Author: Dirk Frickenschmidt
Date: 16:18:50 12/28/97
Go up one level in this thread
On December 28, 1997 at 08:56:20, Kai Lübke wrote:
(move to bottom)
Hi Kai:
Here's the human theory concerning the game, based on Chessbase Big
Database 98: about 875.000 games, clean database, no doubles or currupt
data seen so far as in many other big databases; so the empirical basis
seems to be quite sound.
I'm of course not shure if there is a rare source for more games
anywhere out which contains more theory on this, but I think this is not
too probable.
You find the theory lines in brackets.
They stop about where Rebel theory stops.
I have no idea what Mchess had in its book, but it's *probably* anything
else than human theory. I guess it's simply a drawn game from
autotesting which shall help to gather Swedish Elo points without
keeping the - in my eyes very attractive - Mchess engine too busy :-)
I already joked that the programmers should use a new type of message
box if this becomes common practice, something which may read like this:
"Dear operator: this is a won game against Hiarcs6" or "this is a draw
against Rebel8", and then the choice: "Would you like to
a) save the game as won/drawn right away?
b) jump to the first move out of book and study the endgame?
c) have the game displayed automatically move by move in a few seconds
witout having to wait for the opponents moves which are known anyway?"
:-)))
Like Thorsten, Don Dailey, Fernando Villegas and others I don't want a
heated debate. I would even prefer if Peter Schreiner, Sandro Necchi and
Marty himself would express their point of view. We certainly can debate
all that in a calm and polite manner with some humor, but without any
hidden or open will to do any harm to Marty's program selling. I had
some very attractive games using Mchess6 - which I bought because it
plays strong chess! - and hope to see more of this kind from those who
bought Mchess7. And we should perhaps post some of these brillant games
played by the Mchess *engine* (not by the book) right here to make it
clear that the question is *not* something pro or anti Marty and his
fine program.
Anyway it must be possible to debate where book learning development
might lead us and what we users expect a program to do and what not.
Maybe this differs from user to user, maybe a majority likes this or
that...
But let's first view the the facts...
Rebel8,P - Mchess7,P [B12]
Kai Lübke Dec 28 97 CCC
1.e4 c6 2.d4 d5 3.e5 c5 4.dxc5 Nc6 5.Bb5 e6 6.Be3 Nge7 7.Nf3 Nf5 8.Bd4
Nfxd4 9.Qxd4 Qa5+ 10.Nc3 Qxb5 11.Nxb5 Nxd4 12.Nbxd4 [12.Nfxd4 Bxc5
13.Nb3 Bb6 14.a4 (14.c4 dxc4 15.Nd6+ Ke7 16.Nxc4 Bd7 17.Nxb6 axb6 18.0-0
Rhc8 19.f4 Bb5 20.Rf3 Rc2 21.Rf2 Rac8 22.Rd1 Be2 23.Rc1 Bb5 24.Rd1 Be2
25.Re1 Bb5 ½-½ Jepson,C-Engqvist,T/SVE-ch 1997/EXT 97) 14...0-0 15.a5
Bd8 16.0-0 a6 17.N5d4 f6 18.f4 Be7 19.exf6 gxf6 20.Rae1 Kf7 21.Re2 Bd6
22.g3 Bd7 23.Nf3 Bb5 24.Rfe1 Bxe2 25.Rxe2 Rfe8 26.Rd2 Rad8
Renner,C-Treffert,P/BL2-S 1994/BL2/0-1 (43)] 12...Bxc5 13.0-0 Rebel out
of book at -0.16 [13.c3 Bd7 14.Kd2 Ke7 ½-½ Westerinen,H-Arkell,S/London
WFW 1988/EXT 97] 13...Bd7 14.Rfd1 0-0 15.Nb3 Be7 16.c4 Bc6 17.Nfd4 Rac8
18.Na5 Bd7 19.cxd5 exd5 20.Rac1 Rxc1 21.Rxc1 Rc8 22.Rxc8+ Bxc8 23.Nab3
a6 24.f4 Bd8 25.Nc5 g6 26.b4 Kf8 27.Kf2 Ke7 28.Ke3 f6 29.Ne2 b6 30.Nd3
Bf5 31.a4 Kf7 Mchess out of book at +0.17. If Rebel had played 31.Nd4
(like it would have up to close to 10') Mchess would even have stayed in
book 3 more moves. =
Kind regards from Dirk
>OK, some more facts about this problem:
>
>I'm currently playing a game at 10 min/move fixed between MChess 7 and
>Rebel 8 on my P6-200, 60 MB hash each.
>Rebel left the book at move 13. MChess left the book at move 31!
>And that happened only because Rebel moved away from 31.Nd4 (which it
>considered from 0:00 on) close to the 10:00 mark; had it played Nd4,
>MChess would have stayed in book for another 3 moves! I.e., in a normal
>tournament game, Rebel would never have reached the point where it moved
>away from Nd4.
>So this _could_ mean it's an anti-Rebel8-line (though the eval out of
>book in the "31.Nd4"-line was only +0,68 by MChess).
>
>here's the game:
>
>[Event "Long time control tournament A"]
>[Site "SHEP'S SERVER"]
>[Round "2"]
>[White "Rebel 8.0"]
>[Black "M-CHESS 7.0"]
>[Score "*"]
>
>1. e4 c6 2. d4 d5 3. e5 c5 4. dxc5 Nc6 5. Bb5 e6 6. Be3 Nge7 7. Nf3 Nf5
>8. Bd4 Nfxd4 9. Qxd4 Qa5+ 10. Nc3 Qxb5 11. Nxb5 Nxd4 12. Nbxd4 Bxc5
>13. O-O {Rebel out of book at -0,16} Bd7 14. Rfd1 O-O 15. Nb3 Be7 16. c4
>Bc6
>17. Nfd4 Rac8 18. Na5 Bd7 19. cxd5 exd5 20. Rac1 Rxc1 21. Rxc1 Rc8
>22. Rxc8+ Bxc8 23. Nab3 a6 24. f4 Bd8 25. Nc5 g6 26. b4 Kf8 27. Kf2 Ke7
>28. Ke3 f6 29. Ne2 b6 30. Nd3 Bf5 31. a4 Kf7 {MChess out of book at
>+0,17}
>
>However, I've recently posted a game Comet A.80 - Fritz 5 to r.g.c.c.
>where
>Comet came out of book at +3,15. I called this a "Fritz-killer line",
>but
>Comet's author Ulrich Tuerke told me the opening came from a book on the
>Slav Gambit and was automatically compiled into the opening book.
>
>So there's always a possibility something is _not_ a killer or anti-XYZ
>line...
>
>---
>Shep
This page took 0.01 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.