to make a good multiplayer game, whereas it can take years to make even a short, good single player game. Technology is at the point where any small team can make something acceptably shiny, but a populated world that feels even remotely realistic is out of the reach of most developers. Skyrim and the new Deus Ex were in development for years with huge teams - far longer than a Call of Duty game or Left for Dead 2. Warren Spector said almost exactly this when I saw him last month, and he's been saying it for the last 12 years, and it has always been relevant - then and now - games need to change so desperately, but they won't.

You do realize people have

You do realize people have been saying VR is the next big thing since the 80's right? No one wants to wear a dorky and cumbersome helmet for extended periods of time. 3D became a fad because all it needs are glasses and even so there are people who give up on it because they dont like wearing them. Glassless 3D is coming and it will be a thing, after that we have AR(augmented reality) and holographic displays to look forward to.


VR was unattainable until now because when it was touted as the next thing 3D graphics were in their infancy. Home computers had barely been introduced to the 3D card market until the mid 90s. Now we are living in a very mature industry, the only way forward is VR, Glassless 3D is not 3D until you have true multi directional holographic projectors that are at least 20 years away. Oculus Rift on the other hand will be the Voodoo Rush of the Virtual Reality world. When it comes out in a year or two it will change the direction of gaming forever.

...maybe. Or it'll be a cool

...maybe. Or it'll be a cool gimmick, stay around a few months while it still impresses friends, then be buried under 'enthusiasts' empty cheetos packets or hocked on eBay. It's often not a medium, but what is done with it that changes the world.

Add new comment