Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty 15.18 Vs Fritz 5 Game 2

Author: blass uri

Date: 21:36:39 07/29/98

Go up one level in this thread



On July 29, 1998 at 13:18:14, Robert Hyatt wrote:

>On July 29, 1998 at 11:39:08, Mark Young wrote:
>
>>On July 29, 1998 at 07:29:30, Robert Hyatt wrote:
>>
>>>On July 29, 1998 at 06:21:58, Robert Hyatt wrote:
>>>
>>>>On July 29, 1998 at 00:43:48, Didzis Cirulis wrote:
>>>>
>>>>>Hi,
>>>>>
>>>>>When will the next game take place?
>>>>>
>>>>>Regards,
>>>>>
>>>>>Didzis Cirulis
>>>>
>>>>Mark is playing one per day.  Both have lasted basically all day so I assume
>>>>he will kick off another one this morning, but am not sure.  It takes a *lot*
>>>>of time for a human operator...
>>>
>>>
>>>I've puzzled over game 2 a bit, because coming "out of book" a pawn down
>>>really made no sense...  When we started this match, we started off with
>>>the idea of level 0 120 180 (2 hours on the clock, 180 seconds added after
>>>each move.)  I wanted to run Crafty in "tournament mode" and with "book
>>>random 0" (which says search the book moves and ignore any that lose
>>>material or have bad scores).  Since I couldn't make this the default and
>>>leave it logged on, I chose to set those inside crafty when it noticed
>>>"level 0 120 180".  Of course, after game one, Mark and I agreed that the
>>>time control was *way* too long and reduced it to 30 180...  But yours
>>>truly forgot to change the code that sets tournament mode and book random 0.
>>>
>>>Which at least explains how Crafty could leave book at -1.5...  I found
>>>it this morning after game 3 had started..  and have corrected the "patch"
>>>so that crafty should generally not fall into a -1.5 hole.  I can't do this
>>>for "normal" server games because the search is deterministic and would play
>>>the same move every time, which is sure death.  :)
>>>
>>>(against humans)
>>
>>I understand well that a good chuck of a programs strength is getting all the
>>little things right. I have a big advantage here as Fritz is well tuned. Crafty
>>on the other hand is a work in progress. If you want to stop until Crafty is
>>more ready I have no problem with that. When can restart the match at 0-0 at
>>another time. I think it is only right for me to offer this as I have no stake
>>in the out come of this, and the reason for me playing was to evaluate Fritz and
>>Crafty at there best. If you want to restart the match just let me know.
>>
>>I think everyone here knows that you and crafty are much better then a 0-3
>>result.
>>
>>If you want to finish out the 6 games and play a rematch, if you wish. We can do
>>this also.
>
>
>No... keep going as long as you want...
>
>here's analysis so far:
>
>game 1.  good book line that both programs thought white was winning.  Both
>were wrong.  that happens.

In game 1 crafty lost not because of the opening
It could play 54.Qc6 and do a draw against fritz5
(fritz5 against itself in this position is a draw after 54... h5 55.Ka3 Qe6
56.Qc5
Qf5 57.Qc6 Qe6 58.Qc5 Qf5 59.Qc6).

Uri

>
>game 2.  bad book line, good attack by fritz, crafty defended and sort of
>equalized although it ended up with the passed pawn problem again and lost.
>That was simply poor tuning on my part and accurately reflects Crafty at
>present.
>
>Game 3.  Interesting from what little I saw.  Both sides attacked but Crafty's
>came up short.  It underestimated the black pawns and got into trouble.  It
>played not so great, and also probably got out-searched, which happens.
>
>So there's no big problem with the match so far.  I've never said Fritz was
>not strong, if you have followed the comments I've made.  Crafty's "speculative"
>style is not the best way to play against a very fast program, because it tends
>to get into trouble.  That's life.
>
>Carry on... :)
>
>Hopefully, as the week wears on, I will have less stuff to do and can let it use
>the whole machine more, although I don't think that will make a huge difference
>, at least in the games so far...
>
>Bob



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.