Computer chess analysis

Here is a potential example...

13. a6?? (-2.1 Stockfish 13. Be3+- 2.3)

What does all this mean?
13. a6 -> whites move
??-> chess punctuation which means blunder
(-2.1 ->with the move a6 white is behind by 2.1 pts
13. Be3 -> best move
+- -> White has a decisive advantage
2.3) -> with the best move white has an advantage of 2.3 pts.

That is the most complicated it gets. You can always exclude,
the +- or = if you have a point differential being displayed.

I hope im making sense. Would appreciate comment.

Yeah I suppose there are two categories into which you could divide non-English, statistical displays of the text.

The advantage indicator:
+- (White has the upper hand.)
-+ (Black has the upper hand.)
+= (White has the advantage.)
-= (Black has the advantage.)
= (The game is just about equal.)
a few other technical punctuations you could read about on wiki

And the move-specific annotation, per the move currently being analyzed on Lichess (and not just the overall game status):
? (bad move)
?? (blunder)
! (good move)
!! (excellent move)
?! (uncertain move i.e. risky, dubious)
!? (interesting move)
and once again etc.

In my opinion both informations should be displayed. The translation files for Lichess might need to include a key/legend to explain what those symbols mean elsewhere, if it isn't embedded.

that's exactly what I want to do. At the moment I'm trying stockfish in command line and I don't see a way to make it output the punctuations and advandage indicator for a given game.

I keep searching. That's a new world to me.

==> "Play against the machine" does not really accomplish the same thing because if Crafty replies with a different move than the one your opponent did, you lose control over replaying the game in Crafty because the position and the rest of the game is changed.

True, but you still may be able to make this work. I was thinking you could start crafty with each position, and then get crafty's move.

However, correcting myself, there is no reason to do this because crafty has analysis mode:

http://www.cis.uab.edu/hyatt/craftydoc.html#analyze

But since stockfish looks like it would be better, you may want to change to it, although you could consider changing engines after you implement analysis if they have the same commands.

Well, if the punctuations and game advantage indicators are not already annotated from within the source game file, then they have to come from a different source (probably the engine's analysis). I haven't seen a way to do that with any UCI engines yet, though; I just play through the moves individually, and the analysis per move will always tell me what the advantage is.

More info about UCI chess engine commands is elaborated here:
http://www.open-aurec.com/wbforum/WinBoard/engine-intf.html#8

[someonesomewhere]

Yeah, looks like this UCI stuff and integration for doing exactly as thibault wishes does take a little investigation. Unless maybe your info there about Crafty might provide some clues as to how to do it for other UCI engines (not that all versions of Crafty are UCI, if I recall correct).

"In my opinion both informations should be displayed. The translation files for Lichess might need to include a key/legend to explain what those symbols mean elsewhere, if it isn't embedded."

I am okay with this.

Maybe someone can edit my example to be more visually pleasing.

13. a6?? (-2.1 Stockfish 13. Be3+- 2.3)

Not sure if putting the -2.1 and then 2.3 in the same statement is smart.

hmmm

13.a6?? -+ (13.Be3! +-)

But how to fit the numerical score in there is another challenge. I suppose most people don't need to know the exact (possibly biased) numerical score estimates as long as other moves aren't shown in the suggestions queue per the same turn; I find the numbers are extremely useful while doing a relative analysis of a single move as to how various move choices are better or worse in relation than others. Otherwise I suppose it might be possible to get by without the numbers, but still, they could be useful anyway.

People might have some confusion as to units.
UCI engines give the unit in [pawns], e.g. -2.1 pawns, +2.3 pawns (White's score).
Other engines might use [centipawns] "cp", even interfaces for Rybka and such, e.g. -210, +230, etc.

okay how about:

13.a6?? {-2.1 pawns, -+ (13.Be3! +-, 2.3 pawns)}

Would be the PGN-compatible way to do it, unless you're able to wrap lines instead of using single-line output, if you really want to include the score in pawns.

Of course the unit of pawns is slightly misleading since it does not necessarily correspond to checkmate, until engines so advanced as to base on other much more positional factors.

But PGN game text supports annotations/comments in the form of {comment (nested comment)}, so that would be the PGN-compatible way of outputting the information.

What's the status?

I've not found a way to make stockfish annotate a game (yet?).

But thanks to someone's link, I've been able to use crafty to produce that: http://sacem.iliaz.com/sample.pgn.html

now that looks closer to something I could exploit.