Material: B for N. Black has a battery Rf8 and Qf6 attacking Pf2. The White Qh6 has 1 flight square h3. The Black Nd4 can deliver check at f3 and attack Qh6 at f5. The White Be4 is unprotected. Only the Black Rd8 needs activation. The White battery Rd1 and Rd2 threaten Nd4, which is protected once, although Black gets some (insufficient) compensation with Qxf2+. The key to the White defense, however, is Be4, which guards the long diagonal a8-h1 against Qf3+, and more immediately, guards f3 against Nf3+, winning the exchange. Checks and captures go nowhere, so we are left with threats for candidates.

Candidates (29…): Rde8

29…Rde8 (threatening 30…Rxe4 and 30…Nf3+)

If White takes Nd4, there is a mate.

30.Rxd4 Qxf2+ 31.Kh1 Rxe4

threatening 32…Rxd4 and 32…Re1+ 33.Rxe1 Qxe1+ 34.Kg2 Qf1#.

32.Rxe4 Qf3+ 33.Kg1 Qxd1+ 34.Kg2 Qf1#

White must therefore play 30.Bg2, because 30.Bh1 Ne2+ is infeasible.

30.Bg2

I now went for 30…Re2, which gives White pull, but which Toga II 1.3.1 values at about 0.6P (some pull, but insufficient to win). <dzechiel> is quite correct about 31…Nb3 32…Qf5 forcing Rd2 away from the protection of Pf2.

Maybe he peeked at the solution ;>)

Man, this is a monster puzzle... and I was angry about it at first thinking it isn't a friday... and it surely <isn't> a friday if you're supposed to take this thing to the conclusion... anyone who "got" this to the end is literally playing world champion caliber chess: get thee some sponsors and go pro. I looked at this with Toga II from the puzzle move on and neither Tal nor Agzamov make a bad move -- 99% of it is Toga's first choice on both sides...an awesome display of GM strength.

<Dzechiel: Never considered 30...Nb3, and I looked at that position a long time. I'm still not sure I understand it completely.>

I'm not claiming to understand anything completely... but I forgive CG for posting this puzzle today ;) because I actually think I got a glimmer into where good chess comes from. This puzzle _isn't_ about deep calculation! They joke about Petrosian only seeing two moves ahead but seeing everything that matters. I think this problem is an example of that kind of seeing. Difficult as it is, it is all about seeing the position clearly:

Both Tal and Agzamov see with great clarity that Qxf2+ is a killing blow. If you really see this, then it becomes much easier to see that the white Rd2 is tied to to the defense of f2, pinned to the defense as it were... and that's the tactical lens that let's Tal find Nb3. It's NOT super deep (although who knows from how far back Tal saw these tactics!), IMO at least this puzzle is just a very clear sighted use of the position directly at hand.

This is just to get the puzzle as far as 32.Rxd2, which would be full credit IMO... beyond that it is, as someone said, "guess the move" they are just playing great chess with Tal slowly leveraging his up-the-exchange edge (the position evaluates for many moves at around -2.0, which is basically Tal's raw material advantage.)

Great puzzle.

The only place the White Rook can move is 31. Rc2, and then Black plays 31. ... Qf5. Please see <dzechiel>'s earlier post for the rest of the answer to your question.

Hi, <iqbalianpawn>, what is implied is that white cannot defend f2 without giving up an exchange (a rook for a knight), as is seen in the line you propose.

That rook can't move away from defense of the f2 pawn. Qxf2+ is deadly, I believe it forces mate... So Nb3 leaves the rook only one option... c2, and that's no option at all. As Dzechiel has noted, after Rc2, black's Qf5 would leave the rook no safe haven on the second rank. There's no sense in having the rook run, so by allowing NxR, white has RxR recapture and White sacs the exchange but keeps the f2 pawn guarded.

<iqbalianpawn What if 30...Nb3 31 Rc2 Qf5 32 Rdd2 Nxd2 33 Rxd2 And f2 is again defended..?>

This would seem to result in the same position that the game text results in -- black wins the exchange with f2 still defended, but actually there's a critical loss of tempo and this line loses quickly... the difference is in your line white doesn't get in h4 (which serves two functions... it sets up a pawn exchange that lets white bring his queen back into play with tempo, and critically, it makes luft for the king. In your line, after 33.Rd2, black has ...Re1+ 34.Bf1 Qb1 and I think the end is quite near for white. I didn't check that with the computer, though, so maybe black has some other devilish move there.

Black has two major pieces on the f file. The f3 square is guarded by the white bishop, while the f2 square is guarded by a rook on d2. Black also has a knight that could get to f3 or e2. The knight itself is under attack. It looks like white can capture it, but this capture also looks dangerous as it removes a defender (Rd2) from f2. An important, perhaps critical question, is whether or not this knight is immune from capture. Black also has a rook on d8. What can be done with that piece?

From white's perspective there are some serious looking weaknesses but also some threats. Depending on where the black Q moves, it might be possible, for instance, to sac the bishop on g6 with mate threats/perpetual check threats against black. White also has a Q that if attacked has only the h3 square to escape to. For the most part, the white queen is cut out of any action on the f file (cut off from all action in fact). On the other hand, despite her lack of mobility, it is not easy to attack the white queen. Black playing the N to f5 for instance might just allow a trade of the bishop for the knight with the result that N checks and/or forks on f3 and e2 evaporate with the disappearance of that piece. The same seems to go for an immediate Nf3+. It would accomplish a trade of minor pieces and a release of the white queen (who could capture the pawn on g5 after the black queen takes the B on f3).

