Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question to Uri Blass

Author: Mike Hood

Date: 14:47:24 03/27/02

Go up one level in this thread


On March 27, 2002 at 15:44:42, William H Rogers wrote:

>On March 27, 2002 at 15:35:47, Slater Wold wrote:
>
>>On March 27, 2002 at 15:09:05, ALI MIRAFZALI wrote:
>>
>>My responses, based on the hundreds of hours I've spent reading about this
>>project.
>>
>>>As i have read in theses messages;You believe that Current programs
>>>are stronger than the 1997 Deeper Blue.Since that was a dedicated machine not a
>>>program that can be run on different Computers I have the following questions.
>>>           1.What Elo would the 1997 version of Deep Blue get on the SSDF
>>>(this is a hypothetical question;that is if SSDF tested Deep Blue.)
>>
>>2890.
>>
>>>           2.Same question for the 1996 version.
>>
>>2675
>>
>>>           3.How important is pruning? Why or why not do the current programs
>>>have better pruning than Deep Blue?
>>
>>Deep Blue didn't prune.  Brute force, no null move.  Hsu looked into null move,
>>however he felt against Kasparov, that it might be a risk.  Therefore he didn't
>>use it.  On a 1.0Ghz machine, that extra ply from pruning is a make or break
>>thing.  At 200M nps, you've got some nodes to spare.
>>
>>>           4.If current programs (shredder 2715 Elo on 1200 Mhz For example)
>>>are indeed sronger than Deep Blue 1997;What specifically makes them stronger?
>>
>>They aren't.  They won't be for a long time.
>
>That is a moote question or answer as we do not know just how strong Deep Blue
>really was. If during the game with Kasparov, he was not at his best, and if
>they had played more games, say 5 or 10, then we might have a much closer
>guestimate as to the playing strength, but for now we do not. It is also known
>that we have programs now with much better chess knowledge than they had when
>Deep Blue was written. Remember that DB relied mostly on its tremendous speed
>not on a great amount of chess knowledge. It is quite possible that Shredder or
>some of the other programs written today would fair much better than the last
>version of DB, but we will never know as IBM, once the got the title, did not
>want to risk losing it.

If IBM ever has the guts to resurrect Deep Blue, let's hope they play it against
a selection of the best current chess programs for PCs and (money allowing)
several top GMs, not just Kasparov. IMO Anand is a better anti-computer player
than Kasparov. I hope that Vincent Diepeveen could also play against Deep Blue.
That will finally give us a half-way decent rating for Deep Blue, instead of the
bloated Elo figure sometimes thrown about.




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.