<Kola> You're right, that was careless joke that started on the Kibitzer's Cafe. We just wanted to stress that accounts are being distributed on the basis of luck, not on the basis of who can and cannot afford it.

I have to wonder if even the great Tal saw all of the variations through to the end when he made the move 18... Nxg3. It seems more likely that he made this move based on speculation rather than a precise analysis of the position. With this in mind I went looking for the "best" line with the help of ChessMaster 7000 and found that both players made numerous mistakes during the ensuing skirmish.

First of all, 21... Re3 is a better move than 21... Re5. I think someone suggested this already in an earlier post. Here is a likely continuation: 21. Qxb7 Re3 22. Qg2 Qh4+
23. Kg1 Rg3
24. Nd5 Rxg2+
25. Kxg2 Qe4+
and we arrive at roughly the same configuration as with Re5 (assuming best play by both sides of course), but we get there a lot quicker and with far fewer complications (but perhaps it is the complications that Tal thrives on!)

Secondly, white's move 25.Qxa7 is almost a blunder, instead he should have played 25.Qf6 which still favors black, but there is no forced mate after this, and white can keep his chances alive. What they arrive at after the fireworks is a position where black has a pawn and a queen for a knight and a rook. Here is a possible continuation: 25. Qf6 Rg5+ 26. Kf2 Rf5+
27. Qxf5 gxf5
28. Nd5 Qh4+
29. Kf3 Kh8
30. Nf4 Qg4+
31. Ke3 Re8+
32. Kd2
and black has roughly a 2 point material advantage, but no dramatic winnning combination in sight and I like white's pawn structure compared to black's, and white should be able to make black's life miserable by harassing his queen with the two rooks.

Thirdly, black basically lets white off the hook following 25.Qxa7? with ... Qg5+?, a surprising lapse given the person pushing the black pieces. 25... Rh1+! was the best move in this position, as follows: 25. Qxa7? Rh1+!
26. Kg2 Rh2+
27. Kf3 (not Kf1, else Qh3, followed by Qg2# or Rh1#) 27. ... Re8
28. Qg1 Qh5+
29. Qg4 Rh3+
30. Kf2 Qxg4
31. Rg1 Rh2+
32. Kf1 Qf4+
33. Ke1 Qxc1+
34. Nd1 Rhxe2+
35. Kf1 Qxd1#

Finally, after Black lets white off the hook with 25... Qg5+, white still manages to lose in grand style, by playing 29.Qe7? instead of the more solid 29.Qe3. Here is a possible continuation based on this move: 25 through 28 are pretty much forced... 25. Qxa7 Qg5+
26. Kf2 Rh2+
27. Ke1 Qxc1+
28. Nd1 Qc2
and then instead of 29.Qe7, ...
29. Qe3 Qxa2
30. Qf3 Qa5+
31. Nc3
and White holds it together for a little while longer.

For all its flaws, this is certainly one of the most brilliant and beautiful combinations there are.

I respectfully disagree. In my early days of tournament chess, I fell in love with the idea of the dramatic sacrifice and often made speculative sacs in tournaments because it "looked right". It hardly ever worked out for me, especially if I was playing someone on the same level or better than me (once out of every 10 tries, maybe!), but when eventually I did start having some success with it, I was attempting them far less frequently and I noticed that I was almost subconsciously being more discriminating by forcing myself to visualize the resulting positions where I could see a clear advantage, BEFORE committing to the sacrifice.

The other thing about sacrifices is that even in situations where tbey are feasible, there is often another, safer and slower, but less risky way to win. You have heard the old golfing saw "Drive for show, Putt for dough", well there ought to be a similar one for chess, "The sacker looks great, the plodder gets mate".

The kind of positions that we see in these puzzles are relatively rare in actual games, but we get a false sense of the frequency of their occurrence because we are looking at them every day. I wish CG would occasionaly include some "mundane puzzles" where the best move turned out to be a combination to win a pawn or to achieve a slightly better strategic position.

Also most puzzle are attacking combos, some defensive combos might be nice as a change of pace.>

Total agreement here. I have posted suggestions similar to yours and sent messages to CG along similar lines, but so far no response.

CG changed the caption from "White/Black to play and win" to "W/B to play." But in fact, the puzzles are still almost always sacrificial attacks that lead to mate or overwhelming material advantage.

18...Nxg3 Qxb7 meant that 19. Nxf1 wasnt very exciting. I also tried tosoveitwith a White (queenside) rook on d1

but I was looking at 18....Nf4 gxf4 and tring to get win by getting rook over to the King -even looked at f5 as possible move...I then decided Nxg3 was the move but I didn't see the defence White came up with - but I doubt Tal saw much of it either...

Itis impsiosbe eenofr Super GM to calculate all these variations easily OTB -he would have see a lot of course.

And he lost plenty of a games (and like the rest of us he miscalculated and suffered from "chess blindness" on occasion) and he fell victim to attacks himeslf -all GMs do.