Author: Robert Hyatt
Date: 07:19:59 06/13/01
Go up one level in this thread
On June 13, 2001 at 02:26:23, odell hall wrote: > > > > Bob that's just it, Tactics is alot more important then you realize, humans >will never overcome the problem of tactics, some one once said tactics is 90% of >the game. Garry kasparov problem with deepblue was tactics, he had superior >positions most games, but he always had to walk though a tactical minefield, and >the computers had the advantage of having calm nerves, If what you say is true, >that computer play positional chess at the 2000 level, then why does your friend >Roman lose 70% of his games to crafty??? Because the tactics get out of control. Most of his games are 5 7, which is very fast, still. One lapse in judgement and the game gets wild and it ends quickly. Other times, Crafty just out-plays him in endgames, but again, not because it is a wonderful endgame player, but because it doesn't make obvious mistakes at that speed. You are buying into the "urban legend" that says computers are far better than GM players at tactics. This "legend" should be more specific and say "some kinds of tactics". Because in some forcing tactical lines, a human can see 5X deeper than the best program around. The human's problem is to keep the game steered into positions where that 5x is important, not into positions where the machine sees everything while the human has to "calculate". > I think is is hard for a reasonable >person to buy the ideal that humans are better even though they lose more often. > It makes no sense to say the A is better then B, even though A loses 70% of the >time. There are circumstances that make this perfectly normal. If all you want to look at are "results" then for the moment, the machines look very strong. But if you look at the _games_ you will certainly see lots of holes that need patching. When I first moved to Birmingham (first time) in 1970, I played a guy that was a regular in the local chess club. At the time, I was a 1600 USCF player, and he was a 2000 USCF player. He beat me most of the time, but he played some ugly positional chess only to beat me by out-calculating me. But over the next three months, my calculation got better, his positional play didn't, and I ended up beating him three of every four games we played. I just had to give up my favorite openings (say the Goring gambit, King's gambit, and even the Evan's gambit as white, the Latvian or others as black). I have _always_ liked to "slug it out tactically" as the chess is more fun. But I discovered that against _this_ player that was the wrong plan. I adjusted, and I did better. That is where computer chess is today...
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.