Author: Albert Silver
Date: 10:17:14 01/04/00
Go up one level in this thread
On January 04, 2000 at 00:36:41, Charles Unruh wrote:
>On January 03, 2000 at 21:50:02, Albert Silver wrote:
>
>>On January 03, 2000 at 18:49:53, Charles Unruh wrote:
>>
>>> rt Silver
>>>>>
>>>>>Hi Albert,
>>>>>
>>>>>I am OK with you disagreeing with me - but please allow me to put forward some
>>>>>evidence to support my case.
>>>>
>>>>If you didn't, what would I possibly have to disagree with? :-)
>>>>
>>>>>
>>>>>Firstly, from where you've made your interjection, I assume that the point that
>>>>>you disagree with is that computers would be expected to score about 80 Elo
>>>>>points higher at active time controls than at tournament time controls. Please
>>>>>correct me if I'm wrong.
>>>>>
>>>>>Now, Selective Search magazine has been published since 1985 (the web site is
>>>>>http://www.elhchess.demon.co.uk/ ).
>>>>>
>>>>>I can't remember when, but in the past, an in-depth study of how different clock
>>>>>settings affect the expected rating of a computer was published. In the current
>>>>>issue (Dec '99 - Jan '00), as they do in every year end issue, they have
>>>>>published summary tables of expectations how different time controls affect the
>>>>>expected outcome. As everyone knows, computers will do relatively better at
>>>>>faster time controls than human players will. Briefly, the results are as
>>>>>follows:
>>>>>
>>>>>Tournament Chess: Normal
>>>>>Active Chess (G/30): +80 Elo
>>>>>Blitz Chess (G/5 or G/10): +200 Elo
>>>>>
>>>>>Selective Search was originally set up with the specific aim in mind of
>>>>>providing more accurate information to chess computer consumers about the
>>>>>strength of the machines, so they take the issue very seriously. At the present
>>>>>time, their highest rating for a chess computer is 2620
>>>>
>>>>I'm sure they are very serious, but that doesn't make them right. I have never
>>>>had the opportunity to read their publication, but as I understand it, their
>>>>ratings include the SSDF ratings. As I don't think the SSDF ratings have any
>>>>value towards deciding the relative strength of computers against humans, the
>>>>very basis of their ratings becomes valueless in my opinion. Any calculations
>>>>made from them would be equally futile, with all due respect. You want my gut
>>>>feeling? On a 500 Mhz PC, the programs in 40/2 are playing a little over 2500.
>>>
>>>A little over 2500 woul be what most all of us here would consider grandmaster
>>>strength.
>>
>>I understand. It does seem ambiguous of me to say this. Essentially, time will
>>tell, but I won't say PC programs are of grandmaster strength until they can
>>hold their own against grandmasters.
>>
>> Albert Silver
>
>Ambiguous isn't quite the word i think, sort of non sequitor or bullheaded seems
>to fit better "2500 yes, but I won't say PC programs are of grandmaster strength
>until they can hold their own against grandmasters."(paraphrase). 2500 on
>500Mhz at that 2500 is holding their own(what about on a Gigahertsz machine?
Why would doubling the speed make much of a difference in 40/2? It isn't
bullheaded, it's logical. So far I believe it is capable of PERFORMING at 2500
against WEAKER opposition. I am not convinced it will perform the same against
Grandmasters. If it performs 2500 against 2300 players but 2300 against 2500
players, it isn't playing at grandmaster strength.
Albert Silver
>Even if such a machine gave only 5-10 points and comps were at the bare bone
>flat 2500 then they would be higher rated by your reasoning than several if not
>many GMs. In fact what most of us here have been looking for, is not for comps
>to hold their own but actually beat GM's we've seen them "hold their own"(draw)
>more than enough.
>>>
>>>>They are outstandingly consistent in what they do, but conceptual chess will
>>>>still be our greatest weapon against them. Conceptual positionally, and
>>>>conceptual tactically. That's why some combinations, though relatively simple to
>>>>calculate are beyond them for the moment.
>>>>
>>>> Albert Silver
>>>>
>>>>
>>>>>(though they state that
>>>>>up to 60 Elo points could be added if one posseses a 500 Mhz PC).
>>>>>
>>>>>I'll leave it at that for now - but if this isn't good enough, I suppose I'm
>>>>>going to have to rummage through my pile of old issues to look for the original
>>>>>article about how these calculations were made.
>>>>>
>>>>>-g
>>>>>
>>>>>>> , then we must be saying
>>>>>>>that right now the computers are about 2620 Elo - which isn't bad (if a little
>>>>>>>lower than the 2664 - 2674 that SSDF seem to be saying Tiger can achieve).
>>>>>>>
>>>>>>>-g
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.