Procedural generation is laziness LOL. Brian Eno and Will Wright in one room!
Written by MM. Posted at 3:56 pm on November 23rd, 2009
9 comments.
Post a comment.
Status Updates
- Hey, we searched all over the database, but we couldn't find any status updates. Have you posted any yet?
Procedural Laziness.
What I’ve learned at University (I’m learning to be an Eletrical Engineer) is that engineers are lazy people (Our teachers atcually taught us that), cause why would you need to put more effort into a project that it requires? Instead, put effort in planing and simplifying it (and later improving it), so you get a easly useable structure that can be modified and improved at later stages easily. If you create something that is too compilcated, it will end up in endless debuging and bittersweet results (Such as spore, witch is in theory the god of games, but in reality, it’s a boring ocean of undetailed randomness), as you mentioned earlier: Don’t do something if a machine can do it for you. Of course in programming (especially in game programming) every algorithm takes cpu time so you have to draw a line between perfomance and complexity (I’ve learned that while studying Digital Technics). So you must not rely entirely on procedural on the fly stuff. Yes, it can save time and help game development to be faster, but it can also make a game too random and undetailed. A game has to look good. And that can only be achived with drawing. Did u ever played metroid? It’s not a complex engine, but it was hell of fun, and it had it’s grapics back then. Balance is the key.
P.S.: Sorry for my bad english
If you want to know my philosophy it’s that human thought is a procedural generator. If so, we might as well program an algorithm that copies human thought. And from what I learned in my short life (I majored in Computer Science) is that mathematics is an approximation of the language of nature. Since brains come from nature, math is the language of our minds.
Also I am a fan of the theory that chaos is just undiscovered patterns. This means that if a game seems too random, it is just because we don’t have enough intelligence to discover the patterns. What we have to do in that situation is actually dumb down the algorithms.
Brian Eno talks about this in the lecture talking about procedural music. The randomness must occur in constraints that our minds can process – for it to be enjoyable for us.
Too bad maths is taught in a really boring way.
1. You claim that the human mind is a procedural algorithm. No, it is not. Not in any understanding of what we consider to be an algorithm. Human-made algorithms make sense when read. They’re followable. You can understand what’s going on.
Nature-creted algorithms aren’t readable. They’re an incorehent mess that never had any creator, any thought or any intention. No engineering, just random noise. As a human, you see patterns only because seeing patterns is a human trait (humans see patterns in literal random noise and clouds too)
Nature’s algorithms just work. Scientists have managed to recreate a fly’s neural net repsonsible for detecting alignement. It works great, but even though they copied it, they have no bloody clue how it works. It just does.
2. “This means that if a game seams too random, it is just because we don’t have enough intelligence to discover the patterns.” – no, you’re dead wrong, and stupid. That means that the game is composed of a lot of patterns that don’t fit in a general, “overview” pattern. It isn’t that I’m stupid, you see, it’s that the overall image, design of the game is incoherent. That the whole game didn’t stick to one particular pattern of thought, but instead drifted off. Individual elements may be well defined but seem random in relation to each other, just because the global design for the game was screwed.
Shooting, throwing objects at enemies, confined enemy confrontations and static storyline events are all patterns that Half-Life 2 used, which created the global pattern of a linear, storyline-oriented FPS with a high level of narrative and production quality. If I’d see disjointed polygons or enemies that came from my ass, I wouldn’t assume I’m too stupid to see the pattern, I’d assume the design team was smoking pot and not there to get QA to a decent level.
3. Procedural content sucks. The strength of procedural content is that it can generate content much faster than a human being. The downside is that learns much slower than a human being. A human can quickly learn how to design maps to include the tactical placement of a shield generator in your newest patch. The level generator algorithm needs a lot more work.
The first ten maps of both procedural generation and human work will be great. After that the next procedural maps will be inferior to human-made ones. Because the algorithm can’t think.
We do have enough intelligence, but the problem lies in the fact that we would need to use our collective knowledge and intelligence in a far more effective way. Isn’t it striking how we’ve always depended on a few ‘enlightened souls’ when it comes to our greatest discoveries? We’ve ran out of the ‘easy stuff’, the big things require scientists to collaborate and perhaps we should first invest in making that work far more effective.
When it comes to recognizing patterns, all you need to do is have enough factual data and a far away enough overview. I think we’ve reach a point where supercomputers solve the mathematical pattern issues for us, or at least gives us hints about potential true patterns.
Anyways, I’m liking the Twitter feed idea, does this mean chats might be broadcast live as well? It should be pretty tempting to make this a game related feature.
Best of luck!
“Nature’s algorithms just work. Scientists have managed to recreate a fly’s neural net repsonsible for detecting alignement. It works great, but even though they copied it, they have no bloody clue how it works. It just does.”
Interesting, but wrong. We do know how it works, we also know why it works. It’s because it was tested and shaped into what it is, by evolution. It may sound or look like total hocus pocus when looking only at the end results, but it’s not, the best idea or call it the best (adapted) algorithm if you like, simply always survives.
It can’t be more simple and effective than that. The beauty is that it’s not complex at all. Very elegant.
To translate all that into a usable algorithm for games, would involve an algorithm that evolves, something that’s actually already being tried within gaming. Mostly in the AI area of gaming. I forgot what it’s called. It’s where DNA computing and artificial intelligence meet up that’s for sure, and it’s very exciting stuff.
Potentially a whole lot more interesting than prcedural content.
archont:
1. Just because you can’t read it doesn’t mean it is not an algorithm in some way or another.
2. You prove my point. If you can’t see the patterns it’s just not fun.
3. You suck. Saying that something sucks doesn’t mean anything, it’s just your subjective opinion.
PHeMoX:
Yeah it’s interesting why there isn’t so much enlightened souls right now. But it may just seem so, we’ll see in perespective in a hundred years from now.
I think you talk about genetic algorithms. They are interesting but take too long for my needs.