Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Estimates of Year Computer Will Defeat World Human Champion

Author: Chris Carson

Date: 07:12:25 04/23/02

Go up one level in this thread


On April 23, 2002 at 09:51:42, Roy Eassa wrote:

>On April 23, 2002 at 00:16:28, Robert Hyatt wrote:
>
>>On April 22, 2002 at 14:58:11, Russell Reagan wrote:
>>
>>>On April 22, 2002 at 13:57:16, William H Rogers wrote:
>>>
>>>>I think that the new worlds champion will be announced just as soon as the top
>>>>human plays a round robin game against the top computer chess programs. I think
>>>>that there should be at least 4 to 6 games with each opponent. When this is
>>>>done, then I think people will see a computer holding the worlds championship
>>>>for the first time( not counting Big Blue). As to when this will happen, it is
>>>>up to the humans, ie. today, tommorow, or 10 years from now.
>>>>Bill
>>>
>>>Part of the reason humans (the _best_ humans) lose to programs is because they
>>>don't have experience playing against them. Computers are still very strong, and
>>>by no means will they ever again be "easy" wins for top players, but if GM's
>>>studied computer play like they would normally prepare for a match with a human,
>>>they would fair better. How much better is hard to say. Maybe they would win 5-3
>>>or 5.5-2.5 in an 8 game match. I don't think a blowout would be common, except
>>>perhaps in the case of Kramnik-Fritz since Kramnik will have had all the time in
>>>the world to prepare and test the program.
>>>
>>>And you mean Deep Blue, not Big Blue :)
>>>
>>>Russell
>>
>>
>>I have used the term "computer shock" for 20+ years.  It still fits.  Computers
>>are not regular tournament participants, which means that meeting one under
>>such conditions is a "shock".  And generally the human does far worse the
>>first time around than the second...
>>
>>This will continue until computers are pervasive in chess.  They aren't,
>>yet...
>
>
>Good way to describe this effect.  Humans do seem to substantially underperform
>(play below their true strengths) against computers at first, leading to
>artifically high initial ratings for the machines.

A really good example of this is the last Rebel GM match where the 2700 GM
(anti-computer expert) played 100 preparation games and scored 2-2 (even) with
Rebel.



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.