Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Anand is right!!

Author: Bruce Moreland

Date: 00:50:51 01/13/00

Go up one level in this thread


On January 12, 2000 at 15:12:21, Rajendran Ramachandran wrote:

>On January 11, 2000 at 21:46:38, Roger wrote:
>
>>Excerpt from the interview:
>>
>>"For instance, if you were to remove the database, you can
>>have a computer ten times faster than it is today. Ten
>>times faster than Deep Blue, easily. If it couldn't
>>consult its opening book, my result would improve
>>immediately. I think most of the top twenty, thirty
>>players could beat Beep Blue if it wasn't allowed to
>>consult an opening database. Or, even the opening
>>database is restricted to a certain size. What happens
>>is, their opening database is almost 400-500 MBs of
>>information. It has access to all the games that are
>>played but we have to remember all that. Or, if I am
>>allowed to have a computer with me, okay, I can't check
>>my thoughts but I can see what was played at any given
>>time. My result would then go up."
>>
>>I think he's wrong
>
>
>
>
>Well...It just could be translation..Or your way of understanding the free
>flowing conversation...
>
>Probably he meant that without database even if you have a computer that is ten
>times faster than Deep Blue, top players could beat the computer.
>
>I think he is right!

It is a two parter.  One is what would happen if the program played without a
book.  The other is what would happen if the book were restricted to a smaller
size.

I think that without a book you are likely to see an increase in inferior moves
right off the top, so as white the computer ends up discarding its advantage,
and as black it ends up getting into trouble quickly.  There is also the issue
of repeatability of lines, it's a lot easier to beat something that plays the
same way every time.

I think he's wrong regarding a smaller book, unless it's super small, like just
a few moves.  With a few thousand moves you reduce the chance of repeated games,
and you give the program a higher liklihood of getting its pieces developed,
leading to a game where the program is less likely to get instantly creamed.

bruce



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.