Chess Evolution: The French Defence (part four)

EVEN more eccentric-looking than 2.f4 as an answer to the French is Steinitz's 2.e5!?
This is sometimes called the Steinitz Variation, but that invites confusion with the much better-known 1.e4 e6 2.d4 d5 3.Nc3 Nf6 4.e5.
Although 2.e5!? looks eccentric, it served Steinitz well - he scored 79% with the move, according to ChessBase's 2020 Mega database, outperforming his results with 2.d4 by five percentage points.
Wilhelm Steinitz - Szymon Winawer
Vienna 1882
French 2.e5
1.e4 e6 2.e5!?
This game was played in round 12. Winawer was so impressed by Steinitz's novelty, which Steinitz had unveiled in a round-seven win against Bernhard Fleissig, that 18 rounds later - it was a 34-round tournament - Winawer used 2.e5!? to beat Adolf Schwarz.
2...c5
Steinitz and Winawer tied for first in the tournament. They had a play-off match, the first game of which saw Winawer innovate with 2...f6!? 3.d4 c5 4.dxc5!? Bxc5, winning in 27 moves (Steinitz won the second play-off game, after which the players agreed to share the first and second prizes).
3.f4 Nc6
Steinitz - Max Weiss in round 11 saw 3...d5 4.exd6 Bxd6 5.g3 Bd7 6.Nf3 Bc6 with a position that looks at least fine for Black (but 1-0, 35 moves).
4.Nf3 f6
In round 12 James Mason played 4...Nh6 and drew against Bernhard Fleissig, another player apparently impressed by Steinitz's 2.e5!?
5.exf6?!
This seems rather cooperative. After ...
5...Nxf6
... Black has a central pawn-majority and an edge in development.
6.g3 Be7 7.Bg2 Qc7 8.0-0

Position after 8.0-0

8...0-0
8...b6 9.Na3 a6 10.Nc4 Bb7 11.b3 0-0 12.Bb2 b5 13.Ne3 was roughly equal in Steinitz - Celso Golmayo Zúpide, Match (Havana, Cuba) 1883, according to the analysis engines Stockfish11 and Komodo11.01 (but 1-0, 67 moves).
9.Nc3 a6 10.b3 b5 11.Bb2 Bb7 12.d3 Nd8!?
A little unambitious. The engines reckon Black has a good game after 12...Nd4.
13.Qe2 Nf7 14.Nd1!?
Another somewhat-strange knight retreat. There seems nothing wrong with the natural 14.Ne4.
14...Rae8 15.Ne3 Bd8 16.c4 Bc6 17.cxb5 Bxb5 18.a4!?
This is not liked by the engines - presumably because it weakens b3.
18...Bc6 19.Rfc1!?
Wrong rook? Maybe.
19...Qb7!? 20.Nc4!?
Black gets an initiative after 20.Rxc5 Bb6.
20...Bc7 21.Ba3 d6 22.Rab1 a5 23.d4?!
White gets rid of his isolated, but not very weak, d pawn, but the move seems premature.
23...cxd4 24.Kh1
It is possible Stenitz had missed that White is in serious trouble after 24.Nxd4? Bxg2 25.Qxg2 Qa7 with ...e5 to follow.
24...d3!?
The simple 24...e5 may be stronger, but the position is very sharp.
25.Qf1?
White had to play 15.Qd3 Be4 16.Qe2 Bxb1 17.Rxb1, when White's bishop-pair and better-coordinated pieces give decent compensation for the exchange.
25...Ne4?!
Black has an excellent position after 25...e5, eg 26.fxe5 Ng4 27.exd6 Nxd6! with strong play against the white king as 28.Nxd6 Nxd6 29.Bxd6 runs into 29...Rxf3! 30.Bxf3 Bxf3+ 31.Kg1 Qd5.
26.Nd4 Bd5 27.Rb2 e5 28.Nb5 Re6??
The engines' 28...exf4 29.gxf4 Nc5 seems roughly level, but still complicated.
29.Qxd3 Nxg3+ 0-1
Winawer resigned before Steinitz could decide whether to go a piece up by 30.hxg3 or 30.Qxg3. There was no defence, eg 29...Bc6 loses to 30.Nxc7 Qxc7 31.Bxe4.