Thursday, February 16, 2006

Models for Games?

Raph covers Will Wright's speech at a conference. I love the subtext to Will Wright's speeches: if you haven't noticed, he more or less skips anything he knows inside and out. He'll go into great detail on a lot of interesting things, but as soon as he comes to a concept he's already applied on a mass scale, he just says a few quick sentences. I like that: it says he's doing this as much for himself as for anyone else.

Anyhow, I think that having different ways of looking at game design is very beneficial. However, in the end, looking at games in terms of cinema, car racing, biology, or stage magic is going to be replaced with looking at games as games. After all, cinema was first looked at as photography or thrill rides. It wasn't until the idea of cinematic timing came in and turned cinema into its own art.

Looking at games in terms of other types of systems is fine. But, in the end, games are games. They have a unique methodology. And the true challenge is not to figure out what they are most like, but what piece of them is unlike anything else. Then, master that piece.

The problem with that is that we're calling lots of things "games". Real-world car racing is a game. Gambling on yak fights is a game. Seeing how long you can hop on one leg is a game.

Similarly, I make quite a lot of games. 95% of them are not computer games! They are role-playing games, or card games, or board games, or anything else under the sun. Do all of these "games" have one, connecting thing which isn't shared in other systems?

Originally, I wrote a long essay about the various things that it could be, but that took up a lot of space. Let me spill it for you: it isn't interactivity. It isn't "fun". It isn't complex systems. It's weighted feedback loops.

All games, from the stock market to yak racing to poker all have feedback loops. But more than that, these are loops with high-grain rewards. You can win or lose in a wide variety of intensities. "Oh, man, Wonder Yak lost by just a spittle's length! I was so close!"

The stock market, by this definition, is a game. You can win or lose in a wide variety of ways and intensities.

But it's only a game if you're playing on those rewards. Many people who have been playing the market for a long time start to get jaded. They set other rewards, such as "continue making at least 12%". The feedback is no longer "won by 11.3%" or "lost by 2.9%", it's now often reduced to "won" or "lost". That's not enough granularity. That feedback is too simple.

Similarly, driving is a complex feedback loop, but there's no real granularity. There's any number of "lose" intensities, from "you made that guy honk" to "you crashed into a yak-racing arena". However, there's no "win" intensities. If you do "good enough", you do good enough. Not enough feedback!

At the heart of every good game is a beautifully reactive reward system stapled to a feedback loop. First person shooters are excessively good at this, with a multi-axis set of reward/penalties. Like health. Yeah, you can run out of health and lose. But more often, taking a hit of any kind lowers your health only somewhat. You can use that as a measure of your success so far.

All good games have these kinds of clear, high-grain feedback.

They don't need "goals". They don't even need to make sense. The things that set games apart from every other kind of system is that weighted feedback loop.

And I'll tell you more about it sometime soon. Or, at least, what I think about it.

2 comments:

Patrick Dugan said...

I think thats one of the best definitions I've heard yet, as "challenge", "goal-orientation" and ect. can be very problematic. Salen and Zimmerman describe an inherent uncertainty as the basis for games, and it seems that what you describe as the basic feature of games is a sculpting of uncertianty.

I've been playing a good indie game this weekend called Tower Defence, and playing through survival mode twice, attempting different phase transitional strategies, I've seen feedback loops occuring right before my eyes. I rescind the idea of a singular optimal play path, but I think that dynamics can be mapped based on how the "hill tops" in the optimality topography bifurcate along dualitistic factors, or (trifurcate?) trinities of factors organized in intransitive heterarchis. For instance, in the Tower Defence game, you can invest Exp in Fire, Cold or Poison, in addition to more generally useful upgrades of fire speed, range, damage, which are another layer of intransitive factors and cost less. I found that a sudden investment in one of the three would yield different benifits (obviously) which when deployed near enough to the hilltop, would set you up to keep safe for a several waves, allowing you plenty of cash to up that form of defence, and bang, feedback loop.

Eventually the rising tide of monster strength requires a transition to another complimentary defense tactic, and these sort of transitions are another thing worth studying I'd expect.

Craig Perko said...

Yup! The transitions from one kind of play to another are, in fact, what the support loops are all about.