17 Qxf6 ? gives a tempo to black for the better IMO 17 .. ♕e7.

To me it should be

"Who says this might be too much"

maybe they were nearing time control, adding more time to their clocks after a certain number of moves. This is a modern game after all.

Or, white was trying to lull black into THINKING that he was just playing for a draw, all the while white is seeing clear forced lines.

It doesnt even look at Nf6+, and when I force it to it gives white only +0.03, Nf6+ is a pretty bad move

CM lets out a condescending sigh and we check out Nf6+. In that line, another interesting point is that CM throws in the in-between move 17.Qg6+ Kh8 and THEN Qxf6+, giving white a tempo.

<aazqua: This is absolutely trivial, but white's play following the obvious knight sack is horrendous. 17qf6??? Uh, how about qg6 ch FIRST and then qf6 ch retaining the tempo?>

14 ♘f6+ <forks ♔ and ♕>

14...gxf6 <wrecks Black's ♙ structure and exposes the ♔ to frontal attack.>

15 ♕g6+ <takes advantage of the pin on the f7♙>

15...♔h8 16 ♕xh6+ ♔g8

<and the following is pure speculation by me.>

17 ♗d3 f5 18 ♕g5+ ♔h8 19 ♕f6+ ♔h7 20 ♗xf5+ ♗xf5 21 ♕xf5+ ♔g7

let's look now

17 Qg6+ Kh8 18 Qxf6+

to retain the tempo and prevent 17... e4! that would derail the white's attack after 17.Qxf6.

<Nostrils: ... In the game line 17 .. kh7 doesn't make sense. Makes we wonder if the script is wrong.>

14.Nf6 gxf6 15.Qg6+ Kh8 16.Qxf6+ Kg8 (Kh7) 17.e4

Now, I have to admit that I missed the idea of Qe7. Of course my (16....Kh7) would lead to Qxh6+, but now it's less clear, but the Bishop enters via g5! (18.Qxh6 with 19.Bg5; Black still can't play f6!) and it's . Any other reply seems to lose because the Rook can enter via d3 (that I noticed).

Indeed, isn't it annoying when people don't read the earlier posts and attribute the discovery of the <17 Qg6+ Kh8 18 Qxf6+> line to someone who was the seventh person to post it, instead to at least noticing the sixth post of the same line :-)

I just hope I got the count right ...

Analyzing this game can give one serious doubts about computer's ability to pick the best moves even in the very forcing lines. Descending up-and down the variants changes the best choices way too frequently, even with >16 plies per step. But I wrote it all up (this is the last paragraph I am typing), so might as well post it.

Moves 14. Nf6+ gxf6 15. Qg6+ Kh8 are either completely forced (for black) or undoubtedly best choices (for white).

Now white can play as in the game:

16. Qxh6+ Kg8 (+7.39)

or a most likely slightly weaker, but still winning line shown by <TheaN>:

16.Qxf6+ Kh7 17.e4 (+4.34)

<<<At move 17 there are two winning moves for white: 17.Qg6+ and 17.e4, and the latter is BETTER - not by much, but a bit better>>>. Come to think of it, this is where I spent most of my time when coming up with my very first post -- I kept considering whether to take the f6 pawn or leave it be, so that it would prevent black Q from getting easy access to the action -- the untouched f6 pawn would prevent f7-f6. OTB, this also would've been a point to rethink the next couple of moves.

After 17.e4, black has a number of similar-quality moves and the analysis becomes really complex. Here are the four best lines as displayed by Hiarcs (15-ply deep):

(+7.39) 17... exd4 18. Qg6 Kh8 19. Qxf6 Kh7 20. Qh4 Kg8 ...

(+8.00) 17... Ne7 18. Rd3 exd4 19. f4 Bg4 20. Rg3 Qd7 ...

(+8.15) 17... Qd7 18. Qg6 Kh8 19. Qxf6 Kh7 20. Qh6 Kg8 ...

(+8.50) 17... Be6 18. Rd3 exd4 19. f4 Ne7 20. Bxe6 fxe6 ...

It is interesting that up to 14 plies, these lines evaluate within 0.25 pawn of each other. OTB, playing 17.e4 could be a way to win on time by having the opponent think about too many variants :-)

<<<OK, back to the game as it was played (if the record is correct) 16. Qxh6+ Kg8 >>>

As noticed by many, 17.Qxf6 is a bad blunder. The best response for the black is, indeed, 17...e4, leading to something like: 18. Qg6+ Kh8 19. Qh6+ Kg8 20. f3 Be6 21. d5 Ne5 22. Qg5+ Kh7 23. Qh4+ Kg7 24. dxe6 Nxc4 25. b3 Ne5 26. Bb2 f6 which Hiarcs evaluates as a dead draw (0.00) in this position:

click for larger view

During the analysis of the line above, evaluation for white never exceeded 0.50, and any exploration of side lines ended in draws as well. Unless I missed something, it is thus quite likely that

    In the line played (allegedly) in the game, 17 ... e4 WOULD END IN A DRAW.

<<<Back to the game again, after 16. Qxh6+ Kg8>>>

The second best continuation after 16 moves is as proposed by many (and not what was recorded in the game):

17. Qg6+ Kh8 18. Qxf6+

and now black has two responses.

18 ... Kg8 19. e4 exd4 20. Qg5+ Kh7 21. Qh4+ Kg8 22. Rd3 (+6.94)

18 ... Kh7 (transposes to the same line as above).

But at least your post has been very much read by yours truly MAJ, and duly acknowledged. The acknowledgment is between extra <<<<>>>> below:

<MostlyAverageJoe: As noticed by <dzechiel>, myself, <Nostrils>, and a <<<<<couple more people>>>>>>

:-)

Okay, so here's the complete list of people who wanted to play 17 Qg6+ Kh8 18 Qxf6+, or noticed the 17. Qxf6?? Kh7?? blunders, or both: <dzechiel>, <chessmoron>, <Amulet>, <MAJ>, <Alphastar>, <Nostrils>, <ConstantImprovement>, <realbrob>, <Marmot PFL>, <psmith>, <znprdx>, <LivBlockade>, <eblunt>, <aazqua>.

Yup, I did mis-count. And wouldn't be surprised if I still missed someone. Oh, well, good job, everyone!