Author: Sune Fischer
Date: 15:18:43 06/18/03
Go up one level in this thread
On June 18, 2003 at 17:20:00, Uri Blass wrote: >On June 18, 2003 at 16:32:33, Sune Fischer wrote: > >>On June 17, 2003 at 14:08:26, Uri Blass wrote: >> >> >>>>The problem is this: If the position _starts_ off with 5 pieces, it will >>>>play _perfectly_. If it starts off with more, it might not. IE it might >>> >>>Here is the relevant part of another post(without a part that can be deleted >>>because of personal attack and with the words "from the original link" to >>>prevent another conversation. >>> >>> I don't know why this conversation is still going on. >>>The position in the diagram has 8 pieces, right? Then there's the comment >>>from the original link: >>> >>>"It's funny that even if we sweep away three white pawns, both engines evaluate >>>White's position as winning." >>> >>>-Tom >>> >>>My comments: >>> >>>I agree that there is no need to continue the conversation because it is clear >>>based on the original link that the engines were wrong in tablebases position >>>and it is also clear that they did not use the full 5 piece tablebases otherwise >>>the problems of not finding draw score even in draw tablebases positions could >>>be solved. >>> >>>I know that at least Junior does not use swindle mode and swindle mode is >>>impossible because swindle mode means that at least one of the engines can >>>evaluate the position correctly >>> >>>Uri >> >>Bob is right that when analyzing a 8 man position with 5 man TBs there is no >>guarantee it will be solved, even if the PV shows only a 5 man position at the >>end, all depends how deep the TBs are probed in the search. > >The point was not about the pv. >Here is the quote again: >"It's funny that even if we sweep away three white pawns, both engines evaluate >White's position as winning" > >It seems clear to me that the author talks about the position after white lose 3 >pawns and not about the pv. It does sound that way. I guess there is no excuse then. >> >>Problem is often that the line is not forced, for instance there is no reason >>the king should eat the pawns and allow a mate score from the TBs. >> >>So if we give Shirov the benefit of the doubt, that he did install the full set >>of TBs, then I think that is what happened. It shouldn't be too hard for others >>to confirm though. > >1)Shipov and not shirov if I remember correctly(I am too lazy to look at the >link) yup, correct. >2)It is clear that in another position in the link the root position was KBP vs >KP and the programs did not evaluate it correctly so it is clear that not all >the 5 piece tablebases were installed. right it's position 6, "most likely due to the defects of the 5-piece tablebase" :) -S. >Uri
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.