Author: Thomas Mayer
Date: 07:05:42 05/22/05
Go up one level in this thread
Hi Jonas, >> I think because players around or beyond 2000 Elo can beat or at least draw >> Hydra quite often. At least more often then the rating difference (they >> claim to be around or even higher then 3000 on their webpage) allows. > If it happens often i would like to see the games (don't forget to include > the losses of these players too) well, Pablo did reach above 2400 once in the machine room at chessbase server. I think that speaks for itself. (Sorry, I do not have the PGN's, but Pablo has them, I believe) >> In my opinion the hole story is about how good are computers in chess >> really. As long players around 2000 with a special system can achieve wins >> and draws as much as they want, they might be not as strong as some rating >> lists, e.g. the SSDF makes us believe. > I think that there is an illusion of some sort here, we only see the odd win > on time (which i am absolutely sure would not occur in a serious match) and > the occational draw, but they don't post the vast amount of lost games that > is a result of their attempt to get a positive result. That's wrong, Jonas, I have seen him play. It's unbelieveable, but it really looks like that he (Pablo) can get at least a draw most of the time against most engines. No matter what the time control is. When the increment is 0 he has usually quite good chances for a win. He even sacrifice something quite often short before the 50 move rule so that the game goes on. Of course this does not work all the time, but if he would not sacrifice anything to prevent the 50 move rule he would have draw anyway... Really unbelieveable, you must see it yourself. > When all this is said i agree that the computers have weaknesses that humans > can exploit through special sytems and such, but i haven't seen one > successful (in serious events at classical time controls) since Kramnik's > Stonewall some time back against Junior i think and a couple of games in the > man v machine games following that. I don't consider winning on time a queen > and two rooks down successful, at best it points out where programmers need > to work. well I have no doubts that something that is possible at 60+0 is also possible at 40/2+20/1+60... Because the problem is not the time (of course because the final 0 increment the engines loose) - the problem is the engine itself, the algorithms have some weaknesses. In my opinion the fight against another engine is in fact something different to the fight against humans. That's what in fact the games by Pablo teach to us. It is right, that Pablo needs against a new engine initially some games to test and find a way to block the position, but finally he can repeat it again and again - even when the opening differs all the time... > I am sure that under serious conditions at classical time controls, they > would not be remotely as successful in getting the odd results as they seem > to be at times now. Kurt Utzinger plays not really unusual chess. Anyway he can achieve many draws in longer time controls (as he said himself about 3 out of 10). The score of 1,5-8,5 is definitely way better then the difference according to the elo system should be when we consider Fritz, Shredder etc. all above 2700. In fact it should be almost always 0-10. Greets, Thomas
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.