lichess.org

What's up with the computer analysis of MVL vs Aronian?

lichess.org/broadcast/2019-fide-world-cup-round-5-tiebreaks-game-2/fh9o8oMq

At move 30, after Aronian played ...Qxe3+ sf says avantage -5.8 and best move for white is 31.Kh2. MVL actually played this move, and then the computer evaluation becomes -0.4. This is already very strange, but what's more, sf says the best move for black is 31...Nxd3, and from there the advantage seems to stay around 0. But Aronian actually played 31...Qe2+, and then the advantage is back at -5.2.

What's going on? It's like after move 30 sf saw the right continuation but then it forgot it at move 31 and was not able to find it any more!?

Huh, are you using a different engine? Here it says winning all that way through. (well, until 32...Nxd3 anyway)

Well what I describe happens with stockfish on the Lichess analysis board.

@TeddyTheTurkey #1
Movement 31 ... Ne4 leads to Black's victory. Move 31. ...Qe2+ leads to the repetition of moves and still does not spoil anything. Move 32 ... Nd3 - mistake. It was necessary to repeat position 32. Kg1 Qe3+ 33. Kh2, and make winning move 33. ... Ne4!

@ujcn ok but what I don't understand is why stockfish suddenly says only -0.4 after 31 Kh2.

1. Change the number of analysis lines to 1 in the hamburger menu
2. Reload the page (there's a bug that prevents cloud evals from being re-fetched otherwise)
3. Go to move 30.

After white's 30th move, the evaluation is -4.9, depth 26, best move 30...Qxe3+
After 30...Qxe3+, the evaluation is -5.8, depth 26, best move 31. Kh2
After 31. Kh2, the evaluation is -0.4, depth 38, best move 31...Nxd3. I don't know why it doesn't see Ne4, especially with the higher depth.

This topic has been archived and can no longer be replied to.

Reconnecting