And, if I may make a note to <Chessgames>, it also seems to me that the jump in difficulty from Thursday to Friday tends to be too great. Thursday is always a puzzle where it can be a challenge, you might have to get creative and clever, but there is a clean set of variations that demonstrate the win (or draw). Lately it seems Friday has been thrusting us straight into the realms of major calculating complexity. I would prefer to see on Friday one more puzzle with a fairly clean solution. Ideally I think it would require more ingenuity and maybe a bit deeper calculation than a Thursday. I'm happy for it to be peppered with side variations and multiple branches so long as it doesn't get really messy. Then save the really complex ones with tonnes of branching variations for the weekend.

1: Nigel Short - Lev Psakhis, Banja-Luka 1985

click for larger view

Analysis by Rybka 2.3 LK mp 32-bit (23-ply):

1. (3.57): 33.Nxc4 Rd7 34.Nd6+ Rxd6 35.exd6 Nxc3 36.Rxc3 Qd5 37.d7+

2. ± (1.25): 33.Ng4 Qe2 34.Re1 Qd3 35.Nf6+ Ke7 36.Re3 Qf5 37.g4 Qg5 38.Qxg5 hxg5 39.Kg2 Bg7

3. ± (1.25): 33.Bd2 Qxe5 34.Nxc4 Qg7 35.Qxg7 Bxg7 36.Rb1 Rd7 37.Be3 Rd5

4. ± (1.10): 33.Rd1 Nxc3 34.Rd8+ (22-ply)

Your argument is premised on the notion of black's rook having a 'second life' after sacrificing itself on a3!!!

In your line after 36...Nd7, one simple way to win is:

<37.Bxf8 Nxf8 38.h6 Ke7>

Unpinning now is necessary to stop white getting a second queen. 38...Rxa3 can be ignored with 39.h7 and queens

<39.h7> and black has to give up his knight for the pawn .

<34...Nxb2> does make good sense. It tempts white into taking a poisoned queen (see my previous analysis). Alternatively, if white goes ahead with 35.Bxh6, the white b-pawn is removed. Presumably white wins amongst the thickets of variations, but at least the water is a bit muddy due to the advanced black passed b-pawn.

On black's previous move, 32...Ke7 undoes the pin and keeps the match even.

click for larger view

Then I saw Nxc4 and thought everything fell into place.

The variation I counted on:
33. Nxc4 Qxc4
34. Bd2 Q-moves
35. Bxh6 with Qxf8, xd7, xR to follow.

I did not see 34. ... Nxb2. Perhaps a critical mistake. Upon reaching that point in replaying the game, I did think, go ahead with Bxh6, but I cannot say what I would have chosen if unexpectedly faced with the "free" queen in the game.

However, UdayanOwen, or MAJ or anyone else who can lead me to the light:

Is 35. Rxc4 Nxc4 losing for white?

UdayanOwen then gives 36. Bxh6 b2 37. Qh7(why?) Rxa3 as losing for white, which it is. But why not this:

36. Bxh6 b2
37. Qxf8+ Kd7
38. Qxf7+ Kc6
39. QxR b1=Q+
40. Kh2!

While this is much, much messier than simply 35. Bxh6, it hardly looks losing for white. Black may be able to weasel a draw out of it, but it certainly is not losing, is it?

While this is much, much messier than simply 35. Bxh6, it hardly looks losing for white. Black may be able to weasel a draw out of it, but it certainly is not losing, is it?>

The answer to your (why?) question about 37.Qh7 is obvious.... It is motivated by preventing black from queening.

Of course, as you reveal, there is a flaw in my analysis. Again, I have succumbed to chess blindness - I should not have automatically covered the queening square since white has a check and this should be followed up... and as it turns out, there is another check, then a rook win, and who cares if he queens cos we've won so much stuff that we come out with a winning endgame anyway........

Thanks for your response. Just note, based on your comments here, you are a much stronger player than am I. I noticed later that Qh7 covered the queening square. I meant absolutely no offense.

I really, truly, just wanted to make sure that I was not missing something obvious.

Thanks also to Magic Castle.

None taken. I like getting tripped up in my analyses. Getting things wrong is as much a part of the art as getting things right, as this is how one hones their craft.

I think your solution is from one angle a bit superficial. There are many queen moves on move <34...> and 3 ways I can see to maneuver the queen to defend the a7 rook prior to the skewer. In many variations the b2 pawn disappears, creating an advanced passed pawn for black. One could argue that a solution should provide some concrete analysis to demonstrate that in the different variations black could play, this counterplay is nothing to worry about.

forward to: TPOTD

Black is well into white's territory.White has the black bishop pinned to the King by the Queen. Could get an attack on f8 from h6. Let's try,

33.Bd2...Qd3 34.Nxc4 ...Qxc4 35.Bxh6...I wish

PM=

Yes of course, I missed the obvious 33 ...Qxe5

Still the idea was right, but the moves were in the wrong order.

Brain L0.15 R 0.75

Material: even. The White Qg8 pins the Bf8 to Ke8, and Pe5 forms a wedge into weak dark squares around Ke8. The White pieces are somewhat uncoordinated and need to be activated. The Rc1 should be on an open file, and Ne3 should be at d6 or f6. Close examination indicates that Ne3 does not generate critical threats at f6, but does at d6. The Bc3 is bad, except for the support it lends e5 in threatening Black's weak dark-square complex. The Black Na4 is out of play for defense of Ke8, but it and the queening of Pb3 pose a significant counter-threat.

Candidates (33.): Nxc4

33.Nxc4, threatening

34.Nd6+ Ke2, then

35.Nxe4 [also, Qxf7], or

35.Nc8+ Ke8 36.Nxa7.

If Black refuses the passive sacrifice of Nc4, White has won Pc4 outright and proceeds to plant Nc4 at d6 like a house, with all the threats described above.

34.Qxc4 Bd2 (threatening 35.Rxc4 or 35.Bxh6 36.Qxf8+)

The Black Qc4 can retreat or counterattack by taking Rc1. If Qc4 retreats, Black must maintain control of the d-file, to prevent Rc1 from joining the attack on Ke8 along the d-file. In case of either retreat or counterattack, Pe5 remains to support the attack on the Black K-side, the second threat above wins Bf8, and Black has insufficient time to generate adequate counterplay with Pb3.

Time to peek. I am surprised that the threats were enough to induce immediate resignation, but Psakhis obviously has more confidence in my pessimistic judgment of his final position than I did. Time to check the kibitzing. It looks interesting.

As CG points out, you do get better with practice...

<<JG27Pyth> wrote: Arrgh! [snip] but I thought Ng4 [snip]>

If you want company, <JG27Pyth>, I almost started looking seriously at 33.Ng4. The only advantage we humble pieces of wetware possess over the machines, however, is our ability to pick out "important" points in the position. In the present case, "Close examination indicates that Ne3 does not generate critical threats at f6, but does at d6." The only useful candidate is therefore 33.Nxc4, not 33.Ng4.

To take a (small) page out of <UdayanOwen>'s book, I recall a study on GMs' eye movements that said they generally look over everything on the board for a couple of minutes, and then focus intensely on a single area. For what it is worth, my initial overall evaluation of the board has become very valuable to me, and certainly, inculcated as a good habit, has permitted me to locate correct candidates more dependably than before.

[ply 15/48, time 00:53, value +0.50]

32.Qg8 Ke7 33.Ng4 Nd5 34.Bd2 Ra8 35.Re1 Qf5 36.Bc1 c3 37.Nxh6 Qxh5 38.Bg5+ Ke8 39.g4 Qg6 40.Qxg6 fxg6