PawnSac: well, better than his 9.Ng5 but still worse than black
< 7.e3?! > a3
< 8.ed? > Bb2
< 9.Ng5/ > good grief he is going down the exchange already.
28/44 -1.26 9.d4-d5 exf3 Qxf3. With Ng5 white's eval drops to
the board after < 13.ab3 > looks absolutely horrible for white.
the only hope might lay in a d pawn march (see the note above
with 9.d4-d5 ). He will have a difficult time developing his Q knight,
which shows the positional plus of the dark pawn on b4, and why a3 is necessary in this line.
< 9...Bf5 > stockfish gives ..Be7 Nxe4 Nxe4 Bxe4 0-0
with a clean +2.14 @ 35/51+. Of course the computer sees farther
out over the horizon than a human player and apparently
Surgey thought it was clearer to maintain the advantage,
restrain the d pawn, and pressure white's weaknesses with
..Bf5. Fair enough. No criticism here.
< 14.Bf6? > stockfish gave d2-d3 with about -2.0 for white but after Bf6
his eval immediately drops another point ( -3.15 @ 30/45 )
< 16. ..Kc8 > parrying a potential N check? a waiting move?
SF gives ..Be7 +3.37 @ 29/47+
< 17...Bg6 > stockfish simply takes the pawn with Bxg4 +3.78 @ 32/54
< 21.Rhd8 > stockfish gives ..Rd3 +4.78 @32/58+ with a long line in
which black maintains the exchange + pawn but i'm not convinced it
is better, pragmatically speaking, than what Sergey played.