Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Stars Only-5a (at tournament time control, List 512)

Author: Uri Blass

Date: 05:17:33 09/02/04

Go up one level in this thread


On September 02, 2004 at 08:06:05, Uri Blass wrote:

>On September 02, 2004 at 08:00:57, Jorge Pichard wrote:
>
>>On September 02, 2004 at 07:43:29, Uri Blass wrote:
>>
>>>On September 01, 2004 at 15:07:06, Jorge Pichard wrote:
>>>
>>>>On September 01, 2004 at 11:40:32, Igor Gorelikov wrote:
>>>>
>>>>>On September 01, 2004 at 09:36:17, Kurt Utzinger wrote:
>>>>>
>>>>>>Hi Igor
>>>>>>A somewhat disappointing score of List_512 with
>>>>>>three (3) won and seven (7) lost matches. Which
>>>>>>was the book used for Fritz Reul's engine. BTW:
>>>>>>great tournamant, please keep up your good work.
>>>>>>Regards
>>>>>>Kurt [http://www.utzingerk.com]
>>>>>
>>>>>List doesn't have its own book, so it plays without book.
>>>>>
>>>>>Igor
>>>>
>>>>
>>>>Any program without an opening book, no matter how strong it is, will play at
>>>>least 40 Elo rating lower. Therefore, please provide List512 an Opening Book.
>>>>
>>>>Jorge
>>>
>>>I disagree.
>>>
>>>If the program can find the better opening moves by search then it can play
>>>better with no opening book.
>>>
>>>It is also possible that a program without opening book can go to positions that
>>>it understand better relative to book positions.
>>>
>>>Uri
>>
>>
>>Can you tell me which program you have in mind that can play better without
>>opening book. I can test your program movei with Dann Opening book, versus movei
>>without an opening book. I will not expect more than 45% out of 40 games for
>>Movei W/O Opening against Movei with Opening. Now is there any other program
>>that you have in mind?
>>
>>Jorge
>
>I have no specific program in mind and I am sure that with no learning programs
>with no book can practically lose every game in a long match after enough games.
>
>Uri
>
>Uri

Note only that in a tournament when programs cannot learn from previous games
things may be different and list did 5/5 with no book against

SlowChess 2.89b
Thinker 4.4h
SlowChess 2.93a
Queen 2.42a
Queen 2.44

for details see http://www.digichess.gr/infiniteloop/nil/nil6_log.txt

I do not claim that list is better with no book but at least in tournaments it
is practically possible that a program can do better with no book if it has
enough opening knowledge to find better moves or better moves for it's style.

In matches it is an harder task and the program need to use some positional
learning or randomness in the opening in order to be better with no book.

Note that book is usually written by humans and it is known that humans are weak
at chess so it is logical to suspect that it may be possible that at least one
program can often find better moves than book moves by itself or at least to
expect it to happen in the future.

Uri



This page took 0.01 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.