Author: Robert Hyatt
Date: 06:54:50 11/02/97
Go up one level in this thread
On November 02, 1997 at 03:10:17, Hristo wrote: >On November 01, 1997 at 17:20:06, Robert Hyatt wrote: > >>On November 01, 1997 at 15:04:45, Mark Young wrote: >> >>>Looking at the scores right now, it seem to be better to have the >>>stonger chess program. Then the fastest hardware. >> >>Not sure what you mean, but Ferret has *both*. Anything can happen >>in a tournament with this many good players, but come visit ICC and >>watch Ferret play the commercial programs. It wins far more than it >>loses. This includes many games from last year against Junior even. >> >> >>So don't conclude that it can't play and only relies on speed. You'd >>be *way* wrong. > >Robert, >he is not "concluding" this. His statement seems perfectly accurate! >It's difficult to understand why are you passing judgment so quickly. >Hope you are not overreacting because "Crafty" is getting slapped around >... >it can happen even to the best! :))) >Why do you think this happened to Crafty?! Is it possible that the other >programs are very good also?! > >Regards. >Hristo I interpreted what he wrote as saying that the programs running on the ultra- hot processors were not that good, which is certainly not correct. As far as crafty's play, I really don't know what happened, yet. I have given two "speculations" already: 1. The programs playing in Paris are *very* good. I've never suspected otherwise. In fact, I'd bet that the top 2/3 of the field are *very* close in terms of chess skill. 2. I was somewhat concerned about how Crafty would do from the beginning. I've been working very hard to solve some difficult problems posed by human players that know how to play against computers. And this over-aggression can easily blow up. Whether this happened or not I don't know yet, but I am going to study the games carefully. 3. The probability of crafty getting whacked is very high, and with this many programs entered, and with the accelerated pairings in round 1 and 2 forcing the stronger programs to face each other two rounds earlier than normal made this a little more likely. IE if you do (as I do from time to time) play a long series of games against another program, you see this sort of unexpected results. I played a 20 game match against gnuchess after seeing Crafty lose 3 in a row in the middle of the tournament. I saw crafty win the first 7 in a row, then draw 1, then lose 5 in a row, before winning all but 1 of the last 8. If those games in the middle are the ones it plays in a tournament, it is going to get whacked. Statistically a few good programs are going to do badly because of this. Not much to be done about it except to work harder. Whether we lost due to poor tuning, bad luck, or crafty simply wasn't good enough to compete, I don't know yet. It's interesting in a way, because thorsten was sure CSTal was going to do poorly based on how it played against Crafty, when, in reality, it ended up playing quite well. Whether it's result in Paris was random luck, or whether its results against Crafty were random bad luck, or whatever, is difficult to say. Ferret is a similar story, although, as I've said many times, I believe it is better than Crafty. I believe it is better than any of the rest, in fact, but that is based on hundreds of games vs commercial programs, not on just eleven. This is the reason that many commercial programs choose to not participate. It is difficult to win an 11 round event with 24+ *very* strong programs playing. Congrats to junior and the rest for "surviving." There is one good conclusion you can draw, however. Every year, a different winner, and a different #2, etc. Things are *very* close now, as opposed to 5-6 years ago when it was commercial at the top, amateur at the bottom. That *has* to be a good sign.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.