October 9, 2003

Simulation Aggravation

by Andrew Stern · , 10:04 am

Greg Costikyan posts a strong, unhappy reaction to newsgaming.com‘s Sept 12. I was glad to read zang.org’s balanced reaction to Greg’s post. (Mind you, Greg is someone who lives a block from Ground Zero in NYC, who saw the towers fall.)

Without getting into the politics (other than to say I find Sept 12 to be a useful, thought-provoking piece, and exciting new genre for games), I’m not sure why Greg and some of his commenters are so vitriolically opposed to calling Sept 12 a simulation. Greg writes, at the height of his vehemence,

But to call this a “simulation,” as the creators do, is fucking obscene. Simulation of what? Where’s the research? What systems are simulated?

Hmm. A common definition of a computer simulation is: “the technique of representing the real world by a computer program; a simulation should imitate the internal processes and not merely the results of the thing being simulated”

Perhaps they object to calling Sept 12 a simulation since, by its own admission and intent, it presents a biased, opinionated “toy world”. It’s not meant to be an objective model of the “real world”.

But this may not be the source of Greg’s reaction. Maybe he sees Sept 12 as simply too “toy”, too small to deserve to be called a simulation, since its underlying control logic may not be much more complex than, say, Missile Command.

But, assuming the implementation is in some bottom-up rule-based way, that theoretically would allow for emergent behavior, it’s a simulation. But what is it simulating? Its own internal world logic. By this definition, a simulation doesn’t have to model “real world” logic, just be consistent within its own world. It’s simulating the logic of a non-purely-realistic worldview.

If Sept 12 weren’t implemented in a bottom-up rule-like way — if it were in fact re-creating the results of the implied internal processes in some far less flexible hard-coded-like way, not generating them from rules — it would be an emulation, not a simulation. For example, if they created gazillions of quicktime videoclips that covered all the possible permutations of when you could shoot a missile, all the patterns of buildings you can destroy, etc., and played back the correct videoclip in response to the player’s actions, that would be an extreme case of an emulation. (I’m sure my friend Adam Frank would like to me point out, if an emulation is good enough, it is indistingushable to the player from a simulation.)

By this definition of simulation, a lot of videogames not called simulations could technically be simulations: Pong, Missile Command, Defender, Pac-man, Quake, Sept 12. Or, if they’re not actually currently implemented as simulations, they could be. But most games (if not all) probably have some emulated behavior mixed in with the simulated behavior, making them what one could call “impure” simulations.

This is a major difference between games and hypertext, by the way. Games often could be implemented as simulations, potentially allowing for emergent behavior; hypertext cannot (unless you adopt a more “exotic”, simulation-like approach). Interactive fiction (e.g., Inform) has simulation elements. Personally, I think the most interesting approaches are hybrid top-down / bottom-up approaches, like IF.

Update: Minutes after posting this (when I should have been getting back to work already), I found a new paper on Gamasutra, by the Interactive Insitute of Sweden’s Craig Lindley, “Game Taxonomies: A High Level Framework for Game Analysis and Design”. In the paper, among many things, he talks about games as / versus simulations.