Author: KarinsDad
Date: 11:11:07 06/10/99
Go up one level in this thread
On June 10, 1999 at 12:36:21, Peter Kappler wrote: [snip] >> >>Garry may be correct. He may also be incorrect. There is no way to know. But >>unless Garry can come up with some supporting evidence, his opinion is no better >>than anyone else's on a game as complex as chess (remember, that the "experts" >>said that you would fall off the edge of the world if you sailed too far). > >I must disagree. Garry's opinion *is* better than anyone else's. He is the >best player on the planet -- quite possibly the best player who has ever lived. >He has insights into the game that you and I can't even comprehend. It doesn't >mean he is right, but it does mean his opinion is the best information we have. > > >As for "supporting evidence", I didn't get to ask him for any, but here is what >I think: In 13 years of tournament chess, I have probably analyzed over one >thousand annotated games, and I have *never* seen a game that was won without a >*clear* mistake by the losing side. I *have* seen countless games that were >drawn even though one player made a significant mistake and was worse throughout >the game. I can definitely understand where you are coming from. But consider this: 1) If a player makes an extremely obvious mistake such as hanging a queen, then it is much easier to win against that player than if he had not. 2) If a player makes a fairly obvious mistake such as going down the exchange, then it is somewhat easier to win against that player than if he had not. 3) If a player makes a potential non-obscure mistake such as allowing doubled pawns, then the position may be such that there is a flaw in the position, but maybe only an expert or better player may be able to find how to take advantage of that mistake. 4) If a player makes a potential obscure mistake such as pushing a pawn, then the position may be such that there is a flaw in the position, but maybe only a GM may be able to find how to take advantage of that mistake. 5) If a player plays nearly perfect superGM chess, then any given position within the game may have a flaw, but maybe only a superGM may be able to find how to take advantage of that mistake. In ALL of these cases, if a mistake is made, a player of sufficiently great enough ability would be capable of capitalizing on the mistake and converting it into a win. It just takes better and better play to capitalize on a smaller mistake and convert it into a win than it does to capitalize on a larger mistake and convert it into a win. And it usually (and this is an extremely important point) takes longer (i.e. more moves) to convert a smaller mistake into a win than it does to convert a larger mistake. What is the difference between a mistake and an advantage? An advantage is what occurs if you can capitalize on your opponent's mistake. If both white and black are playing perfect chess, it seems possible that the small advantage of white moving first (i.e. black made a "mistake" in not going first, so white took advantage of it) may be enough to force a win with a sufficiently strong enough player (i.e. a perfect one). However, the size of this "mistake" (or the size of the advantage) may be so small that it takes 200 moves on each side for white to turn this advantage into a win. It is difficult for Kasparov to force a win against another superGM unless the other superGM makes a mistake that Garry can understand and capitalize on (or Garry accidentally falls into a superior position 6 moves later for each side due to this obscure mistake). What is the definition of a mistake. A mistake is a move which is not the best in the position (there could be several moves in a position that are equally as good). The severity of the mistake dictates how quickly the opponent can turn the game towards his advantage and how much he can do that. I think that Garry plays SO MANY draws against other superGMs because both superGMs keep the position in a state of equilibrium that it is EXTREMELY difficult at that level to find a win when your opponent is playing well. If you played chess and was the best in the world at it and 80% of your good games (i.e. the games where you and your opponent played great chess) were draws, would you not also have the opinion that chess is a drawn game with best play? Just because you yourself could not find a way to convert an advantage into a win because the advantage is just so tiny, does this mean that the win wasn't there? At my low level of play, the mistakes made are large enough that one side or the other can either force a win, or accidentally fall into a win. This can often occur within 1 to 8 ply. At the high level of play of a superGM, the mistakes made are (usually) so small that it is difficult for them to even notice some of these small mistakes. And it takes many ply (12 to 18) before it may become obvious that a mistake was made. I do not think that the superGMs have so many draws because the game is drawn by default with perfect play. I think that the superGMs have so many draws because even they cannot find the best move in every position and even they cannot take advantage of all tiny mistakes. And they especially cannot take advantage of the fact that white moves first throughout an entire game (way too complex). And one other point. When two computer programs play a game of chess, they often play for many moves where both programs think that the game is nearly equal (within a pawn). However, the game suddenly shifts drastically into one direction where one side quickly wins (within 12 to 20 ply). What occurred in this case? Well, one program accidentally made a minor mistake that was beyond it's event horizon to perceive and the other program eventually perceives the consequences of that mistake and takes advantage of it. However, the minor mistake which tactically could not be seen by the first program within 14 ply does not just give the other side a one pawn advantage. It gives him the game. It is a mistake that is so minor that it cannot easily (or even) be seen with 14 ply, but it is so major that the other side can win because of it. Does this not seem to imply to you that maybe, just maybe, even the tiniest mistake made by a superGM (he plays the best moves for an entire game, but played the second best move on just one move) would be enough for a perfect chess playing program to win not just a half pawn, but the entire game? And possibly in a very short period of time (say 20 ply). And if this is conceivable, it may also be conceivable that just the slight advantage of moving first is enough with perfect play to force a win. And even Garry's EXTREMELY great understanding of the game would not be enough to even scratch the surface of comprehending whether this is actually true or not and flipping a coin has as likely a chance of coming up with the right answer as taking Garry's word for it. Remember the example of 4 piece tablebases showing that certain "known" chess knowledge was flawed? Magnify this by the number of possible games that can be played in chess and even Garry is lost in this sea of uncertainty. When compared to this large number of possibilities and breaking it down into a forced win, a forced draw, or a forced loss, the difference between Garry's great knowledge and my miniscule knowledge is so insignificant as to be virtually immeasurable. Garry would never be able to beat a perfect chess playing program just like I would never be able to beat a perfect chess playing program. It's just that the average length of his games would be longer than mine, but they would still ALWAYS be losses. When it comes to unanswerable questions, nobody's opinion is better than anyone else's since the question is unanswerable. Any opinion on an unanswerable question is just an opinion and just as likely or unlikely to be correct. On something impossible to answer, the knowledge of the person with the opinion is irrelevant. That's why I started this long message with "consider this". My opinion here on this subject is no better or worse than yours or Garry's. > >As you say, we will never know for certain. It would be interesting to poll a >large number of GM-level players with this question. Agreed. > >--Peter > KarinsDad :)
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.