Author: Joost Buijs
Date: 02:35:15 07/22/03
Go up one level in this thread
On July 22, 2003 at 05:13:13, Kurt Utzinger wrote: >On July 22, 2003 at 05:05:33, Joost Buijs wrote: > >>On July 22, 2003 at 04:49:27, Kurt Utzinger wrote: >> >>>On July 22, 2003 at 04:41:41, Joost Buijs wrote: >>> >>>>Probably they won't score 30%. >>>> >>>>If we look at the worst case, the Smirin vs computer match, the computers were >>>>having a performance rating of approximately 2610. These were old engines on >>>>outdated hardware. >>>> >>>>Nowadays the hardware is about twice as fast, add 50 ELO points. >>>>The latest engines (Fritz 8, Shredder 7) are at least 55 ELO points stronger, so >>>>a very conservative estimate is that the best engines on modern hardware perform >>>>at least at 2715 level. >>>> >>>>If we look at the ELO percentage expectancy table it is easy to see that a >>>>player with ~2100 ELO, like Mr. Nemeth claims to have, has a chance of scoring >>>>about 3% in a match with a player of 2715. >>> >>> >>> Discussions about Elo are not important. I would admit: Eduard Nemeth >>> will hardly be able to draw one single game vs a 2715 Elo player but this >>> has no relation to fights vs computer programs where it is relatively easy >>> to get a draw. There is a huge difference between playing vs strong humans >>> or vs strong computer programs. >>> Kurt >> >>Personally I don't see the difference. > > Of course can you see the difference. Strong human players may blunder > in tactical situations. But they do never play moves that even 1500 Elo > would under no circumstances do. Just set up some positions with bishops > of opposite colours where one side has one or even two pawns more but > when the game is drawn. Every 1800 Elo player is able to see this and > can sometimes save a game but the computer's evaluation still show > +2.50 or something like that. > Kurt > I agree with you that there are positions that are very difficult to evaluate for a computer program, specially endings with opposite coloured bishop and pawns on both wings. But on the other hand, the midgame is so powerful that it is almost impossible to reach such an endgame. At least for me, I have a Dutch rating of about 1985, thats maybe 2100 Fide. Sometimes I come very close to a draw, but in the end the computer always outsmarts me. > >>But maybe thats just my stupidity. >>We may need a proof of principle, i'm happy to invite you or Mr. Nemeth for a >>controlled match against my dual 2800 on the playchess server.
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.