Game Theory And Markets

Bloomberg (h/t Ritholtz):

 We routinely have things like the crash of 1987, or the Flash Crash of May 6, 2010, shocking events that seem to strike out of nowhere. Yet the nature of this chaos might not be as mysterious as it seems. From the right perspective -- not that of mainstream economics -- it looks like a natural consequence of a market game that often deludes investors into thinking they know how it works.
The idea comes by way of a silly thought experiment invented in 1994 by Brian Arthur, an economist then at Stanford University. Imagine a college bar with music and cheap drinks every Thursday night. Naturally, lots of students want to go. Trouble is, it’s a tiny place, and they will enjoy it only if 60 percent or fewer of them go. Otherwise, they will suffer miserably in the cramped heat. Hence, each week, every student faces a tricky decision: How to do what most other people will not do. (No cheating -- everyone has to decide at the same time.)
By tradition, economists analyze situations like this using game theory, which attempts to explain strategic interactions. It assumes that every student will think hard about the best strategy, and remain aware that others will do the same, everyone taking into account everyone else’s likely actions. By this method, the bar problem is devilish. Deductive thinking can’t give a good solution because if everyone chose one “best” solution, all students would go to the bar or stay at home, and be miserable either way. Strict reasoning falls down.
Drawing on psychology, Arthur argued that people might make decisions in more practical terms using simple theories or hypotheses. For example, a person might think, “crowded last week, should be less crowded this week,” and choose to go. Another might think differently, “crowded two weeks in a row, likely to be crowded again,” and stay at home. Psychologists have shown that people often make decisions by holding a handful of such theories in their minds, using whichever one seems recently to be working best.
Looking at the bar puzzle this way, Arthur found, you can quickly see how what happens at popular nightspots might fluctuate quite randomly from night to night. He used a computer to simulate a group of people using various theories about whether to go to the bar, and learning by trial and error. Quite quickly, the weekly attendance settled at an average of about 60 percent. But -- and this is the significant point -- the number didn’t settle down to 60. Rather, it kept fluctuating above and below in a random way, as people changed their tactics from week to week, responding to others who were also changing theirs.
Notice that there’s no “equilibrium” here of the kind that economists often like and expect.

Pretty cool.  I'd probably just find someplace else to go because fighting the crowd sucks.  That is, unless some girl I like goes there.  Then I'd put up with having a shitty time and try not to blow my stack and get in a fight.  Yep, that's right, I remember college.