During last year i played a dozen of 9x9 games against a 1d (on a turn based site) and won 50% (and I don't think it will improve much if I played some more games). On 19x19 my winning percentage against the same player during the same period was 95% over dozens of games. (all even games with alternating colors and 6.5 komi). The statistics for similar pairings between similar players on this site are similar to these, so strength differences on 19x19 between fairly strong players are much less clear on 9x9.
For me this is an indication than my play on 19x19 is closer to perfect, because in general I would expect that in near perfect play the stronger player would win almost 100% of his games (perfect komi assumed) like Lee Chang Ho 9p will win near 100% against any player who is not one of the world's top 20. My explanation would be that I lack experience and heuristic knowledge on 9x9 openings and I also feel that there is very little room for error on 9x9. So unknowingly I regularly play a game losing move early in the game. Only later in the game (which means too late in 9x9) i realize that the game is lost. This causes more randomness in my results. Dave ----- Oorspronkelijk bericht ----- Van: Don Dailey <[EMAIL PROTECTED]> Datum: woensdag, april 11, 2007 8:01 pm Onderwerp: Re: [computer-go] Sylvain's results > On Wed, 2007-04-11 at 17:49 +0100, Jacques Basaldúa wrote: > > BTW. There is another stone in the way of 19x19 computer go. > > Knowledge. > > Humans play much stronger and do much stronger judgment than in > 9x9. > > I think you said this backwards from what you intended. Obviously, > humans are closer to perfect play and understand 9x9 better than > 19x19. Someone on this group even expressed the opinion that > professional players are close to perfect at 9x9. > > At 19x19 I'm sure there is a great deal of distance to cover even > for the very top players. > > - Don > > > > _______________________________________________ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
