Author: Ricardo Gibert
Date: 20:15:19 06/05/00
Go up one level in this thread
On June 05, 2000 at 20:46:11, Bruce Moreland wrote: >On June 05, 2000 at 08:22:44, Steffen Jakob wrote: > >>Hi! >> >>This is a position from an interesting standard game MissSilicon - Hossa, played >>today at ICC: >> >>[D]5k2/7K/6P1/1p3p2/1P5P/1Pb5/8/8 w > >This is more interesting: > >[D]5k2/7K/6P1/1p3p2/1P5P/2b5/8/8 w > >The reason it's more interesting is that you don't get false evals due to white >being up two pawns to one early on, and later three pawns to one. > >This isn't a pawn race problem. After 1. h5 it should be easy to see that 2. h6 >is a terrible threat that costs a bishop. > >Once you get to that point, white gets the black f-pawn and we are left with a >position a pawn up. > >It could be possible to mis-evaluate this as a win for white, due to the fact >that white is a pawn up. The only obvious thing preventing this mis-evaluation >is the fact that it really is a win for white. > >Here is a much harder position, especially without tables or a little tricky >pawn knowledge: > >[D]5k2/7K/6P1/p4p2/P6P/2b5/8/8 w - > >Let your program chew on that for a bit. You can achieve a similar position >here, but the difference is that the q-side situation does not allow for a win >once the g-pawn is sacrificed. > >My program has been chewing on this for a while on a slow machine and still >thinks it is a win for white. > >I've seen several people hail various programs as brilliant in the original >position, but I suggest that a combination of smart and coincidentally lucky >doesn't add up to real brilliance. Your point is well taken, but there is nothing wrong with being lucky! You must not forget that one program can be "lucky" more often than another if it has a better eval, so perfection, though desirable, is not required. Very hard to do as I'm sure you can atest. How one program does on one position is pretty meaningless, while how it does in the long haul is what counts. > >bruce
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.