On 12/16/2010 10:07 PM, Matthew Twomey wrote: Hi,
> Ok, this still doesn't quite click for me - but I may still be > misunderstanding: > >> What we do now, is compare your scores of the position before 46.Ke6+ >> was played and after it was played. If the score-before is better (for >> white) than the score-after then, depending on the score difference, >> we annotate a variation for white's game move. > This comparing of the score before the move and after the move still > seems strange. What if as white, you were losing and facing a mate in 5 > from black. In this case, no matter what move you make - the score > before each given move will be better than the score after (even when > it's the "best" move)? This comparison may seem strange, but it is the best we can do. A move that is not in the game cannot be put on the board, and thus cannot be analysed. What you could do (to ease your thoughts a bit) is allow the engine to put his variations in the game and then run another annotation in which also variations are annotated. This will result in the figures you initially expected, as it allows you to compare the score after 46.Ke6 to the score _after_ 46.Ne4 As far as your "no matter what move" is concerned: If your white man plays the best move (allowing the mate in "only" 5) his score goes down by 1 cp indeed, but he is not criticised for that. Even not in annotate-all-moves mode, which must be enabled to anyway annotate the moves for a player who is in this dead-lost position. If white has multiple moves that all allow this mate in only 5, an alternate move may be given indeed, depending on the engine's preferred "defence". Code could be added to even prevent this, ..., but note that we are talking score differences here of 1 cp only; under normal conditions such differences are absolutely irrelevant during the course of the game... If white allows a faster mate an annotation may be justified. In Fritz-style something witty like "Even <bestmove> would not have turned the course of the game"... :-) Cheers, Joost. > -Matt > > > > On Thu, 2010-12-16 at 15:00 -0600, Matthew Twomey wrote: >> Joost, >> >> Apologies for the slow reply here. I'm still digesting what you've said. >> Regarding the specific method of annotation, I think I get it now. The >> key, for me, was here: >> >>>> What we do now, is compare your scores of the position before 46.Ke6+ >>>> was played and after it was played. >> What I though would happen is that it would compare the score of the >> position after 46.Ke6+ was played with the score after 46.Ne4 (the >> engine's suggested move) was played. >> >> Thanks for the clarification. >> >> -Matt >> >> This reply is taking me a long time because I'm still trying to get my >> >> On Tue, 2010-11-30 at 08:48 +0100, Joost 't Hart wrote: >>> On 11/30/2010 01:47 AM, Matthew Twomey wrote: >>>> On Mon, 2010-11-29 at 09:10 +0100, Joost 't Hart wrote: >>>>> On 11/29/2010 02:44 AM, Matthew Twomey wrote: >>>>> >>>>> Hi, >>>>> >>>>> Good that it works a bit more to your liking now! >>>>> >>>>> <zip> >>>>> >>>>>> I also did an analysis with Stockfish, and I am noticing one strange >>>>>> behavior. In the annotation, it's showing a Mate-in-X that is one more >>>>>> that it should be. For example, it's showing: >>>>>> >>>>>> ##### >>>>>> 46.Ke6+ +-- >>>>>> Stockfish 1.9.1 JA 64bit: 100:M4 >>>>>> >>>>>> (46.Ne4 48:M4 46..Kc8 47.Bg3 Kd8 48.Nc5 Kc8 49.Ra8#) >>>>>> ##### >>>>>> >>>>>> So in the annotation, it's reporting this variation as "M4" but >>>>>> shouldn't it be showing up as "M3"? When I look at the actual engine >>>>>> window, it's showing "M3". It's doing this with all mate-in-x >>>>> I am not sure we are in sync here: >>>>> >>>>> The line starting with 46.Ne4 is a mate in 4 moves, ending with 49.Ra8# >>>>> I cannot see how that should be an M3. >>>>> >>>>> As to the engine window, which position is on the board when it shows >>>>> M3? The position before or after 46.Ne4 has been played? >>>>> >>>>> Cheers, >>>>> Joost. >>>> Joost, >>>> >>>> The engine window shows M3 after 46.Ne4 has been played. I definitely >>>> could be misinterpreting or misunderstand the exact nature of the >>>> annotation, but it still appears inconsistent to me. Let me try to >>>> elaborate. In the example above, I understand (correctly?) that the >>>> first part of the annotation shows the evaluation of the move that was >>>> actually made: >>>> >>>> ##### >>>> 46.Ke6+ +-- >>>> Stockfish 1.9.1 JA 64bit: 100:M4 >>>> ##### >>>> >>>> This indicates that *after* 46.Ke6+ the position evaluates to M4 (this >>>> is confirmed via the engine window). Now the second part of evaluation: >>>> >>>> ##### >>>> (46.Ne4 48:M4 46..Kc8 47.Bg3 Kd8 48.Nc5 Kc8 49.Ra8#) >>>> ##### >>>> >>>> This is showing an evaluation of M4, but here it's showing the >>>> evaluation *before* 46.Ne4. So in the first case, the evaluation is >>>> based on the situation after move 46 for white, and in the second case >>>> it's showing the evaluation before move 46 for white. >>>> >>>> So it's possible that this is the intended behavior, but I'm just not >>>> sure. What I though the annotation would tell me is the "score" of the >>>> move I made, and the "score" of the suggested move. Please let me know >>>> if you believe I'm simply misunderstanding. >>> Matt, >>> >>> It is just a detail. And no misunderstanding, I think. >>> >>> And I must ask you the wrong question: picture yourself in the position >>> of the engine :-) >>> >>> What scid does is just autoplay the moves from the game. So every once >>> in a while you are told to stop thinking and tell us what you consider best. >>> >>> So after black's move 45, you start thinking about white's 46th. And you >>> see a mate in 4 line, starting with 46.Ne4. We stop your thought, you >>> tell us the M4 line starting with Ne4 and we restart your thinking after >>> putting 46.Ke6+ on the board. In this position you look for the best >>> line for black and - after we stopped you - you report your verdict on >>> the position. >>> >>> What we do now, is compare your scores of the position before 46.Ke6+ >>> was played and after it was played. >>> If the score-before is better (for white) than the score-after then, >>> depending on the score difference, we annotate a variation for white's >>> game move. >>> >>> Got it? So the score you see after 46.Ke6+ is really the score after >>> 46.Ke6 was played (and assuming black will answer best). The score you >>> see after 46.Ne4 is what the engine saw coming before 46.Ke6+ was played. >>> >>> As to the mating length: UCI reports the length in moves, that is, plies >>> by both black and white. This may look confusing, depending on who has >>> the move in the mating position. Put a position on the board in which >>> white can produce a mate but it is black to move. >>> >>> Start from (edit/setup start board) W: Kb6, Rb2; B: Ka8 with white to move. >>> >>> The engine will report 1.Rd2 (or whatever) ... Kb8 2. Rd8# >>> Now make the move 1.Rb4 >>> >>> There is still a mate (... Kb8 2.Rc4! Ka8 3.Rc8#). Look at the engine's >>> M numbers. >>> >>> Cheers, >>> Joost. >>> >>>> Thanks, >>>> >>>> -Matt >>>> >>>> >>>> >>>> >>>> >> >> >> ------------------------------------------------------------------------------ >> Lotusphere 2011 >> Register now for Lotusphere 2011 and learn how >> to connect the dots, take your collaborative environment >> to the next level, and enter the era of Social Business. >> http://p.sf.net/sfu/lotusphere-d2d >> _______________________________________________ >> Scid-users mailing list >> Scid-users@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/scid-users > > ------------------------------------------------------------------------------ Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d _______________________________________________ Scid-users mailing list Scid-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scid-users
