Author: KarinsDad
Date: 14:52:47 01/10/99
Go up one level in this thread
On January 10, 1999 at 13:57:19, blass uri wrote: [snip] > >I know that the loser is always losing by doing a mistake but the point is that >kasparov did mistakes that he usually does not do. How often does Kasparov make a mistake that nobody catches? Probably quite often, but of very minor consequence. The reason is there are very few individuals (and programs) that could analyze to the level that they would catch something that Kasparov does not catch over the board. Any minor mistake that Kasparov makes is fair game for a program as powerful as Deep Blue to capitalize on. Even if only in an extremely minor manner. > >Resigning in a draw position is not a mistake that kasparov did in the past. Except that Kasparov changed his tactics based on the fact that he was playing Deep Blue. That worked in the first match, but failed in the second. Presumably, the Deep Blue team learned something from the first match. Humans make mistakes for different reasons than computers. Humans make mistakes because they are tired, low blood sugar, emotional considerations, mis-calculations, and not truly understanding the position. Computers only make mistakes because their algoritms did not truly understand the position and all of it's implications. Kasparov lost because he thought he was lost and could not swindle the computer. If he had been in that same position against another human, his thought processes may have been different and he may have thought that he could have played on. His mistake was not just in not seeing the draw, but rather also in assuming that if he himself could not make a mistake in his analysis and that the computer would not miss the win. He made a typical mistake that humans make against computers. He thought that within a given position, the computer would be infallible in winning. > >Going to a line that he was not ready to go to is not a mistake that humans >usually do. Humans do this all of the time, the instant they get out of their own book (or at the GM level, when their opponent plays a theoretical novelty). If not at move 5, at move 10, or at move 15, or whereever. > >I saw the games and I was not impressed by the level of deeper blue. >I expected them to play better with their hardware. Deep Blue did not have to play brilliantly. It did not have to come up with deep complications or sacrifices. It just had to play solid chess and wait for Kasparov to make a mistake. No matter how poorly you thought Deep Blue played, it played well enough to win the match and I guess that's all that matters. > >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.