Yes, gxh4 does. Why don't we start two continuations of the game against each other. You play black against gxh4, and I will play black against Qd6. See who wins faster. By CG rules, this should be done on my forum, not here.

After Nf4 and the ensuing exchanges in which black also has to give up the knight, both sides have a rook, but in this position, white's rook is locked IN FRONT of the pawn wall behind black's position, yet black's rook is outside the pawn wall, and black's king is safe on h6. Black has play! So in this position, white is up a full piece (the knight), but there are nuances which cant be calculated by computers or explained in numbers. For example, 35.Nh4 hxg3 36.hxg3 a5 37.Kg2 axb4 38.axb4 and black must give up the knight for any type of play. (CM says blacks best move is 38...Rb8)39.Qxe7+ Qxe7 40.Rxe7+ Kh6

Incidentally, after all of this stemming from Nh4, white can STILL play Qd6! after any rook move other than Rd8, which is winning as well. For example 38...Rd8 39.Qd6! Qxd6 40.cxd6 and white will queen the pawn.

Karpov cuts through all of this and finds the best winning move, Qd6!

35...♕xd6 36 cxd6 wins the pinned knight---or

The final statement is of course the point. Humans are fallible. (Fear anyone who claims otherwise.) Let's model human play as computer play with an occasional random suboptimal move, with suboptimality increasing with positional complexity. Who would deliberately risk being struck by a random lightning bolt after 35.Nf4 or 35.Ne5 when 35.Qd6 is available?

Qd6 is a good move (good enough for an easy win), but it is not the best move.

Dynamic Evaluation: Checking the Black King with Qe5 removes both Queens from the board after pd4xe5. That should be enough for this puzzle.

PM+

No, I was wrong, on that line 36. ...h4xg3 gives Black a -0.03 advantage.

On the given line, 36 ...Kf7 gives White an advantage of _3.82.

To answer my own question: not even Karpov (granted, at 11 years old). Many people posting, rightly or wrongly, are expressing greater confidence in their own chessplaying abilities than Karpov did.

"A man's gotta know his limitations." - Dirty Harry

Nf4 and Ne5 both look like great squares for the knight to move to but niether look forcing(if white didn't have his queen on e6 then Nf4 would create the immidiate threat Ne6+, but there appears to be no way to explot this)

No rook moves look good and white has virtually no good pawn moves

Also, it may have an application to real life where there are often different ways to solve problems, which allow for the diversity, creativity and uniqueness that makes life more interesting and enjoyable.

The last quoted statement (where I used <<>> for extra emphasis) is incorrect.

The most forcing variants after gxh4 do not include Qd6, therefore the possibility of Qd6 is irrelevant and has no effect on the score.

Your proposed continuation <Kf8 Qf6+> (this forces black to play Kg8) <Qd6>, is slower to win than playing h5 instead of Qd6.

I posted 7 lines with higher evaluations than 35.Qd6, and only one of them (35. Nf4) includes Qd6 in the mainline.

<znprdx: <MostlyAverageJoe: ....Why don't we start two continuations....> Listen if you say gxh4 does it all...that is good enough for me....it is in effect a zugszwang. If you found this right off – that is fantastic>

Frankly, I *am* peeved at myself that I did not look at Qd6. I just got lucky with gxh4. Looking at the game, I was ready to despair that I missed an easy move, but then Hiarcs supported me. It is not frequent that one finds a move stronger than Karpov did, even if he was 11, so there's no way I can let it go without a fight to prove my line to be superior :-)