Modeling the Future

Many people didn't like the movie "Starship Troopers". I liked it, however. It was entertaining, if not exactly the pinnacle of awesomeness people wanted. Denise Richards was especially cute, and I really emphasized with the brainy guy.

The reason I bring this up is because, in the movie, the brainy guy joins the military and is in "games and simulations" (or something along those lines) and everyone was very impressed.

I've always known that simulation was important, but it was when I watched this movie on the big screen that I realized games were integrally related to simulation. I was only seventeen, still in high school, so I can be pardoned for my blindness before that moment.

NOW we find out that NBA Live 2005 predicted who would get into the finals and even predicted that the underdogs had the potential to win (which, evidently, they did).

I don't follow sports. I like PLAYING sports, but just watching sets my teeth on edge. I always want to see "how would it have turned out if they had done THIS instead", but that's never possible. So, without the interactivity that actually playing gives, I find sports both boring and irritating.

But as a case study, sports are exquisite. This is because sports are a closed system full of known entities. Also, the games are so popular that they are extremely realistic. Averaged over a huge number of games, you can actually see the correlation between your game and reality - and tweak accordingly. Since this has been going on for at least a decade, every year getting more tweaking and more accurate, it is now unnervingly prescient.

When you think about it, it makes sense. You're simulating a game with a game. It's not surprising that it is pretty solid.

Except that those are HUMANS your codifying. Codifying ACCURATELY. Within the boundaries of the game, the humans are grossly predictable, even if their fine behavior will behave differently. They are automita.

Other fields - such as wars, politics, and economics - are not easy to simulate. They are open worlds with hidden variables and very limited information. They also cover a much larger span of time, making it more likely that the outside world will interfere in a way not related to the simulation's primary goal. Like, say, the president having a heart attack. Or a tsunami wiping out an oil platform.

So you can't get accurate simulations.

But you CAN get RESILIENT simulations.

By setting the bounds of 'random crap' in your simulation, you can simulate a wide range of conditions and come up with the best possible set-up to deal with unforseen circumstances.

For example, you could run the NBA simulator eight trillion times with a high level of static. The static will injure a given player, or have a player's personality change due to media/police interactions, or have the coach get sick, or any number of other situations which could easily happen. The game could then tell how well the team reacts to any given misfortune. And, in fact, the coach could use this information to, say, shore up his team with an extra forward. Or he could find out that if he benches player X and puts in player Y, his team improves - even though player Y's ratings are actually lower.

THIS kind of simulation could be extended to businesses, wars, and governments. You can't tell exactly what is going to happen, but you'll know how much misfortune you can handle, and where the stress points are.

Cooool.