>>498631081>What are the chances that Naroditsky was running Leela while playing?the Leela team started contributing to Stockfish around the time the NNUE update was made after TCEC (was it 17?) demonstrated Leela had significantly superior understanding of pawn structure and absolutely trounced the fish in French-like structures. In that sense, Leela demonstrated a permanent win over the old Stockfish and the last of the vanilla AB engines decided to incorporate a neural network.
All this is to say that Stockfish and Leela will largely see the same lines, here. It won't matter which one he runs. They'd see both, they'd just disagree (maybe) on which is better.
>>498628358This move ended up being very hard to analyse. It's a great move for multiple reasons. The more I look at it, the more I think it's absolutely insane to see this in a few seconds.
Suppose we don't play Bc8 first. Then best seems to be ...h6, Bf4, Nxf4, gxf4, then Bf6 and black is winning. Here are what I think are the main ideas:
White needs activity, so Ne2 is your main idea to bring the knight to a better square (say d4).
Black has an advanced pawn, and can start marching on the kingside (push g5: surely white dare not let black play gf without knight on e2).
So let's say white plays Ne2.
Now if ...g5 immediately then f3 picks off the e pawn and white might even draw. Instead we might try Rd8, and then maybe if Nd4 then Bxd4 ed4 and we can double up on the d pawn
Now the only counterplay white has is this awkward f3 move to grab the e pawn. Note that if the bishop stays on the e file, black can't defend it and has to allow the trade (but is still better)
