Author: Christophe Theron
Date: 15:06:59 10/18/00
Go up one level in this thread
On October 18, 2000 at 17:04:44, Eelco de Groot wrote:
>>If it is dead even after Rc6, then this move is highly justified.
>>
>>It offers opportunities to white, and as it is dead even white is not taking
>>risks himself.
>>
>>Instead, black can reach a position where it takes a too deep search in order to
>>find the refutation, and lose.
>>
>>That's probably what happened (I'm not even pretending that Rc6 was correct),
>>and white won.
>
>Hi Christophe,
>
>I don't know if you meant that as an exact statement, but I would think that,
>what I believe Amir also said, is that the opportunities come at a cost. Okay,
>here speaks the beancounter, you forego on playing the "most correct" move so a
>perfect evaluation would see the resulting position go worse. However you create
>practical opportunities in return. I think the opportunities will often be
>present for both sides because the position is made sharper. Like Spasski said
>about the Kings Gambit, that in that opening maybe more than in any other, the
>moves come at an incredible cost. Meaning there I think that both sides have to
>be extremely careful not to make a mistake. Just as in this position, where in
>Thorsten's variation Amir gave a possible 45... b5 46. Qe2 Rd5 47. Bb6! as
>giving more trouble for Black but which may be hard to find for any program or
>person playing the White pieces. So it works both ways, as you no doubt will
>have seen much more examples of in testing. If the position was dead even after
>Rc6, it must have been in theory a little bit better for White just before. If
>the attack fails, the opportunities and the initiative will surely go to the
>defending side. A principle not unknown to footballers playing Cattenaccio I
>believe? After Rc6 it just depends on whether Black or White can play without a
>fatal mistake, much more than the exact evaluation. And about what Johan (not de
>Koning) always wanted to see in soccer with Ajax and FC Barcelona, trying to
>keep the iniative, the attack! But saying that White can play without risk
>because the position is even, that sounds more like the old paradigm again.
Of course I know that. But if I follow this principle I come back to what chess
programs do currently: play boring.
I have two good reasons to go for this kind of positions:
1) my program is no patzer tactically. I have worked for years on improving my
search algorithms and as a result my program is often able to compute deeper
than its competitors. So if there is something (a combination) to find in the
continuation, my program has maybe an advantage.
2) if the win is to deep to compute, both opponents are left in the fog. In this
case I believe that the defending side has bigger problems to solve than the
attacking side and is more likely to crack under the pressure. The game against
Nimzo demonstrates my point. Of course it can backfire sometimes, and with lots
of games we will be able to say if it works more often than it fails.
>BTW Christophe, I have a difficult(?) test position for Tiger, well, more for
>Q5t really.. It's from a game Kasparov - Timman, Wijk aan Zee this year (TWIC
>273).
>
>
>[Event "Corus"]
>[Site "Wijk aan Zee NED"]
>[Date "2000.01.25"]
>[Round "9"]
>[White "Kasparov,G"]
>[Black "Timman,J"]
>[Result "1-0"]
>
>1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.Nxd4 Nf6 5.Nxc6 bxc6
>6.e5 Qe7 7.Qe2 Nd5 8.c4 Nb6 9.Nc3 Qe6 10.Qe4 Bb4
>11.Bd2 Ba6 12.b3 Bxc3 13.Bxc3 d5 14.Qh4 dxc4 15.Be2 Nd5
>16.Bxc4 g5 17.Qd4 Bxc4 18.Qxc4 Nf4 19.Qxe6+ Nxe6 20.O-O-O Ke7
>21.Rhe1 Rhd8 22.Rxd8 Rxd8 23.Re4 Rd5 24.Kc2 c5 25.Ra4 Nd4+
>26.Bxd4 cxd4 27.Rxa7 Kd7 28.Kd3 Rxe5 29.Ra4 c5 30.b4 Ke6
>31.Ra6+ Kf5 32.b5 Rd5 33.Rc6 c4+ 34.Kxc4 d3 35.Kxd5 d2
>36.g4+ 1-0
>
>[D]r3k2r/p1p2ppp/bnp1q3/3pP3/2P4Q/1PB5/P4PPP/R3KB1R b KQkq -
>
>I found it very hard to dissuade Q5t from playing 14. ..Qe6-f5 instead of the
>more correct I think 14. ..d5xc4. At twelve ply it finds dxc4 often with a
>beautiful variation, a correct "prediction" of the game for ten full plies! The
>needed positional knowledge is in the Search here I think but the computer at
>the moment is just not fast enough yet. All I need is a factor ten in speed!
>Half an hour for a move now but beautiful variation. Whenever there is new
>hardware, whole new sorts of programs become feasible, that is my personal
>believe. And we don't even have eval in hardware yet, like Deep Blue but much
>more tunable, it will come maybe:
Gambit Tiger 1.0 would play dxc4 instantly and has kept the move for 3m 30s (I
have stopped the analysis at that point, it was starting iteration #13).
N6 0.27s dxc4 Be2 Nd5 Bxc4 Bxc4 Qxc4 Nxc3 Qxc3 O-O -0.40
N7 0.87s dxc4 Be2 Bb5 a4 Nd5 Bxc4 Bxc4 Qxc4 -0.38
N8 1.59s dxc4 Be2 Bb5 a4 Ba6 O-O O-O Qg5 -0.36
N9 3.95s dxc4 Be2 Bb5 a4 Ba6 O-O O-O f4 Nd5 -0.42
N10 7.90s dxc4 Be2 Bb5 a4 Ba6 O-O O-O f4 Nd5 Bg4 -0.38
N11 25.97s dxc4 Be2 Bb7 Bg4 f5 Bh5+ g6 Bd1 O-O Qg5 Nd5 Bd4 -0.40
N12 64.31s dxc4 Be2 O-O O-O Bb5 a4 Ba6 f4 Nd5 Bd4 Nb4 Bg4 -0.42
Christophe
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.