Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Another Look at Deep Blue Prototype Vs Fritz

Author: Robert Hyatt

Date: 06:00:44 01/18/02

Go up one level in this thread


On January 18, 2002 at 03:06:00, Joshua Lee wrote:

>
>This is mentioned in the Monty Newborn book Kasparov Versus Deep Blue
>but there is no mention of the problem existing on move 16. It is impossible to
>even guess How Deep Thought or DB would do against any Top program based on one
>game this i realize then so is there a game which you think better to look at
>for a good example of the Deep's definately being better than commercials?
>
>I had meant to ask you a few questions reguarding the Deep's  In your opinion
>what did they do better than current software and Reguarding a Performance list
>of Computers. This list shows the performance Ratings of Deep Blue/Thought, the
>ratings are low

Low?  Deep Thought produced a real USCF rating of 2650 over 25 consecutive
games to win the stage III Fredkin prize.  That is not very low.  :)

> but the game number is high compared with commercial software
>Chess Tiger having 2788 against an average FIDE elo of 2497. Doesn't it make
>sense that this is just too high and with the same number of games as for
>instance Deep Thought of 1989 (29 games) that this rating would come down?
>There is a problem with Chris Carson's list Fast games are included for deep
>thought which most likely messes up the ratings.


I don't give a lot of thought to any of those ratings, personally.  Too many
variables.



>
>Since DB and Deep Thought Searched differently if not better and their Search to
>a Depth of 10 would be different from Crafty , Hiarcs , or Fritz how can one
>know that just because i let these programs search the same number of nodes that
>the results would show beyond doubt that one is better than the other?
>Do i need to look twice as deep what are the guidelines?


Totally unknown, unfortunately.  It is hard to compare today's programs that
use null-move to today's programs that don't, much less comparing to a 1989-era
program that didn't...




>
>I can say that Commercial's are 100% better if they could find the Bishop Sac
>that Kasparov Could've played then i can say well if the two were playing the
>Commercial software would win ...and that's about the only decent example i have
>but that sac turns out to be harder than most Nolot Positions due to the Depth
>needed.

I believe that bishop sac would have led to a draw, at least according to
Murray Campbell DB was expecting the sac and had a 0.00 score...  Kasparov
wisely chose to avoid the complications...  and his pawn move made the sacrifice
threat real as after his move the sac actually has "teeth".






>
>Do you think Computers are IM or GM strength? I remember you placing then at
>around 2450 or so within the past few years maybe i am quoting you wrong


My opinion hasn't changed much.  They may _possibly_ touch 2500 now, unless
the opponent plays to the machines significant weaknesses.



>
>I could babble on but then i don't want to annoy.
>
>Thankyou



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.