Rebel wrote: ↑Mon May 12, 2025 8:29 pm
Note 14.Nf5 a knight sacrifice with a positive score of +1.92 black happily accepts with Qxf5 also with a positive score of 1.96, so a difference in evaluation of almost 4 pawns.
15.c4 Qd7 the difference in evaluation score is now 5 pawns.
16. Nb3 Nxc4 the difference in evaluation score is now 6 pawns.
17. Nc5 evaluation score difference is now 8 pawns, black has no idea what's coming to him.
20...Qc8 finally SF12 HCE (the search wonder) sees the loss.
Classic search vs evaluation example.
[Q] which engines find move 14, 15, 16 and 17?
Huh? I can't see any of the differences you give in the PGN you provided. What am I missing?
Ed's PGN evals are from side to move POV, not white's POV. Just take a look near the end of the game, with mating evals from both sides, each with a different sign:
Ajedrecista wrote: ↑Mon May 12, 2025 6:39 pm
Enclosing the quoted post with PGN tags for a better visualization of the game. Friendly reminder:
There is algebraic notation in German for the suggested moves between brackets: K for King (the same than in English, hence been recognized the king figure), D for Queen, L for Bishop and S for Knight.
Regards from Spain.
Ajedrecista.
Dear Ajedrecista,
I am aware of this option, but thanks for your friendly and well-meaning reminder. Unfortunately posting like you suggest goes with a major loss of information on the game. All my ugly PGN go with times and evaluation for every single move, and in case of doubt I could always provide the PGN from the other computer that would look slightly different but otherwise similar to add the opponent's view in the same way.
Does anyone even look at games you post these days? Who knows ( I certainly don't). But in case someone does: why not just copy-paste the game to a chessprogram of your choice instead of clicking through it on a forum?
I have established my own standards on doing and posting games on CCC more than 20 years ago. I am too old to change them now without good reasons.
Ajedrecista wrote: ↑Mon May 12, 2025 9:09 pm
Ed's PGN evals are from side to move POV, not white's POV. Just take a look near the end of the game, with mating evals from both sides, each with a different sign:
So the PGN has been tampered with (though certainly with good intentions) to make them less reliable to trust and understand.
BTW: evals from side to move POV are heavily old-fashioned, aren't they?
I am aware of this option, but thanks for your friendly and well-meaning reminder. Unfortunately posting like you suggest goes with a major loss of information on the game. All my ugly PGN go with times and evaluation for every single move, and in case of doubt I could always provide the PGN from the other computer that would look slightly different but otherwise similar to add the opponent's view in the same way.
Does anyone even look at games you post these days? Who knows ( I certainly don't). But in case someone does: why not just copy-paste the game to a chessprogram of your choice instead of clicking through it on a forum?
[...]
Thanks for noting it was well-meaning.
Clicking on c8 and d8 squares of the generated board with PGN tags recovers the info, including times and evals.
Rebel wrote: ↑Mon May 12, 2025 8:29 pm
Note 14.Nf5 a knight sacrifice with a positive score of +1.92 black happily accepts with Qxf5 also with a positive score of 1.96, so a difference in evaluation of almost 4 pawns.
15.c4 Qd7 the difference in evaluation score is now 5 pawns.
16. Nb3 Nxc4 the difference in evaluation score is now 6 pawns.
17. Nc5 evaluation score difference is now 8 pawns, black has no idea what's coming to him.
20...Qc8 finally SF12 HCE (the search wonder) sees the loss.
Classic search vs evaluation example.
[Q] which engines find move 14, 15, 16 and 17?
Huh? I can't see any of the differences you give in the PGN you provided. What am I missing?
1. The game I posted is from the UHO-rating list, not your game.
2. Click on the square c8 to view the raw PGN.
3. Moves are followed by a comment something like 14. Nf5 +1.92/19 5.8s
+1.92 score from the side to move
19 search depth
5.8s time used
4. As far as I know no engine (or at least the very vast majority) uses the Chessbase white point of view scores.
90% of coding is debugging, the other 10% is writing bugs.
Rebel wrote: ↑Mon May 12, 2025 8:29 pm
Note 14.Nf5 a knight sacrifice with a positive score of +1.92 black happily accepts with Qxf5 also with a positive score of 1.96, so a difference in evaluation of almost 4 pawns.
15.c4 Qd7 the difference in evaluation score is now 5 pawns.
1. The game I posted is from the UHO-rating list, not your game.
2. Click on the square c8 to view the raw PGN.
3. Moves are followed by a comment something like 14. Nf5 +1.92/19 5.8s
+1.92 score from the side to move
19 search depth
5.8s time used
4. As far as I know no engine (or at least the very vast majority) uses the Chessbase white point of view scores.
All of the games of my UHO ratinglist are played, using cutechess-cli. And cutechess just puts the numbers, it gets from the engine, in the comments of the played move. So, the engines decide, if their eval is WhitePOV or Player POV. Not me. And not cutechess.
Rebel wrote: ↑Mon May 12, 2025 8:29 pm
Note 14.Nf5 a knight sacrifice with a positive score of +1.92 black happily accepts with Qxf5 also with a positive score of 1.96, so a difference in evaluation of almost 4 pawns.
15.c4 Qd7 the difference in evaluation score is now 5 pawns.
1. The game I posted is from the UHO-rating list, not your game.
2. Click on the square c8 to view the raw PGN.
3. Moves are followed by a comment something like 14. Nf5 +1.92/19 5.8s
+1.92 score from the side to move
19 search depth
5.8s time used
4. As far as I know no engine (or at least the very vast majority) uses the Chessbase white point of view scores.
All of the games of my UHO ratinglist are played, using cutechess-cli. And cutechess just puts the numbers, it gets from the engine, in the comments of the played move. So, the engines decide, if their eval is WhitePOV or Player POV. Not me. And not cutechess.
Standard is POV side on move, it’s only rare non-standard ancient engines that differ. Any other (random) POV is going to break early draw, early win etc adjudications, no?
Ajedrecista wrote: ↑Mon May 12, 2025 6:39 pm
Enclosing the quoted post with PGN tags for a better visualization of the game. Friendly reminder:
There is algebraic notation in German for the suggested moves between brackets: K for King (the same than in English, hence been recognized the king figure), D for Queen, L for Bishop and S for Knight.
Regards from Spain.
Ajedrecista.
Dear Ajedrecista,
I am aware of this option, but thanks for your friendly and well-meaning reminder. Unfortunately posting like you suggest goes with a major loss of information on the game. All my ugly PGN go with times and evaluation for every single move, and in case of doubt I could always provide the PGN from the other computer that would look slightly different but otherwise similar to add the opponent's view in the same way.
Does anyone even look at games you post these days? Who knows ( I certainly don't). But in case someone does: why not just copy-paste the game to a chessprogram of your choice instead of clicking through it on a forum?
I have established my own standards on doing and posting games on CCC more than 20 years ago. I am too old to change them now without good reasons.
Ajedrecista wrote: ↑Mon May 12, 2025 9:09 pm
Ed's PGN evals are from side to move POV, not white's POV. Just take a look near the end of the game, with mating evals from both sides, each with a different sign:
So the PGN has been tampered with (though certainly with good intentions) to make them less reliable to trust and understand.
BTW: evals from side to move POV are heavily old-fashioned, aren't they?
Peter
No they aren’t, they’re standard. As ever the accusation is a confession
