1

I've had a few looks at the Traxler counterattack before. It seems fine for Black, but somehow the engine (Stockfish) says that White has a +2.57 advantage after Nxf7, but not much advantage after Bxf7+. Why is this so, when this opening is accepted by masters?

3
  • 3
    This could be a viable question, but you need to specify which line / which position (and which engine would be nice as well).
    – Glorfindel
    Aug 17 '20 at 15:19
  • 2
    I think this is answerable, so voting to reopen.
    – Allure
    Aug 18 '20 at 7:25
  • 2
    I've examined this from the white side, and I can say with absolute certainty that the Bxf7+ line is the line that you should play. The Nxf7 line is evaluated by Stockfish as winning initially, but when examined more deeply things get real murky down the line. Bxf7+ on the other hand is just a clear edge for white, even though it's not enough to claim that white is winning.
    – Scounged
    Aug 18 '20 at 9:14
2

I can't find it now, but I remember reading this exchange on Chessbase. A player just played ...e6 on one move, and then ...e5 on the next, seemingly losing a tempo for no reason. It was clearly computer preparation however. An interviewer asked GM Viswanathan Anand what he thought about the moves. Anand replied that if the computer thinks it's fine, then it's fine. Instead you have to analyze the lines afterwards to see why it's fine.

In other words: these days computers are so strong that their judgement trumps human judgement, including those made by world champion-caliber players.

Now the fact is that computers don't like the Traxler, even at high depth. It was played in the TCEC S17 superfinal between Leela and Stockfish. Here's the first game, and here's the reverse. Note:

  • These engines are playing on super hardware and achieve speeds and depths that would take very long for ordinary machines. When Leela played her fifth move, she was at depth 22/62. When Stockfish played its fifth move, it was at depth 43/62.
  • Both finalists thought 5.Bxf7+ is the best move.
  • Both finalists evaluated the position after 5.Bxf7+ as +1.5 for White (more precisely, Stockfish gave +1.5 for White; Leela thought White had a 60% chance to win).
  • Comparatively, in the starting position, White has roughly a +0.5 advantage.
  • There is some fluctuation in engine eval, so e.g. if Stockfish thinks a move that is +0.1 pawns better than the next-best move, the next-best move might still be playable. However, +1 is a pretty large gap.

Therefore the only conclusion here is that the human masters are wrong. The Traxler is objectively dubious at best. Playing this can work against humans because humans cannot find a string of "only moves" in a row to hold a position, plus it's psychologically difficult to defend against a strong attack for a long time. That's why humans - even top-level humans- often crack under pressure. The Traxler also offers plenty of opportunities for both sides, which is great if you want to play for a win and/or you think you are better than your opponent. But if you play the Traxler as Black in correspondence chess, you are asking for trouble. Maybe 32-piece tablebases will show that your position isn't lost, but you would have to fight hard for equality.

Finally I'll add that it's misleading to say the Traxler is well-regarded by masters:

  • Wikipedia's article on the Two Knights includes this quote: "No grandmasters have regularly adopted the Wilkes-Barre as Black, but Alexander Beliavsky and Alexei Shirov have played it occasionally even in top competition"
  • Chess Openings Explorer gives by far the most common response to 4.Ng5 as 4...d5 (which is also the main line of the Two Knights).

I'd call the Traxler playable in human games, but only as an occasional weapon.

2
  • 1
    Are you claiming that computers are better at everything than we are? Apparently Watson is an abysmal Stock Exchange investor. As an aside Beliavsky drew comfortably against Karpov following Bxf7.
    – Philip Roe
    Aug 18 '20 at 17:00
  • 1
    @PhilipRoe computers aren't better at everything, but they are better at almost everything that's chess-related. Beliavsky making a comfortable draw isn't sufficient evidence of viability - in the same way Miles beat Karpov with 1. e4 a6, and it's more likely that Karpov was caught by surprise than this is a good opening.
    – Allure
    Aug 19 '20 at 0:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy