Author: Sune Fischer
Date: 14:13:36 04/02/02
Go up one level in this thread
On April 02, 2002 at 14:41:49, Uri Blass wrote: >On April 02, 2002 at 12:14:47, Sune Fischer wrote: > >>On April 02, 2002 at 11:16:23, Otello Gnaramori wrote: >> >>>On April 02, 2002 at 05:08:08, Martin Andersen wrote: >>> >>>>"I was totally unprepared to play against a modern playing program. My >>>>experience from the mid-1990s turns out to have been completely irrelevant >>>>today" >>>> >>>>This is a surprise to me. He didn't know that there has been a big increase in >>>>playing strength for chess programs in the last 7-8 years, of course due to >>>>faster hardware but also better software. >>> >>>Do you think that Gulko didn't know of the hardware and software progress...I >>>seriously doubt about this, probably it's just a good excuse ? >>> >>>>Then I would say his score of 3-5 without preparation, is excellent. >>>> >>>>http://www.kasparovchess.com/serve/templates/folders/show.asp?p_docID=20790&p_docLang=EN >>>> >>>> >>>>Martin. >>> >>> >>>In this case he affirms that only playing anticomps and with well preparation he >>>is going to prevail in some way and this is another 100 elo points for the comps >>>IMHO. >>>Humans are piece of cakes if Gulko is right... :) >>> >>>w.b.r. >>>Otello >> >> >>I think what he is saying is, that he needs a bit of time to get used to a new >>way of thinking. Computers do not make tactical blunders, so the time he spent >>looking for that was just wasted. >> >>IMHO the rating of programs would drop 100-200 elo if the programs could enter >>regular tournaments. The GMs would quickly discover some of their weaknesses, >>they would learn good anti-comp strategies and would always be able to at least >>draw the things. >> >>Does a few more plies/speed change this? >>I'm not sure, I don't think 2 plies means a lot in turns of better king safety, >>pawn eval and what else in the eval(). >> >>I have often seen very strong programs become very passive in closed positions, >>they simply do not steer clear of this, it means they might draw low-rated >>playes who just happen to know a bit of anti-comp. strategy. >> >>The fact is, that humans still have a far better positional eval than programs. >>I think even low-rated (>1600) playes has a better positional understanding, but >>they just get killed in tactics (probably why they can win with the fischer-move >>rule:). > >No >I do not think that low rated humans have better >positional understanding than computers. > >There are positions that they understand better than computers but there are >positions when the computer evaluation function is superior and in most of the >positions that practically happen the computer's >positional evaluation is better. > > >Humans may overevaluate or underevaluate some advantage when >computers often know better to give the right weights >for the different factors in the evaluation Absolutely not. If we talk only of a static eval, then it is much harder for a program to "understand" when a knight is better than a bishop. The problem with programs is they add so many small things together and do not understand the critical point of a position, e.g. they don't understand the fight over a dominating square, humans know when a passed pawn is strong and when it is weak, this is all very hard to put in a static eval. Much of this is just intuition, if we can't even put that stuff into words, then how are we supposed to program it? I imagine there must also be great problems in programming what kind of endgames that win or lose. This is something you need to know way in advance before going into the endgame. Complex pawnstructures can only be approximated in a static eval, the GM will have a much more intuative feel about what is drawn and what is won. IMO programs get by with the use a tactics, you never see them sac. if they can't see the reward in the end, strong players don't need very much calculation in when to take on h3 and when not to. The initiave is no easy to eval, but it is something humans always play for. Sure tactics is part of the game, but my point was, that humans could learn to play computers much better than they do, they still act as though it was a human opponent. Take Gulko for instance, first round lost, second tie, who knows next maybe won? He thought so, and I believe him because he was learning fast. In just a few games he will figure out his opponents, and play against _them_. So I think the programs fly high in elo because their style is unknown and surprising to many. If a human player A playes human player B both with 2500, they will remain about 2500 both of them no matter how many games they play. I would like to see if same hold true for computers, would they be able to keep their rating or would the human slowly increase his win/lose ratio to above 50%, he would IMO. -S. >Uri
This page took 0.01 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.