Author: odell hall
Date: 04:35:00 07/24/01
Go up one level in this thread
On July 23, 2001 at 14:52:19, Christopher R. Dorr wrote: >On July 22, 2001 at 15:59:59, Uri Blass wrote: > >>On July 22, 2001 at 10:44:24, Robert Hyatt wrote: >> >>>On July 22, 2001 at 08:57:56, Drazen Marovic wrote: >>> >>>>Not Fritz 7!, Not Deep Fritz!, but Deep Fritz Grandmaster 1.0!! This should be >>>>the name of the next version of Fritz. Just to take a slap at all of the non >>>>believers in fritz's grandmaster strength. Mogens you said you didn't believe >>>>in psychics, but you would believe in amazing luck time and time again, isn't it >>>>amazing how yet again a non GM strength program could hold a 2600+ GM to an even >>>>score- just luck i'm sure..... >>> >>> >>>Did you ever stop to think that a 2600 player _could_ hold a 2400 player >>>to a drawn match result just as easily? If a GM wants to draw, it is _very_ >>>difficult to prevent it. >> >>Yes but a GM is not going to play for a draw against 2400 player in a 6 game >>match. >>He is going to play for a win because he knows that even after losing one game >>he has very good chances to get at least +2 =3 in the other games. > >This is not necessarily true. History has shown us that white has a 52% to 54% >edge against black's 46% to 48%. While certainly not decisive, it would make >sense in many match situations for a 2600 GM as black to head for a draw against >the 2400 GM with white. It depends on many things; the GM, the match length, the >score, the prizes. Especially against a strong computer this strategy would make >sense because the computer's enormous book would give it an automatic edge in >terms of finding a line that was advantageous to it out of the opening. If you >can head for a relatively safe position, logic would dictate that you do so, as >many of the sharp lines for black have a high loss rate. One loss in a six game >match, and the GM *has* to push hard for a win. Against a computer, this is >deadly, as one of their great strengths is to take any offered material (often >necessary to offer in an absolute-win game), hold it, and win in 50 moves. Also, >since even very good computers *still* do some amazingly stupid things So what's your Point? Are you saying Humans never do amazingly stupid things? >sometimes, it is a very reasonable strategy to take no risks, and wait for the >box to do something silly, and then crush it like a bug. > > >Chris > > >> >>> >>>I explained the problem a month+ ago... to win a match, the GM has to first >>>not lose it, which means he will play carefully. And wait for the machine to >>>make a horrible mistake that gives him an easy way to a technical win. >>> >>>If the match were set up so that the GM gets $100K for each win, nothing for a >>>draw or loss, then the match would be different. And it is very likely that the >>>GM would play far differently since only a win makes him any money. >> >> >>It is very likely that Deep Fritz could win 4-2 with no draws or 3-2 with one >>draw in this case and people are going to complain that the GM could do better >>result of 3-3 at normal conditions. >> >>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.