Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: GM Blatny Vs Ferret(C) Games: questionable time control

Author: Robert Hyatt

Date: 14:06:29 02/20/99

Go up one level in this thread


On February 20, 1999 at 12:34:15, Bruce Moreland wrote:

>
>On February 20, 1999 at 04:25:58, Mark Young wrote:
>
>>I agree, and is why I always thought it unwise to compare and draw conclusions
>>between manual and automated program ratings on ICC, FICS etc. at blitz time
>>controls.
>>
>>Being able to play 5 0 and 3 0 games without loss of time and without losing on
>>time is a big advantage for the automated programs when playing strong humans or
>>other manual programs.
>
>FYI mine doesn't play 3 0 r or 4 0 r, and unlike some others, mine doesn't have
>a limit on how slow it will go for rated blitz games.  It will happily play a 2
>19 as well as a 5 0.
>
>There are advantages and disadvantages that accrue to either side in this issue.
>
>The automatic programs can play lots of games and they never mouse slip, they
>gain a few seconds on the clock every move, they rarely if ever flag, they can
>play shorter games, and they can make instantaneous moves.
>
>They are also subject to repeated games while the operator is asleep, they are
>challenged by people when their rating is up and ignored when their rating is
>down, they don't challenge, instead they are challenged by people of all rating
>levels, people can pick the time control to play against them that suits them
>best (without having to negotiate with the operator), and people can get a lot
>of practice against the automatics and understand their style.
>
>A manual operator can (but doesn't necessarily) receive assistance from the
>operator during the opening phase and occasionally in the middlegame, can pick
>the time controls that suit it best, can change programs in order to exploit
>perceived weaknesses in specific opponents, and can cherry pick high-rated
>opponents (you don't often see manual programs playing 10-game matches with
>2400-rated people, which hurts your rating often).
>
>The manual programs lose 2 or 3 seconds per move if they are fast operators, and
>more if they aren't.
>
>I could argue about which is better but I'd probably be wrong.
>
>bruce


I've said many times that ICC/FICS/chess.net are _really_ interesting places
to play chess.  But when you go 'automatic' you enter into a _hostile_
environment, because your program is totally on its own.  It has to allocate
time with zero input from a human, it has to choose openings with zero help
from humans.  It has to handle repeated games by the same opponent with zero
help from humans.  All in all a very hostile environment.  But once it can
handle these things, it can pretty much handle _any_ environment with little
worry by its author.

IE I was doing 'book learning' pretty much before anyone except Marty (that I
know of).  And the reason was the ICC environment where a human will play the
same opening over and over if given a chance.  I decided to (a) not complain
about it and (b) take action on 'my end' to stop it.

A little 'human guidance' can be important.  But when automatic it is not
possible.




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.