1

I am Black and my opponent is White. Here the computer recommends Rxd4, which I have done (the rook captured bishop here).

[FEN "4r1k1/1pp2pp1/p6p/4q3/P2rPn2/2P1QP1P/1P4P1/R4RK1 w - - 0 1"]

It's obvious then that Qxd4 is a blunder, because then I play Ne2 to fork the king and queen and the game is over. I would expect the computer to play cxd4 here to go up material, but the computer plays Rfe1.

Why not cxd4 here? What future sequence am I not seeing?

4
  • 4
    My first instinct would be cxd4 Qg5, threatening Qxg2# and Nxh3+ (the latter winning the queen). Not computer checked though.
    – koedem
    Commented Jan 13, 2021 at 22:13
  • 4
    I checked SF12 and it thinks cxd is best for White. Still lost after ...Qg5 though
    – B.Swan
    Commented Jan 13, 2021 at 22:21
  • SF12 is the open source engine Stockfish.
    – Timo
    Commented Jan 15, 2021 at 14:40
  • @koedem: I think your instinct is right, I would only check a second if White has anything (Rf2, Qf2, Qd2 - nope, the only move is Kf2, but then Qxg2 followed by Qxb2, curtains.) Qg5 is usually a motive against an undefended Qd2, but here it works wonders too. Commented Feb 14, 2021 at 20:00

1 Answer 1

3

Stockfish, at depth 20, evaluates Rfe1 with a score of -8.5, while cxd4 has one of -2.6. With so large a difference, the explanation is that you were playing against a weak computer. There is usually a gradient (e.g. from 1 to 10) of difficulty to choose from when playing against a computer, so you have probably chosen something far from 10. Chess.com (guessing from the screenshot) also has adaptative opponents that will grow stronger/weaker according to your performance.

For what it's worth, shortly after cxd4 White gives up a rook and a pawn for Black's knight and has an exposed king and piteous pawn structure —so Black wins anyway—, as follows.

4r1k1/1pp2pp1/p6p/4q3/P2rPn2/2P1QP1P/1P4P1/R4RK1 w - - 0 6

1. cxd4 Qg5 2. Kf2 Qxg2+ 3. Ke1 Qxb2 {Double threat: Qxa1+ and Ng2+ (a royal fork)} 4. Qxf4 Qxa1+ 5. Ke2 Qxd4

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.