Raph Koster pointed out how skill usually obeys a power law: for example, if you plot out the number of people who achieve certain scores in golf, it is not a bell curve, but most people are down in the sucky range and it drops off until you get to the score that only Tiger Woods achieves.
Chess has a ranking system, and I wondered if the ranking system followed a similar law, so I did a little quick and dirty research on the Internet Chess Club server - I sent ads to play with people ranked from 600-700, and it reported how many people were eligible. I sent ads for 700-800, ditto. 800-900, 900-1000. The lowest end of the curve had 2 people in the 700s - at the high end there was one guy in the 2800s.
So - was it a power law?
No. It was a bell curve.
I don't understand. But I guess I'm not very bright; that's why my 1300 is in the bottom 10% of ICC players. All I can think is that it's something to do with the ranking system itself, which doesn't really measure skill but measures your position relative to other players. If, instead, we measured chess player's skill by some kind of more linear metric - (How about: how many CPU cycles a computer program has that they can beat half the time?) - then maybe we'd see Raph's power law.
By the way, Raph's power law has implications for those of us making mass market games. We tend to tune the difficulty of our games to the point where we think it's fun; what we don't realize is that because we've been playing our game for months or years we're at the far end of the power law and most of humanity has no interest in those ridiculous levels of difficulty. Adding nightmare difficulty levels and extreme challenges adds value for a very small percentage of our users. So why do it? Because it's cheap, and because that small percentage may be the people who recommend your game to others. If it's not cheap...you probably shouldn't bother.