Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: a couple of hard positions (material gain that disappears)

Author: Robert Hyatt

Date: 18:55:52 06/12/01

Go up one level in this thread


On June 12, 2001 at 10:22:52, Jon Dart wrote:

>Here are a couple of positions from recent Arasan games:
>
>[D] 1n2r3/5Nk1/p4b2/2Qp3p/6p1/P5P1/P1B1q1PP/6RK b - -
>
>Here taking the piece with Kxf7 is fatal. It appears
>however that after Be7! Black doesn't have anything better
>than a draw.

Here, on my P3/750 Sony notebook, I get the following:

               10    25.21   0.00   1. ... Kxf7 2. Qxd5+ Kg7 3. Qxh5 Bb2
                                    4. Qg6+ Kf8 5. Qd6+ Kg8 6. Qg6+ Kf8
               10    37.35  -0.05   1. ... Qc4 2. Qxc4 dxc4 3. Nd6 Re2
                                    4. Bd1 Rxa2 5. Nxc4 Be7 6. h3 Bxa3
                                    7. Nxa3 Rxa3 8. hxg4 hxg4 9. Bxg4 Rxg3
               10->  38.20  -0.05   1. ... Qc4 2. Qxc4 dxc4 3. Nd6 Re2
                                    4. Bd1 Rxa2 5. Nxc4 Be7 6. h3 Bxa3
                                    7. Nxa3 Rxa3 8. hxg4 hxg4 9. Bxg4 Rxg3
               11    41.63   0.06   1. ... Qc4 2. Qxc4 dxc4 3. Nd6 Re2
                                    4. Bd1 Rxa2 5. Nxc4 Be7 6. Bb3 Rf2
                                    7. Re1
               11     1:27   0.00   1. ... Be7 2. Qxd5 Qxc2 3. Qe5+ Kxf7
                                    4. Rf1+ Kg6 5. Qe6+ Kg5 6. Qe3+ Kg6
                                    7. Qe6+
               11->   1:36   0.00   1. ... Be7 2. Qxd5 Qxc2 3. Qe5+ Kxf7
                                    4. Rf1+ Kg6 5. Qe6+ Kg5 6. Qe3+ Kg6
                                    7. Qe6+



>
>This one is an opening trap sprung by the program
>"Postmodernist":
>
>[D] rn1qk1nr/ppp3pp/5p2/2bP1bB1/3p4/1Q3N2/PP2PPPP/RN2KB1R w KQkq -
>
>Most programs will play Qxb7, but this loses material as the
>Queen can be trapped. There isn't really a good move here -
>White's mistake was earlier in the game, and he's already
>worse. Crafty eventually selects Bf4 (at ply 12)
>with a minus score.

Current program finds problems at 11 plies, almost 2 minutes on my notebook.
Once it fails low, it will search to resolve it and find a better move so
if it has 2 mins/move it would find it on my notebook.  On my quad, 30-40
seconds would be enough to fail low and bail out of Qxb7.  Score after the
fail low is -2...





>
>This is another example where grabbing material is the
>wrong choice. A human might well avoid Qxb7 on the general
>rule that Queen sorties in the opening are a bad idea,
>but most computers won't weight that factor heavily
>enough and must discover the badness of Qxb7 by search.



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.