Author: Robert Hyatt
Date: 19:58:36 07/21/00
Go up one level in this thread
On July 21, 2000 at 21:14:46, Ralf Elvsén wrote: >Trying to forget the NPS-discussion a second, I think one should >also look at the chess played. I am just a clubplayer and my >opinion is essentially worthless. Nevertheless... > >The chess Junior displayed at Dortmund was IMHO the same thing that happens >on my own PC but one or two (or three?) plies deeper. When I look >at the DB - Kasparov match 97 I think I see something different, >another dimension (yeah, I can't defend this point). > >I think it is possible that DBs potential performance against humans >wouldn't be extremely much higher than DJs. But DB could still possibly >crush DJ. These two scenarios aren't incompatible. Isn't it the >old discussion that rating differences are exaggerated in >comp - comp games? > >It would be interesting to hear what strong chess players think >(especially Kasparov :) > >my zero cents > >Ralf One point Hsu made when he told me about the original 10 game match with the 200K nps DB, was that it won ten games non-stop due to one thing: a lack of king safety understanding. He played 5 games vs Genius and 5 vs the rebel version that was current back then. And he was quite specific about the problems they had defending against an attack. The problem back then was that computers didn't attack very well (they still don't). And they were (and still are for the most part) horrible at detecting attacks until it was too late. I think in normal games, 10 0 would be nearly impossible to do. But if your opponent cheerfully allows you to attack game after game, and sits there happily while you do, then 10 0 is quite easy to understand. I would shudder at the thought of a player like cptnbluebear, or insight, or several other very strong humans on ICC, if they attack like they attack, but if they had the tactical acuity of a computer to prevent the occasional blunder that loses won games. Hsu said he was surprised at first, but once he saw what was going on, it was obvious the micros were simply missing some important evaluation that he was doing after working with Joel and others. Just some more data to consider...
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.