Author: KarinsDad
Date: 13:39:42 06/09/99
Go up one level in this thread
On June 09, 1999 at 16:01:13, blass uri wrote: [snip] > >I think that in most of the positions there is more than one perfect move > >For example it is possible that all the legal moves lead to a draw in the >opening position. > >It is possible that a short perfect game is 1.b3 b6 2.Ba3 Ba6 3.Bb2 Bb7 4.Ba3 >Ba6 5.Bb2 Bb7 6.Ba3 Ba6 with draw by repetition > >It is not clear to me that there is a losing blunder in this game. > >Suppose that there is no losing blunder in this game. > >If this is a perfect game then I believe that there are many perfect draws >between GM's and if it is not a perfect game then it is not clear to me what is >your definition for perfect game. > >Uri Well, I guess we could disagree on this until the cows come home (or even longer). But why would a perfect program make an inferior move like b3 when there are most likely better moves (at least according to all chess theory in history, which of course may not be correct)? And why would black as a perfect tablebase program play for a draw when it would consider 1.b3 a slightly inferior move for it's opponent? The perfect program would go for the win at that point. Following current chess theory, it would try to control the center, develop pieces, protect it's king, and win. On such a theoretical issue, people can rationalize just about any stance. But if you take a step back and just look at simple openings such as: 1. e4 There is theory here to indicate that this is a good (not necessarily the best) move. Chess is a wargame where the side with more space and material normally has enough of an advantage to win (there are exceptions due to the imperfect play of humans and computers). In a fully developed tablebase of all legal games/positions, it seems unlikely that the basic premises of chess would totally fall apart. Granted, certain variations which are considered good today would be proven to be inferior, but the bottom line is that chess may be a drawn game if both sides play perfectly (I think that white will win in all "perfect" variations, but that is even more up to debate than thinking that a perfect program would never lose or draw to a human). Playing a person against a perfect tablebase program would appear to be a win (virtually automatically) for the program. Let's say that on AVERAGE, there are 2 perfect moves out of any given position (more than 1 like you claim). There are also 3 great moves, only 2 of which are perfect. In a 60 move game, the chances of a human picking one of the 2 best moves throughout the entire game when he has a choice of 3 is one chance in 37 billion. And that is with an assumption of an average of 2 perfect moves (probably on the high side) and only 3 great moves (i.e. GM level, which is probably on the low side). But even with these extremely conservation assumptions, a superGM would on average make a slightly inferior move once every three moves. Even assuming that he does not do this until after move 15, this means that he would be making 15 slightly inferior moves. I would think that 15 slightly inferior moves would lead to a loss in any variation. And even if he only made the slightly inferior move 1 time in 10, he would still have 4 or 5 slightly inferior moves which would lose the game for him. I guess I just have to agree to disagree. KarinsDad :)
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.