Author: Anthony Cozzie
Date: 10:04:38 06/01/04
Go up one level in this thread
On June 01, 2004 at 07:04:26, Vasik Rajlich wrote: >On June 01, 2004 at 04:52:32, Sune Fischer wrote: > >>On June 01, 2004 at 04:45:12, José Carlos wrote: >> >>>On June 01, 2004 at 04:39:10, Sune Fischer wrote: >>> >>>>On June 01, 2004 at 04:04:57, José Carlos wrote: >>>> >>>>>On June 01, 2004 at 03:44:59, Sune Fischer wrote: >>>>> >>>>>>On June 01, 2004 at 03:27:37, José Carlos wrote: >>>>>> >>>>>>>> >>>>>>>>It's a very powerful feature, too powerful IMO if not all engines have it. >>>>>>>>I'm quite sure even Ruffian would lose 10-90 if Crafty had aggressive learning >>>>>>>>and Ruffian just used a small book without learning. >>>>>>>>You can be of the opinion that's a fair result, I think it is pure nonsense. >>>>>>>>Granted, it demonstrates that Crafty has learning that works, but what other >>>>>>>>conclusions can you hope to draw from it? >>>>>>> >>>>>>> >>>>>>> I disagree but I think we can agree that it's a matter of taste. IMO, Ruffian >>>>>>>has a very good selective search. Using your reasoning, we could say "if Ruffian >>>>>>>beats Crafty we can draw the conclusion that Ruffian has a much better selective >>>>>>>search, but the result is not fair, it should use only null move. Otherwise, the >>>>>>>comparison is nonsense". :) >>>>>> >>>>>>Yesterday I played a few games on fics against a Crafty clone, I think it was >>>>>>already game 5 where Crafty managed to repeat a won game. >>>>>>I was very close to resigning already at move 10, the position was not lost at >>>>>>that point but I knew the game would be of course. >>>>>> >>>>>>More importantly, where is the _fun_ in that, why even play the game? >>>>>>Who in the world gets a kick out of seeing the same games over and over? >>>>>> >>>>>>-S. >>>>> >>>>> In my opinion, the fun is exactly in figuring out an algorithm to avoid that >>>>>Crafty clone beating you twice with the same line. Don't you think it is fun to >>>>>be smarter than a smart opponent? >>>>> >>>>> José C. >>>> >>>>No I prefer to focus on the algorithms and evaluation. >>>> >>>>Book learning is "fake elo", you only cheat yourself into thinking the engine is >>>>better than it really is. >>>> >>>>-S. >>> >>> Human elo is also "fake elo" by that reasoning. >> >>Yes it is in a way, IMO. >>It's hard to prevent with humans of course, the solution could be FRC :) >> >>> To me, chess is much more than >>>search and evaluation. To you, it isn't. >> >>To me chess is so much more than memorizing book lines. >>To you this is the main thing. >> >>> Ok, that's your opinion and I repect it. >> >>Ditto :) >> >>-S. >>> José C. > >Ok, it's a matter of taste of course. I'm with Sune on this one. > >Note that being a chess engine developer is different than being a chess player. >As a chess player, if you don't memorize a certain amount of theory, no matter >how much you dislike doing so, it will cost you some games. > >As an engine developer, it's perfectly reasonable to restrict yourself to the >engine algorithm itself, and let the existing tools handle the opening & >associated learning issues. It's what software engineers call "modularity" :-) > >Vas To me, book learning is just another AI opportunity. Whether or not you consider this part of the engine is a matter of taste. I think it is not that far-fetched that one could create a reasonable opening book from scratch using a BStar-ish search. anthony
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.