Author: martin fierz
Date: 12:52:03 09/17/02
Go up one level in this thread
On September 17, 2002 at 13:25:19, Dieter Buerssner wrote:
>On September 16, 2002 at 19:17:08, martin fierz wrote:
>
>>>>>>> 8/8/4K1p1/4PpB1/3p1P1P/5n2/4k3/8 b - -
>>3 after all :-) or some form of knowledge that the knight can come to the
>>rescue. anyway, it gets a bit complicated :-(
>
>Yes. One can also construct positions with connected passed pawns of rook and
>knight pawns, that are drawn. But your condition 3 also seems not fullfilled.
>I also try to be careful with such rules (that give really high positional
>bonus). I try to think, either the rule should have essentially no exceptions,
>or the search will find the exceptions fast.
>
>>i still like the idea of pawn on 7th rank = +3 in positions where the total eval
>>is >+7 or so (or +4 with >+8 or whatever). should make you find the 2...Nd4
>>line, right?
>
>Yes and no. 2...Nd4 is seen fast in the mainline:
>
> 320305 2.476 5.26 9t 1...d3 2. Kf7 Nd4 3. e6 Nxe6 4. Kxe6 d2 5. Kf6
> d1=Q 6. Kxg6 Qd5 {571}
>But what happens here?
>
> 567274 4.610 4.91 10t 1...d3 2. Kf7 d2 3. e6 d1=Q 4. e7 Qd7 5. Bf6
> Nxh4 6. Kf8 Qd6 7. Be5H {991}
>
>Well, now the search sees lines like (after 1...d3 2. Kf7 Nd4):
>
>3. Kxg6 d2 4. h5 d1=Q 5. h6 Ne6 6. h7
>
>And again we have the large positional score for the pawn on 7th rank. This is
>(disregarding extensions) already 10 ply after the root position. Some more
>plies are needed to see, that h7 won't help here. After all, still a rather deep
>search is needed, and probably it would still make my engine draw this position.
>Instead of +7 or 8 scores, I'd see only +5 scores, but nothing more may change.
shit :-)
i thought it was a good idea with the pawn on the 7th, but i realize that it
gets too complicated :-(
the reason i as a human am not as concerned about this version of the pawn on
the 7th than in the first example is that a pawn on the 7th against a queen and
nothing else is relatively hard to stop. in the other position, i have N+Q which
generally will mate a king, unlike a Q alone. plus i might have the possibility
of saccing the N for that pawn. but obviously these are too many rules to
implement, besides that there will be numerous examples where they don't work!
aloha
martin
>Regards,
>Dieter
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.