Friday, December 23, 2005

Step One: What's a meme?

I was once asked to clearly explain how I'm using/planning on using what I have occasionally called "memetics". "How can you control gameplay with memetics, exactly?" Today seems like a good day to start explaining that. Think of it as a Christmas present.

First things first, you'll have to learn what I'm actually saying, as opposed to what it sounds like I'm saying. That means learning which version of "memetics" I adhere to.

Of course, I call it "pattern adaptation control", not memetics... but in a sense, PAC can be considered a subset of a certain kind of memetics. Most of you are probably familiar with the concept of memetics, at least in passing: the idea that ideas can propagate and mutate. A kind of memocology.

Memetics is, at this stage, only a way of picturing something we can't properly see. The "field", if you could call it that, can't even agree on exactly what they are studying. And I disagree with just about everyone under the sun, as expected.

Most definitions of "meme" equate to "idea that can propagate" plus some additional stuff depending on who is doing the defining. The problem here is the assumption (sometimes impicit, sometimes explicit) that there is a kind of atomic "idea" format, like "atoms". Something which is a clear-cut chunk of mental real estate, like "guns are dangerous".

Of course, atoms aren't clear-cut or atomic, and neither are genes. So why is everyone assuming there's some kind of magical atomic structure for memes?

The first thing to do, when you are reading me, is to forget that idea. The idea is sillier the more you look at it. We can't store atomic ideas in our brain, even if they exist: we store a pattern of neural linkages. This is a fundamental flaw with 99.99999% of attempts to manufacture AI: the designers get this idea that they can state concepts in a clear-cut fashion and then manipulate them. They can't.

Some people might argue that even if memes aren't clear-cut in our minds, they are clear-cut in our language. That's bull, too. "Guns are dangerous." Like water pistols? Or perhaps like fire, in that even touching them harms you? Obviously, that's not what the phrase is intended to mean, but the language doesn't clearly state that. You'd have to extend the phrase into lawyerville with something like "devices producing high-velocity ballistic projectiles can cause harm because those ballistic projectiles physically damage any material they encounter."

Even that doesn't really do the concept justice, since "dangerous" isn't always synonymous with bad. Policemen have guns, policemen are dangerous - but the police are, in general, a good thing, especially in situations which require them to use those guns.

No, there are no atomic ideas, or at least none which memetics utilizes at the moment. What I mean when I say "meme" is "mental pattern". There's nothing necessarily inherent in the pattern which causes it to replicate itself, although obviously replicating patterns are more widespread than nonreplicating patterns. The very nature of "mental" allows for the re-interpretation, change, and mutation of these patterns.

The thing you need to remember is that "meme" scales. All memes are meme complexes, because there is no such thing as an atomic meme. You can always dig deeper, because every meme is built on top of and intertwined with other memes.

For example, "guns are bad" can be zoomed in on, to discover the inherent ideas of violence, accident, crime, death, and physics. You can zoom back to see that "guns are bad" is part of a much larger pattern - perhaps a holistic pacifist pattern, or the pattern of someone mentally scarred by a horrible childhood accident.

Each of these patterns is complete in and of itself - but when you move your head a bit, turn the gem, you see different facets - each of which is complete in and of itself.

The obvious problem is representing these patterns. "Memes", you could still call them, although they are so far from standard memetic theory that it would be better to call them "dingfangles".

The obvious solution is "language". Language is built over thousands of years specifically to represent these complexes. In doing so, it facilitates their transmission and error-correction. You get drift, like generations of animals, but the wild mutations are suppressed because they (A) can't be clearly expressed and (B) can't be clearly understood.

This is fine for a situation where you want to represent a pattern which already exists. For example, "she has hated guns ever since her brother accidentally shot himself when they were kids." The language is perfect for the situation. You know the pattern exactly.

Often, that's good enough. It's good enough for 90% of game work, for example.

But if you want to represent a pattern which isn't in the linguistic culture, you'll have a much harder time of it. That is why genres and technologies take a generation to be accepted: you have to "flush out" the conflicting language and "flesh out" the new language. Your grandparents probably don't like scifi. Your parents might, but they probably don't like superheroes. These are genres and concepts which took decades to come into being.

Nowadays, I can say "It's a rift in spacetime", and people will have a clue what I'm talking about. The remainder of the pattern will probably involve time travel or distant planets. If I'd said that fifty years ago, people would have looked at me blankly.

This isn't because "rift in spacetime" is an atomic concept. Far from it. "Rift in spacetime" is a particular piece of a peculiar and very idiosynchronous pattern which can be looked at in any direction for a radically different pattern.

If you want to develop a unique pattern, you have to introduce it slowly enough to build the pattern in the mind of the audience. Whether that pattern is gameplay skill or some freaky metaphysical voodoo, you take baby steps until the pattern is secure in their minds, at which point you can turn it and see something new, maybe build off that.

Pattern adaptation control handles the tracking of existing patterns, the building of new patterns, the alteration of patterns, and the changing of viewpoints. That is the core idea, explained as best I can in this language at this moment.

Now, exactly how I do that is actually simpler than you might think, and I'll discuss it next time I have the time to sit at the computer. Probably Monday.

I'll explain how to track play skill, player preference, and player capacity. I'll explain how you can use this to fuel your game or plan your spreadsheet layout or create an interactive help file.

Now that I've built a new little pattern in your minds, I'll turn it and expand on it until it is an integral part of your thought process. Then, whenever you want to communicate your reasoning behind your new methods, you'll have to explain this very idea and build this little pattern in their minds...


Craig Perko said...

Some people might say, "You can't express new ideas in old language, so you'll never be able to usefully explain PAC."

Of course, simply by saying that, those people have proved that their argument is wrong. :D

Darius Kazemi said...

You should put a link to this entry in your blog's sidebar, under a "readme first" kind of section. Since this is sort of key to 50% of all the entries you write.

Craig Perko said...

Will do.