Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: At what time control could a commercial program be World Champion?

Author: Robert Hyatt

Date: 17:02:10 03/24/99

Go up one level in this thread


On March 24, 1999 at 18:24:40, Todd Durham wrote:

>On March 24, 1999 at 08:23:16, Albert Silver wrote:
>
>>On March 24, 1999 at 01:15:13, Peter Kappler wrote:
>>
>>>
>>>Most of us agree that the top human players are still quite a bit stronger than
>>>any commercial chess program at regular tournament time controls (40/2).
>>>
>>>It is also not debatable that as the time control gets faster, the relative
>>>strength of the computers increases.  And this leads to my question:
>>>
>>>At what speed are the computers clearly stronger?  How fast would the game have
>>>to be for you to put your money on a top-micro instead of a top GM like Kasparov
>>>or Anand?
>>>
>>>Here are my opinions:
>>>
>>>40/2  super-GM wins pretty easily.
>>>G/60  super-GM, but the games are interesting.
>>>G/15  super-GM, but now the games are very well-fought.
>>>G/5   super-GM, but I would be nervous about my money...
>>>G/3   computer
>>
>>Hard to say. Depends on the length of the match and of course the player's form.
>>In 5-6 games, I'd give the micro a clear nod anyhow, but in 20 games I think the
>>Super GM would come out on top. Of course, it also depends on the micro. Throw
>>the thing on an Digital Alpha and I'll bet all my chips on the machine. The
>>biggest problem with g/3 and g/5 especially, is the time lost making the move. A
>>player like Hawkeye (IM Schmaltz) or Dlugy (GM Dlugy) from ICC are monstrous g/1
>>players (you have to see them to believe it. No pieces hanging and they actually
>>play chess) but they can NEVER compete against a program because every move they
>>make will lose time and is therefore a death warrant. Curiously I don't think
>>that in g/1 comps play much better than them. They don't hang pieces as well,
>>but they don't see very far and here the GM's positional instincts might become
>>a factor.
>>
>>                                     Albert Silver
>
>I've watched Dlugy play 1 0 and you're right, it's astounding! So what about
>Dlugy (or other strong human) v (random strong computer program) at 1 1 so that
>the human doesn't perforce lose on time?
>
>Todd Durham


Here is a bit of data.  Dlugy plays lots of 3 0 and 5 0 vs crafty, but he has
recently tried some 5 2 games:


Crafty has won 6, lost 2 and drawn 1.  Most were 5 2, with a couple of 5 3
games to boot.  I don't find the 5 0 vs 5 3 scores that different.  Players
like Dlugy aren't losing on time very often.  They get into tactical trouble.




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.