Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: I Dicover a new Setting for CM6000, much stronger than CM6666

Author: Jorge Pichard

Date: 06:52:02 04/17/00

Go up one level in this thread


On April 15, 2000 at 23:05:52, Robert Hyatt wrote:

>On April 15, 2000 at 12:18:39, blass uri wrote:
>
>>On April 15, 2000 at 12:00:48, Jorge Pichard wrote:
>>
>>>On April 15, 2000 at 11:30:02, James wrote:
>>>
>>>>On April 15, 2000 at 09:18:28, blass uri wrote:
>>>>
>>>>>On April 15, 2000 at 08:30:42, Jorge Pichard wrote:
>>>>>
>>>>>>So far this has been the best Setting for CM6666, which sometimes even beat
>>>>>>Fritz 6a.
>>>>>>
>>>>>>                   Setting for CM6666
>>>>>>
>>>>>>Style          Positional            Material
>>>>>>
>>>>>>AD   3         M/P  -3               OQ   9.9
>>>>>>SOP  100       COC 103               OQ   9.7
>>>>>>
>>>>>>R    0         MOB 104               OR   5.0
>>>>>>MSD  20        KS  134               OR   5.0
>>>>>>
>>>>>>SS   10        PP  100               OB   3.3
>>>>>>CFD  0.0       PW  134               OB   3.3
>>>>>>
>>>>>>TT = ON                              OK   3.2
>>>>>>DT = ON                              OK   3.2
>>>>>>
>>>>>>                                     OP   1.0
>>>>>>                                     OP   1.0
>>>>>>
>>>>>>--------------------------------------------------------------------------------
>>>>>>By mixing the original Anand Setting with the Material setting from CM6666,  I
>>>>>>finally discovered a much stronger setting, which just won a mini match of 50
>>>>>>games in 15 minutes per side, using an AMD K6 II 500 Mhz with 128 MB of RAM.
>>>>>>When I matched CM6666 setting against my new CM8888, which I just renamed,
>>>>>>CM8888 won
>>>>>>38 out of 50 games.
>>>>>>
>>>>>>              I Should call this setting  the CM-Queen
>>>>>>
>>>>>>Style          Positional            Material
>>>>>>
>>>>>>AD  5          M/P -5                OQ   9.5   (My new Adjustment)
>>>>>>SOP 100        COC 105               OQ   9.4
>>>>>>
>>>>>>R   0          MOB 120               OR   5.0
>>>>>>MSD 20         KS  100               OR   4.9
>>>>>>
>>>>>>SS  8          PP  110               OB   3.2
>>>>>>CFD 1.0        PW  100               OB   3.2
>>>>>>
>>>>>>TT = ON                              OK   3.1
>>>>>>DT = ON                              OK   3.1
>>>>>>
>>>>>>                                     OP   1.0
>>>>>>                                     OP   1.0
>>>>>>
>>>>>>If you already have CM6000, it is very easy to set up a new setting, Just click
>>>>>>under Game and Click the set up a Personality, once you set the new values for
>>>>>>your new Setting Don't forget to save it. If you are having hard time to set the
>>>>>>new setting using your mouse and Bar , simply Click the bar and use your arrows
>>>>>>key.
>>>>>
>>>>>I think that using different values for your queen(rook) and your opponent's
>>>>>queen(rook) is a bad idea.
>>>>>
>>>>>Maybe it works against most programs because they have no idea how to play when
>>>>>there are queens in the board(I see that programs without the name chessmaster
>>>>>are often bean counters in queens endgames and do not understand the importance
>>>>>of passed pawns in these endgames) but I do not believe that it is a good idea
>>>>>against chessmaster personalities.
>>>>>
>>>>>It is not going to be a good idea against other programs when programs get
>>>>>better so I suggest to stop suggesting not equal numbers for material.
>>>>>
>>>>>Uri
>>>
>>>I also own Fritz 6a and Junior 6a for more than a month, plus I own an AMD
>>>ATHLON 800 Mhz, which I am planning to pit against my new setting of CM8888. I
>>>do understand that traditionally the queen and the rook has a value of 9 and 5
>>>points, but I  do not see the reason for saying that changing these values will
>>>make a program play worst chess.
>>
>>I only said that using different value for white queen and black queen is not a
>>good idea.
>>
>>I did not say that changing the queen value from 9 to another number is a bad
>>idea but only that
>>the values 9.9 9.7 are not logical and something like 9.8 for both queens seems
>>more logical(it may help to beat other programs but I do not believe that it is
>>a good idea against other chess master personalities or for the future when
>>programs get better)
>>
>>Uri
>
>I don't like 9.8 or any such number.  I have watched several programs play
>crafty and get into a 3 piece vs queen where they have the queen and think
>they are a pawn up.  Generally they are lost.

You are definitively wright about this high queen value Mr Hyatt, I Just found
out that CM8888 lost to CM6666 in game in 60 minutes per side where CM6666
setting had two Bishops a knight and a pawn, Versus CM8888 setting which had a
Queen and a pawn and at that time CM8888 thought that it was winning by a small
fraction of a point. My solution was to change the value of the queen to 9.5,
since the bishop is already set up to 3.2 and knight to 3.1. I started the same
game before the critical exchange took place and observed the game carefully,
sure enough the small evaluation problem was corrected, and the same game was
won by CM8888, (or should I rename it CM-Queen Setting.)

PS: The score at this point is CM8888 W3 L1 D1 at game in 60 minutes per side,
CM6666 is loosing game number 6 so far, by having a bishop and two pawns Vs
CM8888 a rook and 1 pawn and the bishop will soon have to be exchange to avoid
the pawn from promoting into a queen next move. Ah Ha , I just glanced at the
game and the exchange took place, leaving CM6666 with a lost position.



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.