Schelling may well be the original practitioner of Freakonomics. Tim Harford's video is a nice exposition. It not only explains a specific phenomenon, it hints at something profound.
Ever hear the phrase "greater than the sum of its parts?" This is one of the many interesting phenomena of emergence. Economists and other social scientists are only now starting to cotton on to these insights and they will do well to keep on cottoning. Let's break this meta-lesson down into simpler parts.
First, you have actors with preferences. These preferences either create rules or have to operate within certain rules, which are devised separately. In Schelling's game, the preferences created the rules.
Second, you have the actors behaving according to said rules, which - before Schelling came along - were unpredictable. The macro-effect of the behaviors do not resemble the individual behaviors, much less the original preferences of the actors.
In other kinds of rules/preferences scenarios, the macro-effect can be totally unpredictable -- even if the rules are simple and the actors' choices limited. So we get different things happening at various levels of description, but these levels are interdependent. Economies, ecosystems and other complex systems work this way, too.
This is similar to the lessons of chaos theory offered by Ed Lorenz: small changes in initial conditions can mean big changes throughout the system. Likewise, small changes to the rules can result in consequences we did not or could not anticipate. When one couples the latter insight with the idea of different levels of description (say, micro, mezzo and macro), we have a lifetime of questions for political scientists and economists to try to answer -- hopefully before they have any power.