A few weeks ago I checked a game I played with Stockfish and I was a bit surprised to see the move 5...Nbc6 having a question mark. This was a recommended line by several people in a forum thread (elsewhere) some time ago.
I just compared a few games :
en.lichess.org/v7fT6LMj/black#10Two days later it gets a ?! mark :
en.lichess.org/n3dFvHvb/black#10And today there's no remark from SF, and it gets a +0.7 :
en.lichess.org/Ew0X7ZJg/black#10What's going on ? :-)
According to the Universal Chess Interface protocol, scores returned by Stockfish are subjective (or loosely defined):
"cp: the score from the engine's point of view in centipawns"
http://wbec-ridderkerk.nl/html/UCIProtocol.htmlTherefore any annotations generated from them are also subjective.
It's going from (by date) "?" to "?!" to "nothing".
it seems Stockfish is learning from your games.
:-)
Maybe stockfish is using different evaluation time different times. Just an iidea, I don't know.
For me the move looks weird. Like black is sacing a piece for absolutely nothing. He doesn't even get centre control in compensation.
I put the position into my local standalone stockfish 7.
It says after thinking a while:
Depth 27, Tme 4282.51, Nodes 240.817M
+1.09
6. fxe5
#4
I agree. Just 6.Nc3 seems very nice for white. Also accepting the piece seems ok.
p.s. Mr zombie. I should have played Bxc6 in the simul.
That's the problem with too little time/too many boards combo.
No time to consider double-edged long-term implications. Result is superficiality.
Yes, time seems a problem for many simul players on lichess. You can try adding extra time for the host because it's often needed.
Bxc6 is good? Exchanging the fianchetto bishop for the knight?
Small differences in timing can result in large changes in the evaluation returned, especially at the relatively low iteration counts used for the analysis.
One of the many reasons we have to use our brains and not just accept shallow SF analysis at face value :)
@ #7
Yes, Bxc6 (instead of the silly Re1) after your 0-0 was slightly better for white. Even possible the move before that instead of d3.
You should have gone 9..., Bd7
Not really on topic, sorry OP. Last time, I promise :-).
#8 Honestly, I *never* trust engine evaluations at face value. They're only meant for itself and the programmer trying to fix bugs.
