Material: N for B. The Black Qe4 has one legal move, to g4. The Black Re7 and Ba6 are both unprotected, and the White Qc6 is pressing Black's Q-side. Both sides need to improve their exceptionally poor piece coordination. The salient feature of the position, however, is the White Rf6, which at first glance is playing Don Quixote, tilting at Black K-side, supported only by Sancho Panza and a mule, the Pe5 and Pg5.

Candidates (30.): Rf4, Rg1, c4, g6

30.g6, threatening

31.Rxf7+ Ree8

[31...Rxf7 32.gxf7+ Kxf7 [else, 33.Qxe6] 33.Ng5+, winning the Qe4].

32.gxh7+ Qxh7 33.Rg1,

and White has a raging attack with a P up.

Black cannot play 30...fxg6 or 30...hxg6, which block the Qe4's escape square on 31.Rf4. Acceptance of the Rf6 sacrifice loses:

30...gxf6 31.exf6 (threatening 32.fxe7 and e8=Q, winning 2Rs for Rf6)

31...Re8 32.gxf7+, and the threatened fork 33.Ng5 wins 2Rs for Rf6 again.

Black must therefore refuse the Rf6 sacrifice.

30...Rf8 31.Rxf7 (threatening 32.Rxf8+ Kxf8 33.Qa8+ Kf7 [or Re8] 34.Qxa6)

In the threat, White takes Rf8, not Re7, so that Black cannot counterattack along the f-file.

31...Rfxf7 32.gxf7+ Rxf7 (to maintain the counterattack along the f-file)

33.Qe8+, and the threat of

34.Qxf7+ Kxf7 35.Ng5+ and 36.Nxe4

leaves a placid position, with White an exchange up against a bad B.

Time to peek. Timman's solution was much more elegant, and I should have explored the vulnerability of the Black Qe4 more than I did. No doubt the variation forcing Geller's 31...fxg6 has been much discussed. Time to check the kibitzing.

Fascinating. Although I took a little longer calculating than Geller (probably :), I also "panicked" with 30...Rf8. Although I overlooked <MAJ>'s 30.h6 as an alternative refusal of the sacrifice Rf6, it is completely logical: it eliminates all tactical strokes based on the fork Ng5+. Likewise <MAJ>'s 31...Rd7 after accepting the sacrifice Rf6 looks as though it just throws a R, whereas it avoids a later fork Ng5+ after 32.gxf7+ in one of my lines.

<<MostlyAverageJoe> wrote: ...CG unleashed it as a puzzle to confuse everybody. I claim bogus...>

I guess we pieces of wetware just do not compete with <MAJ>'s hardware. His computer analysis looks fine (of course). Personally, I like the idea of an occasional "bogus" puzzle. I am not proud, I like having to be on my toes, and the closer the puzzles are to game situations, the more realistic everyone's answers have to be.

Watto came a gutser there old pal.

<<sombreronegro> wrote: I think 31...Qxg6 is playable.>

Definitely not. See <Ron>'s post, the first one on the game.

<<Samagonka> wrote: Any better chance for Black winning, had he captured the Rook with 30.gxf6?>

<MostlyAverageJoe>'s first post shows that with best play, White still has some pull after 30...gxf6. I give several lines after 30...gxf6, leaving the strong impression that accepting the sacrifice loses for Black. Geller (and I) missed 31...Rd7 in <MostlyAverageJoe>'s line 2, "panicking" and rejecting 30...gxf6 for 30...Rf8, because Geller (and I) also missed <MostlyAverageJoe>'s line 1 30...h6. So, the answer is "No".

<<JG27Pyth> wrote: Is that an type of "interference">

I was not sure, so I checked. CG's scripts do not permit me to give the correct link, but you can find (with <CTRL>-F) the correct link on this page

http://encyclopedia.thefreedictiona...

by searching for the word "chess".

Actually, <RandomVisitor> was the first one to post that line (and 30...gxf6 as well). While his Rybka's and my Hiarcs' lines differ somewhat, it is fairly clear that white's win is far from obvious.

<Occasional bogus puzzles also remove the need for <MAJ>'s unlabeled dual puzzles (one real, one bogus). CG is just taking your suggestion, <MAJ>, just spreading the bogus puzzles out over a longer time.>

Thanks, <MAJ>, I missed <RandomVisitor>'s post. I agree, White seems to have pull, but not a won game, in both lines, accepting or rejecting the sacrifice Rf6. It's interesting that humans (i.e., Geller and all of us posting today) overlooked both drawing lines.

<<MostlyAverageJoe> wrote: The occasional spoiler puzzles are normal (see Daily Puzzle F.A.Q., at the end of the "What kind of move am I looking for?" section), but too infrequent, IMHO.>

Yes, that's the concept I was trying to put into words -- I think perhaps this is sometimes called, informally at least, 'suffocation' -- as it turns out though, in this puzzle the escape square isn't really removed anyway.

Toga II 1.31. gives

[ply 15/44, time 01:04, value -0.49]

30.Rg1 Bc4 31.Qa4 c6 32.Rf4 Qg6 33.Rg2 Ra7 34.Kg1 b5 35.Qa3 b4 36.Qa4 Bb5 37.Qb3 Qh5

[ply 15/64, time 03:03, value +0.86]

30.g6 gxf6 31.exf6 Rd7 32.gxh7+ Qxh7 33.Qxd7 Kh8 34.h4 Qg6 35.Ng5 Qxf6+ 36.Kg3 Rg8 37.Qxc7 Be2 38.Qf4 Kg7 39.Kf2 Qxf4+ 40.exf4 Bh5 41.Rg1 Bg6 42.Nf3

Note that the ends of the lines become inaccurate. The valuation shows that 30.Rg1 is worth about 1.35 P less than 30.g6.

Note that Toga is freeware. My chessforum contains instructions for downloading everything you need to do your own analysis.

30.g6 is a wonderful move which threatens to ruin the Black position. 30...hxg6 loses to 31.Rf4 when the queen is trapped while 30...fxg6 ruins the pawn structure by leaving e6 undefended.

are you an old man who smokes cigars? It's a little saying we have over here in Denmark, which I believe originates from Bent Larsen. The London, Torre, Veresov and Colle/Zukertort systems have a certain degree of popularity at club level here and there is more than a little truth that the older generation play it a lot. (They are decent enough lines, of course, as Kovacevic and Jussupow have shown countless times.) Only the youngsters play the Trompowsky!

I have never really liked the positions with ...d7-d5 in these lines, not that it is bad for black or else no-one would play the Queen's gambit. There is just a certain rigidity to black's structure I can't stand. Two sterling, classic examples of the approach I like are:

Colle vs Capablanca, 1929

oh heck, I don't play the old cigar smoker lines in the Queen Pawn -- I face them as black.