So it looks important to not give up the fight to control the light squares around the white king. The black N looks immune to capture (this still needs to be tested) but immediate moves like Nf3, e2 or b3 look unconvincing. (The Nb3 move is interesting, but I don't think it forces white to move the rook off the 2nd rank. It also takes the knight away from where the action seems centred).

Black can try and target the pawn at d5 but that looks hard to attack (it could give white enough time to rearrange pieces to bettter squares). It was at this point in my thinking - looking for instance at moving the d8 rook to d7 or d6 that I started to think about how the bishop at e4 is unprotected. So in addition to white defenders doing a lot of work and being somewhat tied down, the important light squared bishop is not defended by any piece at this point in the game. Perhaps something obvious to others but to me it took awhile for this to sink in. Also if black could safely capture f2 with his queen (after the white rook captures the N), then shifting the black Q to e2 sets up possiblities such as bringing the black R to f2. Bringing the white R back to d2 at that point leaves the white bishop open to capture. Bringing the d1 rook to d2 allows the f8 rook to come to the first rank and that would be curtains for white.

So after looking at lots of possible first moves with knight and queen - here is the first candidate move with any real teeth:

29 ... R(d)e8
[black would be happy if white defended the B with a rook - a) 30. Re1 would be followed by black taking the bishop and then forking the king and d2 rook with the knight - or b) 30. Rd2 would be followed again by the capture of the bishop, then black could play Qxf2+ followed by Qf3+ and the removal through subsequent checks and captures of both white rooks]

30. Bg2
[moving the bishop off the h1-a8 diagonal allows a N fork on f3 and once black has traded his N for a white rook then black will have a queen and two rooks against a white B and R - the black queen is too cut off from play - and this should allow white to achieve a winning postion - it might also be possible and more effective to move 30 ... Ne2+ (see analysis below) wherever the bishop moves; alternatively 30. Bh1 although unlikely still needs investigation]

30 ...Ne2+
[now the black N cuts off the white d2 rook's defence of f2; moving the K to f1 allows mate with the queen; moving the K to h1 allows the same Qxf2 capture. Once that queen is planted on f2, then there are all kinds of nasty threats - for instance, the black N threatens to capture g3 with a check, then black can play Re1+, white takes that rook on e1, then, when black recaptures with the Q, white has only Bf1 with checkmate to follow - or if the B goes to f1 (i.e., after a black rook comes to e1), then black captures that bishop with the e1 rook and checkmate again follows. In other words, there might be some things that white can try, but I don't think anything will work if the white king moves to h1 and black is allowed to capture f2 with the queen. One possibility would be 31. Kh1 ... Qxf2, then 32. Qh3. Does this save white? I wish I had time to figure that one out.]

31. Rxe2
31 ...Rxe2 [again f2 will be targetted and it looks like there is no defence]

So what about the possible move 30. Bh1? Also what about moving the white queen down to h3 (after either Bg2 or Bh1)?

Too bad, I have now run out of time today to look at this some more. Maybe I really missed something obvious, but I am going to go with 29 ... R(d)e8. I think that the threat of Ne2+ (assuming that white moves the bishop to g2 or h1) is a very strong one, followed by black capturing on f2.

I should add one last thing however. If black plays his rook to e8 and white just goes ahead and captures the black knight with his rook, then black can capture f2 with his queen and that queen looks like it has a nice square on e2 (this allows for a black rook to come to f2 or f1 depending on white's response) and it also threatens the bishop on e4.

Seriously, for me Tal moves in the most unexpected ways. So already, I get queasy trying to solve this one. Been staring at it awhile and i've only found the most boring continuation.

Material: N for B. Black has connected pawns Q-side. White has passed d-pawn on 5th rank. White Q is nearly trapped; can only go to h3.

Ideas: Would like to trap the White Q. Would like to trade off the minor pieces. B is an advantage in an open game. Would like to win the d pawn.

29...Nf3+ (forking the d2 R, and forcing the capture.) 30.Bxf3 Qxf3

Now likely 31.Qh3 Rf5 (d-pawn should fall, and R interferes with the Q coming to play. Then simplification and the a & b pawns march down.

Time to see what i've missed.

What the? I considered 29...Re8, but felt there were too many responses. I also looked at Nb3...where is this going? Ok, I must be missing some mate threats or something here. Why couldn't White's R come off the 2nd rank? Or simply move to c2. Must read kibitzing.

Actually, had to throw this in with Fritz to see the continuation. 29...Rde8 30.Rxd4 and I saw 30...Qxf2+ 31.Kh1 but didn't see the continuation 31...Rxe4 which is crushing.

Wow. This was really tough. As soon as I saw Tal as Black, I figured I wouldn't get it. I feel even worse because I actually considered Nb3 to deflect the R but just couldn't see it in concert with everything else. I think the other thing that messed me up is that I gave up when looking at responses. I felt that there were "too many" responses to some moves like Rde8, when in fact, there are not.

First, my line above is a blunder due to: 29...Nf3+ 30.Bxf3 Qxf3 31.Qxg5 . Ouch, The d-pawn is passed and triple guarded, the White Q back in play and it's over.

I see now too that the Bishop cannot leave the a8-h1 diagonal due to Nf3+ winning the exchange. That seems simple enough. And the 30...Nb3 cannot be met with 31.Rc2 Qf5!

This position comes apart rather trivially when the threats appear. The f2 square is weak, and mate follows should the Re2 leave. The R has limited mobility, and Black can squeeze all it's potential squares away. And the N will force the exchange should the B leave the long diagonal.