Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Behind Deep Blue: 3rd print with new Hsu afterword

Author: Vincent Diepeveen

Date: 06:42:09 05/10/04

Go up one level in this thread


On May 10, 2004 at 08:11:34, Sune Fischer wrote:

>On May 10, 2004 at 07:33:38, martin fierz wrote:
>
>>there are instances where pushing pawns in front of your king is ok. there are
>>others where it's not ok. they are rather easy to distinguish most of the time
>>(some simple rules are sufficient for this) and i'm surprised that you don't
>>understand that it's not such a great idea here...
>
>Rules of thumb are good but not great, there are always exceptions and computers
>are experts at finding them. It is possible that all other moves simply lose
>faster.
>
>>i know you are a born
>>skeptic, but perhaps you should learn to trust people with more experience
>>sometimes :-)
>
>Sometimes, but maybe not this time ;)
>Just look at the thread below
>http://www.talkchess.com/forums/1/message.html?364474
>
>I don't 100% trust analysis made without computer assistance, even when it comes
>from GM's. It is only too human to err.
>
>>i'm not going as far as vincent of course to say that "after ...g5 white wins
>>with eyes shut". that is not true of course. but it's obvious that things aren't
>>progressing normally here if you have to play ...g5.
>
>I think the original intent was to show that DB was not at the level of "modern"
>programs _because_ it played g5. Now we have other programs playing it too so
>this bit of "proof" goes out the window, IMO.

However it played like 7 very poor moves in game 1 and that for every game
except game 6 and game 2 it just made a few (shuffle moves are not good but are
not losing the game sometimes).

Antonnie just quoted here what was wrong in his eyes, but there is a shitload of
other very poor moves.

Let me just first tell you that deep blue was very soon out of book.

Starting with move 3 where it already makes a mistake :

3 .. Nd7? best is bxf3 diep plays bxf3 and not nd7.
10.. h6?  very poor move. lucky diep doesn't play that. 10..e5 looks better
          but important is to avoid h6
11.. Qa5? very poor move, better is for example Qe7
12.. Bc7? very poor move. A real NULLMOVE
13.. g5?  very poor move now there is no way to escape from a zero against
          even poor shuffle players
16.. Qb6  this is not the best move
22.. g4?

I hope you realize Seirawan had an easy job commenting on these games. The bad
moves made are *so* poor, that he has hardly time to give better alternatives
for choice mistakes like 16..Qb6.

I am sure you do not realize it, but if you play 2 moves like Qb6 against
kasparov in a normal game you're dead.

Deep Blue plays however 6 completely losing moves in this game so all the poor
moves we do not even discuss here. We just discuss at the utmost beginners level
here. Not even at the higher levels where making wrong choices loses already.

Best regards,
Vincent



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.