Author: Bob Durrett
Date: 18:14:52 11/06/02
Go up one level in this thread
The idea of programming a chess engine to play PROVOCATIVE anti-human chess is very intriguing. Perhaps an otherwise very strong program, like DF, could be programmed to play in "Provocative Anti-human Mode." It might help to provide the engine with the approximate playing strength of it's human opponent so that the provocations would be consistent with the vulnerabilities of the human. Provocative chess against top GMs would have to be much more subtle, for example. Bob D. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ On November 04, 2002 at 19:46:33, stuart taylor wrote: >On November 03, 2002 at 22:06:58, Jon Dart wrote: > >>On November 03, 2002 at 19:29:38, stuart taylor wrote: >> >> >>>Don't you believe that many masters could also fall prey to (atleast my version >>>of) Arasan, if they are not on top form? Atleast, if they don't try to play >>>anti-computer? >> >>Of course. I've seen it happen. >> >>>I'd expect that a master would never play that program in public without having >>>his openings prepared for it. >> >>Lots of Masters and even higher-rated players play on the ICC (Internet Chess >>Club), mostly blitz, and mostly for fun. I guess this counts as "in public", but >>no money's at stake. >> >>>I might also guess, that even if a master could cope with it, he would get tired >>>out very easily, as it is constantly playing tricks, more than other programs, I >>>think. >> >>Amateur players often fall victim to rather shallow combinations. They may be >>unexpected, but in fact aren't difficult for the computer to discover. Many >>programs will find these traps in a few seconds. It's rather uncommon for a >>professional-class player to be tricked this way. They see tactical shots before >>they happen, and are often setting them up for their opponent. >> >>That said, I've seen even some Expert-level players really tear computers apart. >>If they get it in a position they know how to play, especially one that features >>a gradual buildup of attack (the Classical King's Indian is a good example), >>then they can be deadly. >> >>Also if a player really wants to draw and can get the program into a quiet >>position with a static pawn structure, often the program (mine, anyway) can't >>make progress. >> >>--Jon > >I'm saying that of course, all good programs see all these tactical shots, and >so do good humans. But Your program (5.2) provokes them endlessly, which I >havn't seen with others. Possibly, when Arasan gets stronger, it might lose some >of this feature, but the way it is, in Arasan 5.2, it simply thwarts anything >good. Let's put it this way, I might loise quicker than someone 200 elo points >below me, simply because I am more likely to fall into the traps which are aimed >at me, more than they are aimed at the one of 200 points lower. > > Actually, I just now played it and was extremely determined not to be >aggressive myself, but to be closed and cold. What followed was that I was a bit >upset that, after acheiving what I wanted at the beginning, Arasan ripped off >two pawns in a queen excursion. I thought that would be fine with me, but I >didn't want to be tempted to play that type of a game. But I went on, and I >really got Arasan a bit on the ropes. I was sure there must be some great >combinations around the corner, but NO. But then I thought that te first part is >clearly OK anyway. So I offered a piece. It rejected it! So that made me feel I >was right, but since it complicated further, I was suspicous. > Eventually, It suddenly gave up its queen which left me materially superior. >I had been sure there was something even better than what I was playing, but I >decided I'd analyze afterwards. First I want to do it all myself. > Eventually, although I was well up, I couldn't see how to break a certain >cycle, to have more than a draw. But then, the computer suddenly resigned (in a >drawn position). > Then I was very eager to analyze, and went to "start". BUT I WAS SO EAGER, THAT >I PRESSED "GO" INSTEAD OF ARROW. That lost all the game. >I don't know why I saying all this story, but It's a bit exciting, because it >has just happened. >S.Taylor
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.