Author: Uri Blass
Date: 15:02:47 10/27/02
Go up one level in this thread
On October 27, 2002 at 17:29:34, Sune Fischer wrote: >On October 27, 2002 at 17:13:59, Uri Blass wrote: > >>On October 27, 2002 at 16:59:48, Sune Fischer wrote: >> >>>On October 27, 2002 at 15:46:07, Uri Blass wrote: >>>> >>>>My definition of 1 ply error is an error that computer can see a big change in >>>>the score based on a search of one ply. >>> >>>Are you kidding. That sacrifice was very deep, he calculated for 45 minutes. >>>What a computer thinks at 1 ply is irrelevant, it can't possibly have seen a >>>fraction of the considerations Kramnik did. >> >>Irrelevant. > >No, because most sacrifices are too deep for a 1 ply search (that's why they are >called sacrifices!). You give something now and hope later on to gain back more >material or checkmate, or maybe just to get a lot of initiative. > >>Playing a bad sacrifice is still an one ply error even if the selective search >>of kramnik could not see the refutation. >> >>> >>>>The mitsake of kramnik in game 5 was 1 ply error because after the mistakes >>>>programs can see in 1 ply that kramnik is losing a piece(they search the first >>>>move check and extend after the moves because they extend checks and after it >>>>they go to qsearch but it is enough for them to see that kramnik is losing a >>>>piece. >>> >>>I might agree with you if you include regular qsearch on top of that 1 ply, but >>>if they do extensions at 1 ply and actually find the refutation only at 10 plies >>>deep, then it's of course a 10 ply error. The iteration counter is just an >>>arbitrary number in that context. >> >>I think from the point of view of chess programs and most programs can see >>at depth 1 after kramnik's mistake that kramnik is losing because they do check >>extensions and do capture in the qsearch. > >By your definition a 1 ply error would depend on the program, a 1 ply error for >program X might be a 4 ply error to program Y, so I don't like this definition. > >The ply depth of the error must depend on the depth of the refutation line, and >not on the program. > >>>>The mistake of kramnik in game 6 is also 1 ply error because programs can see >>>>big change in the score based on search of 1 ply. >>>>I do not agree that the refutation is more than 1 ply. >>>>Programs usually evaluate from the first ply that the sacrifice is wrong. >>>> >>>>Playing sacrifices for the beauty and not because they are correct is a mistake >>>>of weak players and not of players at the level of kramnik. >>> >>>I don't think so, chess is art and chess players have different styles. >> >>The best chess players see it first as a sport. > >Yes, but also as an art, I know Kasparov has used that word:) > >>>As the saying goes: "when you have found a good move, look for a better one". >> >>Yes but spending a lot of time on analyzing a move that you suspect to be better >>does not make sense when you already have a good move. > >Kramnik must have thought it was deadly. > >>>Clearly Kramnik wanted to play the beautiful sacrifice, but be also wanted to be >>>sure it was correct, hence the 45 minutes of thinking. >>>He must have miscalculated, that's all (I don't conspire to him losing on >>>purpose). If his opponent had been a human, things might have been different :) >> >>I suspect that GM's will have no problem to find the right defence in 120/40 >>time control. >> >>The only case when kramnik did something similiar against humans was against >>anand and anand had no problem to win. > >Computers are well know to have a bad sense of danger, take a look at Nemeth's >games, he lures out the king and checkmates easily, for instance. Nemeth's games are faster time control and at slow time control computers can play better in most of the cases. Even in case that kramnik estimated his chances as 60% it is better for him to take no risks when he is leading the match. 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.