Computer analysis bug in simul game "from position"

I have completely blundered and have a -10 lost position. But the computer analysis is showing 0, 0, 0 and 13 cp for me. I am unable to logically explain this. Seems like a bug.

Should I raise this as an issue in github?

@Aadithya2002 here is what is going on. the engine, when you tell it to analyze, only goes to around depth 15 per move. that is not deep enough to know what is actually going on. Try letting it st=it for a minute and observe what happens. On my computer, it decided white was better.
When the engine analyzed however, it didn't go deep enough and thought white was the one who had blundered,
so you ACP is low even though you blundered.

@ChessMathNerd Hi there! Ok its quite interesting to see that a person who could see 15 moves deep would have played this thinking it to be advantageous. This apart,

But the "request computer analysis" is supposed to give a comprehensive analysis of the game. The fact that it has failed to do so in this particular case is a pitfall of its function that I have never seen till now in 2 thousand of my games.

Is this an isolated case or have you or anyone else experienced this already?

I have verified it and it seems to be a bug. Here's my verification -



Note that the result of the "request computer analysis" is different and accurate here.

And the in-browser engine analysis doesn't think black is better until after depth 15 (like it does in the game analysis) and behaves 'properly'

The two markedly different analyses of the same game.

@Aadithya2002 this is a very rare occurence. The fact that it analyzed differently the second time could be due to the fact that your computer was running slightly faster the second time. It just so happens that in this case, the blunder is exactly the right difficulty to find, that the engine will see it sometimes, and not see it other times.
EDIT: Request computer analysis does not give comprehensive analysis. It gives light analysis. I have had many games where it believed a certain thing to be an inaccuracy, but when I went to that spot and let it go deeper, it liked my move, or my opponent's move.

The "request computer analysis " result is clearly different when it should be the same.

Let me mention a lichess developer - @nojoke

Thanks @ChessMathNerd

@Aadithya2002 request computer analysis uses the local processing power, it is dependent on the speed provided. Lichess does not provide the CPU time for the engine, to my knowledge.
Any two analyses of the same game are likely to be very slightly different.

@ChessMathNerd Oh # 7 (edit) okay... but a blunder is not as subtle though.. I don't know , your points seem to be correct, but still think this is worth to be looked at by a site developer.