Author: Drexel,Michael
Date: 10:32:42 09/16/03
Go up one level in this thread
On September 16, 2003 at 12:55:06, Anthony Cozzie wrote: >On September 16, 2003 at 12:16:19, Drexel,Michael wrote: > >>On September 16, 2003 at 11:54:09, Anthony Cozzie wrote: >> >>>On September 16, 2003 at 11:50:35, Drexel,Michael wrote: >>> >>>>On September 16, 2003 at 08:38:33, scott farrell wrote: >>>> >>>>>On September 16, 2003 at 06:25:59, Drexel,Michael wrote: >>>>> >>>>>>On September 16, 2003 at 06:10:13, scott farrell wrote: >>>>>> >>>>>>>[d] r3r1k1/1b3pbp/2p2np1/1p1p1q2/pP1Pp3/P1P1P1PP/1B2QPK1/3RRNN1 b - - 0 107 >>>>>>> >>>>>>>If a human is playing white, and just moves d1d2, and d2d1, can this position be >>>>>>>broken open? >>>>>> >>>>>>Without analysis I dare to say the position is completely lost for white. >>>>>>h5,Bc8,Bf8,Bd6 >>>>>>Black is more or less a piece up and should win. >>>>>> >>>>>>Michael >>>>>> >>>>> >>>>>Isnt that roughly the line crafty played that I showed below. But I dont see the >>>>>continuation, and neither does crafty. >>>>> >>>>>How do you stop white from playing d1d2, d2d1 ? >>>>> >>>>>Scott >>>> >>>>In this position almost any move order will finally stop white from playing >>>>d1d2,d2d1 because checkmate ends the game :) >>>> >>>>In principle your program should avoid pawn chains against humans. >>>>If your program plays for example 1.e4 c6 2.d4 d5 3.e5 (no book of course) or >>>>1.c3 e5 2.d4 e4 then there is something wrong. >>>> >>>>Michael >>> >>>That is the crafty approach, and it certainly works. I'd rather teach Zappa to >>>understand them, though. I don't see any reason why a computer can't play closed >>>positions well. Its just a matter of writing some eval code. >>> >>>anthony >> >>In closed positions long-term plans are often important and this is not a matter >>of writing some eval code. >>Of course the human opponent will blunder more often in open positions and the >>computer will make less serious strategical mistakes. >> >>Michael > >I agree with you: computers are stronger in open positions. But it is not clear >to me why computers cannot approximate a plan with eval code. Granted this >"plan" may be thrown out the window at any time. In this position, for example: > >1. We have a closed center >2. Black has a large space advantage, and much greater mobility >3. Same-side castling >4. The queenside is completely blocked. > >Clearly the position calls for a kingside attack. White is helpless, as black's >has a spatial advantage on the only open theatre of action. > >If(1,2,3,4) then I need to plan kingside attack A closed position is usually not so simple. Of course this can help in some easy cases. > >is equivalent to >If(1,2,3,4) then (if Rook on Queenside(score -= 10)) or various more complicated >versions. > >anthony > >> >>> >>>> >>>> > >>>>>>> >>>>>>>Do any engines detect this behaviour and play differently? >>>>>>> >>>>>>>There is a player on ICC that is pretty good at doing this to my engine, and I >>>>>>>am trying to stop him with coding. >>>>>>> >>>>>>>I play with white it with crafty 19.3, with d1d2, d2d1, and this is the >>>>>>>continuation: >>>>>>> >>>>>>>[FEN "4r1k1/1b3pbp/r1p2np1/1p1p1q2/pP1Pp3/P1P1P1PP/1B1Q1PK1/3RRNN1 b - - 0 1"] >>>>>>>{-------------- >>>>>>>. . . . r . k . >>>>>>>. b . . . p b p >>>>>>>r . p . . n p . >>>>>>>. p . p . q . . >>>>>>>p P . P p . . . >>>>>>>P . P . P . P P >>>>>>>. B . Q . P K . >>>>>>>. . . R R N N . >>>>>>>black to play >>>>>>>--------------} >>>>>>>1... Bc8 2. Qe2 Qg5 3. Rd2 Bf5 4. Rdd1 Bf8 5. Rd2 Bd6 6. Rdd1 Qh6 7. Rd2 >>>>>>>Raa8 8. Rdd1 Rad8 9. Rd2 Re7 10. Rdd1 Be6 11. Rd2 Ree8 12. Rdd1 Rc8 13. Rd2 >>>>>>>Ra8 14. Rdd1 Rad8 15. Rd2 Rf8 16. Rdd1 Rfe8 17. Rd2 Rc8 18. Rdd1 Red8 19. >>>>>>>Rd2 >>>>>>>* >>>>>>> >>>>>>>with no changes in site: >>>>>>>depth=10 14/36 -1.42 19. ... Re8 20. Nh2 Bf5 21. Qf1 Ra8 22. Kh1 Rad8 23. Qg2 >>>>>>>Qg5 24. Ne2 >>>>>>> >>>>>>>Scott
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.