Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: IPCC 11th Paderborn: Standing after 5 rounds

Author: Vine Smith

Date: 18:01:54 03/01/02

Go up one level in this thread


On March 01, 2002 at 17:20:41, Dann Corbit wrote:

>On March 01, 2002 at 17:12:21, Jorge Pichard wrote:
>
>>On March 01, 2002 at 16:25:08, Kurt Utzinger wrote:
>>
>>>IPCCC 11th
>>>Paderborn, 2002.02.27 - 2002.03.01
>>>                   Score     S F G G B I S Q C D I X P N P M
>>>-------------------------------------------------------------
>>> 1: Shredder      5.0 /  5   X . 1 . 1 . 1 . 1 . 1 . . . . .
>>> 2: Fritz         3.5 /  5   . X 0 1 . = . . 1 . . 1 . . . .
>>> 3: Gandalf X     3.0 /  5   0 1 X . . 1 . . = = . . . . . .
>>> 4: Gromit        3.0 /  5   . 0 . X = . = . . . . . 1 1 . .
>>> 5: Brutus XPa    3.0 /  5   0 . . = X . . . . 1 . . = . 1 .
>>> 6: IsiChess      3.0 /  5   . = 0 . . X = 1 . . . . . . . 1
>>> 7: SOS           2.5 /  5   0 . . = . = X 1 . = . . . . . .
>>> 8: Quark v1.75   2.5 /  5   . . . . . 0 0 X . = 1 . . 1 . .
>>> 9: Comet B40     2.5 /  5   0 0 = . . . . . X . . . 1 . 1 .
>>>10: Diep          2.5 /  5   . . = . 0 . = = . X . . . . . 1
>>>11: Ikarus        2.5 /  5   0 . . . . . . 0 . . X . = 1 1 .
>>>12: XiniX         2.5 /  5   . 0 . . . . . . . . . X 0 = 1 1
>>>13: Patzer        2.0 /  5   . . . 0 = . . . 0 . = 1 X . . .
>>>14: Neurologic    1.5 /  5   . . . 0 . . . 0 . . 0 = . X . 1
>>>15: P.ConNerS     1.0 /  5   . . . . 0 . . . 0 . 0 0 . . X 1
>>>16: Matador       0.0 /  5   . . . . . 0 . . . 0 . 0 . 0 0 X
>>>-------------------------------------------------------------
>>>40 games: +13 =11 -16
>>
>>If I remember correctly a few years ago P.ConNers used to be a la par with the
>>best programs, but nowadays it is standing on the bottom of the list.
>
>I am sure it is still a very good program.
>
>However, there are only 5 games, so it is hard to draw conclusions.
>
>Also, the hardware for P.ConNerS does not seem to have been upgraded to faster
>CPU's, while all of the other programs surely have doubled or tripled their
>speed.  Therefore, I am not at all surprised to see it lose some ground (if,
>indeed, that is what has happened).
>
>Since it isn't SMP, it's pretty amazing that it has done as well as it has.  I
>would like to get a peek at the C code and see what is going on.

P.ConNerS seems to have a poor opening book -- I've looked at two of its
defeats, vs. Comet in Round 1, and Xinix in Round 5, and both feature gross
errors by P.ConNerS upon exiting (apparently) book too early. Comet as White
played a sharp gambit line in the Semi-Slav (1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.Nf3 c6
5.e3 Nbd7 6.Qc2 Bd6 7.g4!?), after which P.ConNers played 7...Bb4!?, a sideline
in which Black tries to liquidate material on c3. Comet's reply of 8.Bd2 was
book, but then P.ConNers erred with 8...Nxg4?, obviously out of book at this
point. This last move seems very wrong because Black has enough problems taking
the g4 pawn on move 7, but here White has been granted an extra tempo to play
O-O-O and connect the rooks quickly. Instead, the usual idea behind 7...Bb4 is
to play ...Bxc3 followed by ...Ne4 and ...Nxc3. The book lines for the program
should have been extended to ensure that this idea was followed through.
Then against Xinix, P.ConNerS played an offbeat Ruy Lopez, which Xinix met with
a less than challenging line (1.e4 e5 2.Nf3 Nc6 3.Bb5 g6 4.O-O [too slow, 4.c3
or d4 are preferred by theory] Bg7 5.c3) at which point P.ConNerS' book seems to
again terminate prematurely, since instead of playing the known 5...a6 with
equality, the program found the suicidal 5...d5?, opening the center while still
two moves away from castling. This mistake suggests that P.ConNerS' code does
not have much by way of opening heuristics (while other unbooked programs might
make the same error in the game against Comet, I don't believe Fritz or Crafty,
for instance, would play 5...d5?), magnifying the problem of the inadequate
book.



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.