Author: Uri Blass
Date: 19:29:48 11/09/00
Go up one level in this thread
On November 09, 2000 at 20:25:39, Chessfun wrote:
>On November 09, 2000 at 19:38:42, Uri Blass wrote:
>
>>On November 09, 2000 at 18:33:05, Robert Hyatt wrote:
>>
>><snipped>
>>>OK... Here is a good point for discussion. My evaluation at this point
>>>is somewhere between -.2 and -1.0, depending on how deep you let it
>>>search. Here is the position:
>>>
>>>[D]2rr2k1/1p1b2p1/p3p2n/2Bp3B/1P3P1Q/PP5R/1q5P/4R2K b
>>>
>>>It is black to move. White has the rook and queen doubled on the h file.
>>>Black has the h-file blocked with the knight on h6. White has no way to
>>>drive the knight off, and no easy way to capture the knight with some sort
>>>of trade. IE the h-file is not going to be used for an attack very easily.
>>>
>>>Do you (or anybody) think that white is really up the equivalent of one
>>>piece (+3.42) here?
>>
>>I think that the last move Kh1 was a mistake and better was Qe7 and white is
>>winning for example Qe7 e5 Rg3 exf4 Rg2 Qc3 Rd1
>>
>>Tiger13 is stronger and has no problem to find Qe7 but gambittiger prefers
>>speculation of 3 pawns and not winning material.
>>
>>Gambittiger did not like Qe7 because of Qf6 and white wins only the excange that
>>is not enough for 3 pawns.
>>
>>
>><snipped>
>>>>70...Ne8 71.Rxc6 { Score: 0,96
>>>
>>>Prior to this point, I would have expected a draw. two pieces for a rook
>>>is generally an advantage. I would not expect to lose such a position very
>>>often since the two pieces are generally much stronger. This "bad trade"
>>>score has been influencing the scores crafty produces for quite a while in
>>>the game. It is one of those 'speculative' type scores that just happens to
>>>be wrong here, but it is right far more often than it is wrong.
>>
>>Gambittiger also expected a draw(evaluation +0.24 before Ne8) and I think that
>>the mistake of crafty was 70...Ne8 and I guess that after 70...Nd5
>>
>>Uri
>
>Uri,
> I read that you now had Rebel 11. I would be interested whenever you
>have time in reading about your impressions. Strength?, interface? features etc.
>
>Sarah.
I do not regret for buying it because I need new programs for my correspondence
games but there are some things that I do not like about it.
The main thing that I do not like is the fact that tiger does significant
preprocessing.
I knew that it does preprocessing before buying it but I hoped that it does less
preprocessing.
Here is an example from the famous game against nimzo after Rc6 Nxc6
[D]1k1r4/p2r4/1pn5/1P2p1q1/P3Qp1p/5P1P/2R3P1/6BK w - - 0 1
The scores of gambit before 44.bxc6
3.54 depth 11
2.96 depth 12
2.40 depth 13
2.30 depth 14
1.80 depth 15
1.68 depth 16
The scores after bxc6 Rc7(I download another engine and downloaded again gambit
in order to make gambit forgetting this information because I did not want to
have influence of learning from previous search about the results)
1.66 depth 9
1.72 depth 10
1.54 depth 11
1.12 depth 12
1.32 depth 13
1.30 depth 14
1.18 depth 15
You can see that there is a significant drop in the evaluation that is more than
one pawn in part of the cases and not because of deeper search.
I believe that correct evaluation is important at long time control because
there are things that you cannot see by search.
I also cannot build a tree and use the evaluations of tiger to decide about my
move when I see this behaviour because the evaluations of the leaves after
search are not based on the same root and 0.6 pawn advantage on one leaf can be
better than 1 pawns advantage on another leaf.
My experience with chess programs is that there are programs without this
behaviour(chessmaster6000 is one example and it seems to me that gandalf is also
better but I did not use a lot of time to test it)
I think that a good idea for the future may be to give the users an option to
analyze with parameter of the root position(assuming that avoiding this
preprocessing without reducing the level of the program is impossible).
I also did not see an option to play engine-engine games on one computer between
gambittiger and tiger13 and I only found a way to play the same engine against
itself.
engine-enigne games is one of the things that I use in my analysis of my
correspondence games.
Inspite of all these facts I am going to use Gambittiger for my correspondence
games because it may suggest interesting moves that other engines do not
consider but I am not going to trust the scores of it and the same for the other
engines in the package of Rebel1.
Uri
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.