Author: Vincent Diepeveen
Date: 16:26:47 04/09/01
Go up one level in this thread
On April 09, 2001 at 13:59:52, Uri Blass wrote: >On April 09, 2001 at 12:21:10, Vincent Diepeveen wrote: > >>On April 08, 2001 at 17:07:14, Uri Blass wrote: >> >>>On April 08, 2001 at 16:38:28, Vincent Diepeveen wrote: >>> >>>>On April 07, 2001 at 13:49:52, Christophe Theron wrote: >>>> >>>>Hell i can even beat it in 15 0 easily as it has too >>>>little positional knowledge. >>> >>>I do not believe that you can beat it in 15 0 easily. >>>You may do it in one game but I guess that you are going to lose in a match. >> >>You do not believe it but so far i have still a 100% score against >>programs. > >100% score against programs? at levels of > 90 minutes a game i have a 100% score yes. Of course we don't talk about hundreds of games. Not even about 50 games. Gotta be way less. Lost count though. But well most GMs have not played many games against programs either. Some games i was real lucky, let's start with that and some games were played years ago when hardware was slower and more important when you could easily trick them by playing 1.e4 and follow mainline of opening then they would play bad move somewhere and you simply mated them or could easily outplay them. With nowadays books that's harder, as the book is usually better as my openings theory. The biggest victory i ever had against a program was when i won my dutch national titel. I had a temporiraly 2300 FIDE rating then which dropped to 2150 when i nearly stopped playing chess for 1 year and i was 19 years old. I won the titel by luck most still say and i can't argue as i wasn't the favourite to win the title so with some draws and lucky wins i won the titel. After that i played a slow game against the fastest computer in those days. It was a dedicated computer. I forgot which computer, probably Jan Louwman still knows it. Because he was amazed that his computer was mated in just around 23 moves after 1.e4 I had white and it was classical french with dxe4 and nxe4 then i castled long and computer castled of course short. A piece sac on h7 and dang a few moves later i could mate it... >Even in 1997 it was hard to get 100% against chess programs. >I know that GM yona kossasvily was the only player who got 6 out of 6 in the >humans against machines in 1997. Kosashvili did real well, seems each country spells different i guess, yes kosashvili profited from rebel not having bishop mobility. Even DIEP at 1 ply would have played that game better. Evaluation :) So against a very aggressive fritz version i'm not amazed kosashvili had a hard time. In 1997 most programs were very passive. Nowadays they are very aggressive tuned. This is also for me a big problem in blitz i lose EVERYTHING in blitz. But note that Jeroen Noomen in Aegon tournament each game got a winning position against programs. He blundered however too many times a piece away to get 6 out of 6. Would be interesting to the lowest scores of the programs each game he blew. >It was in holland and Rebel missed a draw when better hardware could help it. It was a simple evaluation fault of rebel yes. Not a search depth problem. >I read that GM yona kossasvily prepared for this tournament by training with >chess programs. But i bet he didn't see difference playing rebel, the king and fritz. Whereas there is a big difference between these 3 programs. >The same GM lost against Fritz6 in the israeli league. >He played better in 1997 but I think he is still not worse than you. But i do better against todays programs i do better as he does. He's 2600+ FIDE I'll get IM one day but will probably then drop quickly below 2400FIDE. Now i'm 2280 something fide rated (too low fide rated as one needs to play 70 fide rated games or so to get up to 2400 from 2150). So when organizing a match kosashvili - diepeveen that'll be like 6-0 or so. When organizing a match diepeveen - fritz6 kosashvili - fritz6 then i'll look probably better as kosashvili, unless he trains a lot against fritz6. If he trains against fritz6 and suddenly plays shredder then i'll score better though. >> >>How much stronger does a human who externally plays around 2350 play >>against a computer as a 2500 rated human who btw i beat usually >>in blitz easily too as my tactics are better as average. >> >>My problem is simple to define in chess. My own problem is that >>my openings sucks compared to 2500 rated persons. Their openings >>preparement is hell better. > >I do not buy this excuse. Excuse? So you think my endgame sucks for example? hehehehe, let's ask my teammembers about this :) Coming years we play again in masterclass, so i can definitely compare my endgame technique with IMs and GMs then again :) >Karpov lost a game against 1.e4 a6 >GM abir har aven plays in correspondence games the strange opening 1.e4 c6 2.d4 >g6. > >you can get sligtly inferior position and get the opponent out of book. >I believe that players who are more than 50 elo better than you have other >advantages otherwise you could also play the c6,g6 against them. > >> >>Whether i beat it in a match, and i'm around 2280FIDE rated, >>or not this is not the key point. >> >>I think i would, but we can argue about that for a long >>period of time about that and that's what i want to avoid. >> >>The key point is that Tiger has a few weaknesses which are >>so incredible losing against strong players AS SOON AS THEY >>REALIZE THE WEAKNESS, that tiger will forever lose against >>them. >> >>Now the interesting thing is of course that grandmasters do >>not realize the weakness of those programs. >> >>A cool example is blitz. At 21 april it's again that far. >>I always reach the finals. Most GMs and IMs i beat there, >>except some dutch GMs and Vaganian. I do not beat the >>dutch GMs as they invest loads of time in the opening. >> >>Good examples are GMs like Nijboer who simply invest 3 >>minutes just in opening against me, to get a simplistic >>won position and then kill me. >> >>So somehow they have an 'opponent modelling' which is enough >>to beat me easily if they know how i play. >> >>Exactly that is what they lack when playing programs. >> >>Some have heart programs are bad in endgames, so they >>get to a complex endgame where the program is a pawn >>up with GM some compensation, but the program has only forced moves >>as all other moves lose tactical. Then after the game the GM weirdly >>doesn't even ask: "Why did it play so well in endgame?". >> >>Even today most GMs hardly know the weaknesses of the computer. >>Add to that that some GMs are incredible weak tactical and that >>they always give the program easy to play positions somehow. >> >>But now assume the opposite. If we talk about a SERIOUS match >>strong GM versus computer, where the GM is someone who usually >>doesn't blunder away pieces (like v/d Wiel is doing last year, >>he can't even find a team nowadays to play for in masterclass >>i fear for him unless he's at board 10 or so, so he won't >>play me there next season as i'm around board 3 to 5 or so). >> >>SUPPOSE the GM knows the weaknesses of the computer and the >>strong points of it, how would the GM in question play then? >> >>Ok now even worse. Assume we have a GM now whose style is >>very positional and who is tactical absolute real strong. >>Apart from Berlin defense his opening isn't weak (berlin >>defense is really stupid to play against a computer if >>you want to win from it, as you give it for free a majority >>without tough play for white to advance). >> >>What is your prediction then? >> >>>Humans had problem to beat chess programs in the Israeli league in >>>2 hours/40+1 hour/game and got clearly less than 50% and I do not believe that >>>you are a better player than the average of the humans who played in the israeli >>>league. >> >>See above, this argument is completely irrelevant and can even >>be refuted bigtime. >> >>Junior joined top tournament in germany. the worst player of the >>tournament who was kicked butt by all GMs there, this worst player >>of that top tournament completely annihilated junior. > >It proves nothing. >He was simply lucky to catch Junior in a position that Junior did not know what >to do. > >Analysis proved that Junior could play a different move earlier if you gave it >more time or less time and I believe that Junior could avoid going to the >position that it did not understand after a different move. > >Uri
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.