Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: next deep blue

Author: Robert Hyatt

Date: 07:22:27 01/23/00

Go up one level in this thread


On January 23, 2000 at 06:14:19, Alvaro Polo wrote:

>On January 23, 2000 at 02:51:55, Amir Ban wrote:
>
>>On January 23, 2000 at 01:19:04, Bruce Moreland wrote:
>>
>>>On January 22, 2000 at 17:47:28, Robert Hyatt wrote:
>>>
>>>>They were searching about 100K nodes per second, according to Hsu.  yes, they
>>>>do a lot more in their eval than what they could do if they ran on a PC and
>>>>searched 100K nodes per second (this match was in 1995-96 time frame, which
>>>>would have been roughly pentium pro 200 level machines for the PC side).  But
>>>>regardless... if they searched only 100K, and they dominated the commercial
>>>>programs as they have been reporting (again, 38-2 was reported) that says that
>>>>whatever they are doing is pretty good...  as programs like fritz are way over
>>>>100K on a P6/200...
>>>
>>>I think that those matches should be absolutely ignored because they weren't
>>>played under controlled conditions, and they weren't played in public.
>>>
>>>There are enough places that mistakes could have been made, and enough estimates
>>>that could be off, that for these data to be accepted upon faith is completely
>>>ridiculous.
>>>
>>>The machine has existed for some time, and has had ample opportunity to display
>>>its talents on the internet and in international tournaments.  However, we have
>>>zero published games between a "full strength" DB Jr and any other program.
>>>
>>>For all I know it is a wonderful machine.  However, if anyone wants me to accept
>>>that it is a wonderful machine, it is going to have to come out and play some
>>>games, the idea that it should play all of its computer vs computer games behind
>>>locked doors and that we should simply accept the results is just bullshit.
>>>
>>>Sorry.
>>>
>>>bruce
>>
>>The results can be disregarded on these grounds of course, but it's also true
>>that the results, as reported, can be dismissed as being in contradiction to the
>>DB/DT public record, and to common sense in general.
>>
>>Amir
>
>I would like to make a question. Let's suppose that DBjr really beat the other
>programs 38-2. Lets suppose that that means a supperiority of 400 elo. My
>question is, do you believe this 400 elo would be maintained against humans? I
>mean, perhaps comp-comp differences become exagerated, and 38-2 against
>commercials wouldn't mean over 100-200 elo against humans. Do you believe it
>would be possible?
>
>Alvaro


I would guess that "yes, this is possible _and_ probable."  He mentioned (as
did Murray) that the programs had severe king-safety problems vs the one-chip
deep blue machine.  And got tactically blown out reasonably quickly.  I don't
think humans (at least IM/GM players) would get into the same sort of trouble,
since they have a much better king safety 'evaluation'...

comp vs comp and comp vs human are two different worlds...



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.