28 Nov 07
Can anyone recommend decent software for doing "blunder-checks" in post-game analysis? I currently use Fritz 8, but I find that it does not give you any information as to why a move was considered a mistake. the obvious material loss situations are obvious enough, but what I'm really interested in knowing is why engines like Fritz might consider a move to cause, say, the loss equivalent to half a pawn (eg. it scores positions in 1/100ths of a pawn). Is anyone aware of websites or software that can give you more insight into these subtler kind of positional mistakes?
28 Nov 07
Originally posted by rijnsburgerwell I have quite experience with chess softwares, and I never came across any of the type you've described. you just have to figure those positional ideas on your own.
Can anyone recommend decent software for doing "blunder-checks" in post-game analysis? I currently use Fritz 8, but I find that it does not give you any information as to why a move was considered a mistake. the obvious material loss situations are obvious enough, but what I'm really interested in knowing is why engines like Fritz might consider a move t ...[text shortened]... es or software that can give you more insight into these subtler kind of positional mistakes?
28 Nov 07
Originally posted by rijnsburgerEarlier versions of Fruit and some of its spin offs (Toga and Gambit fruit) were open source i think.
I might have to resort to scouring through some source code blegh... Crafty's source is openly available, isnt it? Any other good open source engines around?
28 Nov 07
Originally posted by ChesswickI have CM8K and it has annotation in natural languages but ti doesn't give you the ideas behind the better moves. At least not what I think what ideas are.
Chessmaster does this I believe. I'm not sure what the strength of the engine is, but it's likely more than sufficient for simple blunder checking. It will actually give you a full English analysis of the position.
28 Nov 07
Originally posted by Chesswicki've tried chessmaster's analysis. it's ok, but it's explanations are usually quite one-dimensional or vague ("improves your king safety"... "creates a passed pawn"... "positions your knight at"...) the kind of feedback I'm looking for is probably a bit much to expect from software :/
Chessmaster does this I believe. I'm not sure what the strength of the engine is, but it's likely more than sufficient for simple blunder checking. It will actually give you a full English analysis of the position.
28 Nov 07
Originally posted by rijnsburgerI would love the exact same thing that you are looking for, but I think it might be a bit much to ask for in current software.
i've tried chessmaster's analysis. it's ok, but it's explanations are usually quite one-dimensional or vague ("improves your king safety"... "creates a passed pawn"... "positions your knight at"...) the kind of feedback I'm looking for is probably a bit much to expect from software :/
28 Nov 07
Originally posted by PsychoPawni think i'm going to look into getting source for a few better engines, then seeing if I can modify the any of these engine' evaluation functions to actually keep "reasons" alongside the position score modifiers (eg. if the score is modified by say -0.12 because a certain factor is better for black, I want to log a text description of that reason alongside the score modifier and then view the text descriptions with the score). In essence, I want to try and store the steps that cause a position score's calculation, not just the final score. It will slow the engine down a lot and probably consume los of ram, but it only needs to be done for the best evaluations found. speed is not an issue since I can let it calc as long as needed... seem feasible?
I would love the exact same thing that you are looking for, but I think it might be a bit much to ask for in current software.
28 Nov 07
2 edits
Originally posted by rijnsburgerhey, that reminded me of something. someone has tried to do that before on Toga or Fruit, or some other open source free engine. he had written a program to translate Fruit's evaluation of positions to natural language.
i think i'm going to look into getting source for a few better engines, then seeing if I can modify the any of these engine' evaluation functions to actually keep "reasons" alongside the position score modifiers (eg. if the score is modified by say -0.12 because a certain factor is better for black, I want to log a text description of that reason alongsi ns found. speed is not an issue since I can let it calc as long as needed... seem feasible?
I had tried to take a look at it, but the process was pretty painful so I had given up. look for edits in this post to see if I can find any link.
here comes the edit: I made a deep google search but I couldn't find anything. it was on a personal website. maybe it's down now, sorry.
28 Nov 07
Originally posted by diskamylalso didnt find anything... but worth a try looking into it to see if adding the annotation isnt going to be too difficult. i suspect it will though since I doubt the logic to add/subtract to positional scores is laid out all neatly and ready for annotation code 🙂
hey, that reminded me of something. someone has tried to do that before on Toga or Fruit, or some other open source free engine. he had written a program to translate Fruit's evaluation of positions to natural language.
I had tried to take a look at it, but the process was pretty painful so I had given up. look for edits in this post to see if I can fin ...[text shortened]... earch but I couldn't find anything. it was on a personal website. maybe it's down now, sorry.
28 Nov 07
Originally posted by rijnsburgerAnd chess coaches worldwide are desperately hoping that you fail miserably in your quest to find a silicon positional coach. 😉
also didnt find anything... but worth a try looking into it to see if adding the annotation isnt going to be too difficult. i suspect it will though since I doubt the logic to add/subtract to positional scores is laid out all neatly and ready for annotation code 🙂
29 Nov 07
Originally posted by rijnsburgerWell, in the worst case, you could always put fritz on "infinite analysis mode", and as you play through a game, look at the moves which cause a .5 pawn or more drop in your position. Fritz 10 (im not sure about 8) has a feature (I think) where it can find blunders, and you can set the "threshold" (how sensitive the blunder check it is) to do this, i think you save the game, open it and click tools, analyze, and it will come up with a box with all the analysis settings, blunder check being one of them.
Can anyone recommend decent software for doing "blunder-checks" in post-game analysis? I currently use Fritz 8, but I find that it does not give you any information as to why a move was considered a mistake. the obvious material loss situations are obvious enough, but what I'm really interested in knowing is why engines like Fritz might consider a move t ...[text shortened]... es or software that can give you more insight into these subtler kind of positional mistakes?