Another interesting position for which Stockfish 14 failed to find the best move in the 3 second (depth 22) test. Stockfish did find one of the alternate best moves (h3). However, Stockfish even at depth 37 was still picking h3, though the better move Ne4 had the same evaluation. If Stockfish was somehow programmed with the notion of 'testing line', which means making it hard for the opponent, then it would pick the better move.
My study on this position is here:
https://lichess.org/study/DMOJQHcr/cLzwNkyD#0
The Human Perspective
Black's kingside position is weak; in particular the light-squares and the g-file. White has a well posted Knight on e5, and control of e4. White should look for a way to get a kingside attack going. 1.Ne4 comes to mind as it centralizes the other Knight and creates threats around the Black King. Indeed, analysis shows that White is threatening an immediate win after 1.Ne4 with 2.Qd1. Black has to play carefully for many moves. A variation in which your opponent has to play carefully or lose immediately is referred to by many as a 'testing line', or 'testing variation', because you are 'testing' your opponent. It is very difficult for a human to play the correct moves in such a defensive situation.
Stockfish's Perspective
Even though 1.h3 is the Stockfish choice at a depth of 37, that has the same score as 1.Ne4. The variation from 1.h3 allows Black to get rid of the e5-Knight, and post his Queen Bishop on g6. Stockfish has no notions of one variation being more 'difficult' than another. It picks 1.h3 at depth 37 because it has the same evaluation as 1.Ne4, and, I'm guessing, simply came to its attention first.
EPD Changes Needed
1.Nb5 allows too many exchanges and Black keeps his outside passed pawn. The same can be said for 1.Be4.1.Rg1 is an interesting tactical line.1.Bf3 allows 1...Bxe5 and White's initiative dissipates.1.Qb2 allows 1...f6; which ejects the Knight. I would leave the EPD as it is. Especially if there are multiple moves with the same evaluation, Stockfish should be made to understand that the most 'testing' line is the one to play.
