Author: Robert Hyatt
Date: 06:53:42 07/21/98
Go up one level in this thread
On July 21, 1998 at 02:05:06, Don Dailey wrote: >>>>>What I'm saying is that computers can still overtake us without >>>>>being better in every single way. This is already the case with >>>>>me personally. I am not very strong as a chess player and have >>>>>never been over 2000 USCF, but I can still see my program make >>>>>errors that I would not make. It's getting much rarer now but >>>>>it still happens. And yet I am forced to concede that my >>>>>program is much better than I am. >>>>> >>>> >>>>And I have the same problem. But I don't get beaten because I get killed >>>>positionally, I get beat due to tactical mistakes. That's not nearly so >>>>common at the GM-level, although they do make them. But, when they play >>>>computers, they are also good enough to create positions where they are not >>>>likely to make tactical mistakes, because the computer's position doesn't offer >>>>any tactical chances (Anand vs Fritz, game 1, reference, where fritz had >>>>absolutely no chance whatsoever.) >>>> >>>>If Anand mixes it up with Fritz, he's going to lose some. But if he chooses >>>>to play into positions where tactics don't mean much, he's not going to have >>>>to think much at all. >>>> >>>>That's the point here. If I was good enough to steer games away from the wild >>>>tactical slugfests, I'd be willing to play any micro for money, and go home >>>>rich. Because I know more than any I have seen. But if it turns wild, then >>>>they have a huge advantage. But at the GM level, the GM's are quite good enough >>>>to simply not allow the micro advantage to become an issue. And if you take out >>>>tactics, you take out 90% of what a program is good at. Leaving very little >>>>hope against a good GM (or even good IM) player. >>> >>>Have you read my latest post on the subject? I make the point that >>>HAVING to steer away from these computer friendly positions is a >>>serious handicap in itself. Did you see my analogy with Kasparov >>>vs Deep Blue? You don't get something for nothing. Kasparov >>>had to give up something to get his positional games. It's my opinion >>>that he gave up too much because it's not the kind of chess he is >>>best known for. >> >> >> >>no.. you miss the point. There is *nothing* that says that a GM can't >>"play his own game." Nothing at all. Forget Kasparov. He followed bad >>advice, practiced against Fritz thinking it was preparing him for Deep Blue, >>and he got just what he deserved. But did you watch Anand vs Fritz? Anand >>didn't "play weird chess".. he just didn't let Fritz have a chance tactically. >> >>That's quite different from what you are thinking, I'm afraid. You seem to >>be assuming that avoiding an open position means giving up too much. I'm >>saying that a GM can *do both*. Avoid tactics *and* play his own game. Anand >>proved it against Fritz. Let's watch against Rebel then continue this... > >You are missing the point, not me. I AGREE with your point, you >completely ignore mine. > >Your point is that Anand is skillful enough to avoid tactics *and* >play his own game. This is not the point I was making. > >I will have to resort to a more extreme example. Let's say that >I magically developed the amazing ability to win EVERY game where >my opponent castles king side. But in every other way I did not >improve at all. When you played against me, you would eventually >learn that you could never EVER castle king side. I can imagine >you saying, "Oh, that's no problem, I'll just never castle king >side!" Do you see the problem? Even though you have managed >to avoid my strengths, I still have extracted an advantage from >this ability. You might argue that you are good enough to beat >me anyway, but that is a completely different point. And it's >not the point I am arguing. > >- Don I don't see the problem. Because if you can win every game where I castle kingside, and I figure this out, I take one or two extra moves and castle queen-side. And I roll your program up into a small wad. So such a band-aid won't cut it against a GM player. It would only work *if* you could win every such game, and *if* you could also punish me for taking 1-2 extra moves to make way to castle queen-side. But I don't think you'll find a GM that would say "damn.. I can't castle kingside against this thing... so I'm simply going to give up and move on to the next opponent." And that was *my* point.. that a GM is not so "narrow" in his playing ability that being unable to castle kingside would serve as some sort of psychological blow that would make him feel unable to play or win at all. I know it wouldn't bother me. And I've seen many GM's play computers where they lock the center pawns to block the position, and then *never* castle since there's no way to break through in the center. And it doesn't seem to "psych them out" at all. I believe they are *much* more flexible than you give them credit for being. Otherwise when a program castled long in positions where short was normal, the GM would be confused. Generally they are not. I think a better example of this sort of "fix" is a dog that poops on the side of the street. GM's step over or around it without giving a moment's thought and without disrupting their journey at all... And I think that they would do just fine if the rules of the game said "no kingside castling, ever". Just like they adapt to bizarre games like bughouse or "losers chess" and are impossibly strong in such games (or maybe Fischer-random as another example.) In short, I don't think this would be an issue at all, because you only closed one door, but many are left open. When a program wins every game that the opponent castles kingside in, *and* wins every game where the opponent blocks the center and stays there, *and* wins every game where the extra 1-2 moves to castle are required, *then and only then* does the changes to the program pay off. Does no good to erect one wall, when there are so many other holes that the GM can slip in/out through.
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.