Author: Dann Corbit
Date: 00:24:35 06/27/00
Go up one level in this thread
On June 27, 2000 at 01:01:48, stuart taylor wrote: >On June 27, 2000 at 00:44:20, Dann Corbit wrote: > >>On June 27, 2000 at 00:32:15, stuart taylor wrote: >> >>> Is it not true that human ratings are lower than computer ratings relative to >>>true standard of play due to the fact that humans make many blunders of the >>>nature that computers do not? >>> That is what I Have always beleived as being one of the reasons why computers >>>do as well as they do. >>> In other words, it could possibly be that 2650 on ssdf = 2550 against >>>well-prepared, top humans or even 2500 which is = 2400 in actual standard of >>>play, discounting tactical and mechanical extras. >>>Is this correct? >> >>What you are observing is that the strength of computers and the strength of >>humans is focused in different areas. >> >>Computers are tactically stronger than humans. >> >>Humans are positionally stronger than computers. >> >>If the human can avoid tactical blunders, they will win. However, suppose the >>chance is just 1% that the human will perform a tactical blunder. By 20 moves, >>the chances of not making a tactical blunder are only 80% and 74% by move 30. >> >>In other words, even in very mistake free chess, some tiny slip is likely to be >>generated. Often, this will result in some small material advantage. >> >>If the material advantage is large enough to offset the superior positional >>understanding of the human player, the computer will win. >> >>We might imagine two basketball teams. One has a bunch of short, quick guys. >>They are good at stealing and running the fast break. The other team has tall, >>slower guys. They are good at rebounding and shot blocking. >> >>Both teams play good basketball, but the styles are very different. >> >>Both teams can achieve the object, which is to win, but they go about it in a >>very different manner. > >What really interests me is, how well would a computer rate if its ratio of >tactical/mechanical prowess vs. positional understanding, would be similar to >that of humans? > This, I think is more interesting for a human to know, than when he doesn't >know how much percentage of his effort is competing against something which is >almost like a maths calculator. (as opposed to applied intelligence) Nobody has figured out how to do that yet. If you figure it out, you will be famous. Well, to a bunch of computer geeks you will be famous anyway.
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.