Author: Alastair Scott
Date: 08:47:22 11/28/02
Go up one level in this thread
On November 28, 2002 at 11:00:28, James T. Walker wrote: >On November 28, 2002 at 03:32:44, Alastair Scott wrote: > >>On November 28, 2002 at 00:55:28, Uri Blass wrote: >> >>>On November 27, 2002 at 23:45:47, Robert Hyatt wrote: >>> >>>>On November 27, 2002 at 15:44:03, Alastair Scott wrote: >>>> >>>>>On November 27, 2002 at 10:44:12, Richard Pijl wrote: >>>>> >>>>>>I had special detection code for rambling rooks in the Baron, but removed that >>>>>>in 0.99.3 as it fired wrongly in certain cases. This means that Baron 0.99.4 >>>>>>beta 1 doesn't find the draw. >>>>>> >>>>>>Baron 0.36 (containing the rambling rook detection code) has no problem >>>>>>detecting the draw, it spots it immediately, showing Kxe3 as the PV. When backup >>>>>>up to the diagram position it doesn't show the draw anymore. The reason is that >>>>>>Baron considers 60..gxh3 a mistake, enabling the draw. Instead, Baron 0.36 will >>>>>>play 60..Rge2! >>>>>> >>>>>>I have been thinking a lot about a better way to detect perpetual check's but >>>>>>haven't implemented anything yet. >>>>> >>>>>You're right - I saw Rge2 immediately after I played Rh3, but Ruffian never even >>>>>considered it. >>>>> >>>>>But Crafty 19.1 comes up with a remarkable shot after 33 seconds (at 14 ply) - >>>>>60..Rg1+. I doubt very much whether any human player would ever dream up _that_. >>>>> >>>>>Interestingly, the right move is played for the wrong reasons; Crafty, up until >>>>>then, plays 60..gxh3 with a score of -6.5. It then moves to 60..Rg1+ with a >>>>>score of -10 without ever realising that gxh3 should have a score of 0 ;) >>>>> >>>>>Alastair >>>> >>>> >>>>I'm not understanding the crafty comment above. If Rg1 is -10, and gxh3 is -6, >>>>I can see why it would prefer Rg1 over gxh3, so that white has g1 as a way out. >>>>But why would it want to play a move that leads to a draw when it can get a -10, >>>>which is good for black??? >>> >>>The comment is that crafty does not see the draw >>>for gxh3 and prefers Rg1+ only because Rg1+ >>>has a bigger advantage. >>> >>>This is what was considered as the wrong reason >>>by the poster. >> >>Couldn't have put it better myself :) >> >>This is proving a tricky position. For example, SOS plays ..gxh3 and doesn't >>seem to understand the position at all; it sees Rxe3+ but goes from 7 to 18 ply >>thereafter with a score of -6.6 ... >> >>Alastair > >Isn't that natural since it will probably take 100 ply to see the draw? >Jim Yes - it would seem that either: - a program has code to detect this sort of draw so sees it more or less instantaneously or - it doesn't and is condemned to traverse the 100 ply. It's interesting, as the position is virtually static and the 'pattern' - White Rook continually putting itself next to Black King without deflecting the Rg2 away from the g-file - repeats 50 times ... Alastair
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.