chess or poker

[TimM]

[ QUOTE ]
I can't find it now, but I once saw someone do a statistical analysis on this very question. The logic went something like this:

1) identify a skill interval such that the player at the top of the skill interval beats the player at the bottom 75% of the time

2) count how many of those intervals seperate the worst player from the best player.

The guy's conclusion was that poker had more skill tiers than chess, but Go had like twice as many as poker.

[/ QUOTE ]

I was going to suggest this. In chess, the rating system makes it very easy to count the number of skill classes. A 75% expected score corresponds to a 200 point rating difference. For counting purposes the difference between USCF and FIDE ratings will not be significant, so I will use them interchangeably.

At the top of the human players, we have the 2800 FIDE players. At the bottom, the USCF scale is floored at 100. There are some players (kids) who would achieve a negative rating if it were not for this floor, but we have to question whether they are playing a legal game of chess at this point. Let's just call 200 the bottom for now. That makes about 13 skill classes. Perhaps someday the world will produce a 3000 rated chess player (whether it be man or machine), and that would make 14.

Now for poker, the problem is that unlike chess, the class interval is hard to define. In chess we have single games which can be used a discrete units. In poker it would be absurd to use single hands for this purpose. One would have to define some sort of heads-up freeze-out match to serve the same purpose as a single game of chess. Then the number of skill classes would depend on the exact structure, starting stacks, and poker variant. If we assume NLHE, then we just need to pick a blind/stack structure such that the amount of "play" in this freeze-out match loosely corresponds to the amount of "play" in a chess game.

I think that in this type of poker match, it would be very difficult to even come close to 13 or 14 distinct skill classes such that A beats B 75% who beats C 75% who beats D 75% etc., down to M or N.