Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Brute force base of good game. True or not true?

Author: Robert Hyatt

Date: 06:57:44 09/08/99

Go up one level in this thread


On September 08, 1999 at 09:43:18, Claudio A. Amorim wrote:

>On September 07, 1999 at 10:41:04, Robert Hyatt wrote:
>
>>On September 07, 1999 at 09:36:03, Claudio A. Amorim wrote:
>>
>>>On September 07, 1999 at 06:40:02, leonid wrote:
>>>
>>>>Brute force search is the base of good chess game. True or not true?
>>>
>>>Hi, Leonid,
>>>
>>>Deep Blue team think so, but I think they're far to prove it. In fact, Deep
>>>Blue's game has not a distinctive character, except being extremly efficient. on
>>>handling highly unbalanced positions. The top microcomputer programs (like Rebel
>>>and Hiarcs), albeit much slower, display much more chess knowlegde than Deep
>>>Blue. They surely handle positional play much better.
>>>Brute force, in computer chess, is one of the means to achieve good results, but
>>>must be mixed with other techniques, to deliver great chess.
>>>Imagine Deep Blue without an openings book or a tablebase... It would play as
>>>badly as any club player.
>>
>>
>>Your last sentence it complete hogwash.
>Do you allways use this kind of vocabulary to disagree with, or correct someone?
>I'm not a native speaker, as you can see, but "hogwash" is surely a silly word
>to express "plain wrong". You bring me a new point to think about, but there's
>no need to be unkind, ok?

And you don't think _you_ were being "unkind" to the DB team?  "Imagine Deep
Blue without an openings book or a tablebase...  It would play as badly as any
club player."

You have no idea what you are talking about.  And if you don't, making such
statements can be expected to draw negative attention.  If you watch what _you_
say, you don't have to worry about how _others_ will respond.



>
>DB is just like any other program
>>that uses tablebases...  they are only used in a relatively small percentage
>>of the games, in only a small percentage of the moves in those games where they
>>are used.  In a couple of games, Kasparov took DB out of book on move 2.  It
>>appears to me that it played just fine, judging by the match result against
>>him.
>>
>>The rest of your post has little technical merit, as I know of nothing that
>>suggests that other programs "display much more chess knowledge".
>
>Just see the games. If you like chess, yoy'll agree. Anyway, I never said brute
>force isn't important. I just say it's only a technique, among others, that make
>a program really nice.

I did "see the games".  DB won more than it lost.  That isn't a "club player".




>
>>>When was the last time _another_ program beat Kasparov at 40/2 in a single >>game, much less in a match?
>
>When was the last time Kasparov played so badly and frightened as he did in that
>match? The last game, for instance (I mean the Caro-Kan), was a shame. Crafty,
>Hiarcs, Rebel, Fritz and Chessmaster, all would have won, after Kasparov's
>blunder in the opening. In other games, Rebel 9 suggests interesting
>improvements for both sides, even in tournment time analysis. And we have also
>the clear draw Kasparov blundered away (Qe3+, if I remember well), when he
>resigned that Ruy Lopez...

Again, your data is _wrong_.  Several have tested this, and against top
commercial programs, a strong human (an IM in one case) was able to win with
black or draw _every_ game.  IE white, played by a couple of commercial
programs could _not_ win.

This was discussed here at length.  So your hypothesis has quite a large
hole right in the middle of it.



>>
>>You can't play _badly_ and pull that off.
>
>Deep Blue's play in efficient, but aesthetically unpleasant. As I said, it has
>no distinctive character.


Who cares?  The object is to win games, not be aesthetically pleasant.  Who
knows but what the best games one day might be totally ugly by today's
standards?



>>>
>Sorry for the "hogwash" talk. Since I am new on the field, I have still a lot to
>learn.
>
>Regards,
>
>Cláudio.



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.