Anyway, got many of the crucial moves for this game. I loved how Tal, just like in the Tal-Larsen, 1965, created another invincible barrier that splits the Black camp into 2. BUT this time, with an exchange sacrifice at e6.

It's also worth noting that when the tension at the center seemed to reach a stalemate...and there is no way to breakthrough, Tal unearthed a timely h2-h4-h5 to shatter the Black pawns in front of the King.

Can you guess move where Karpov got into trouble? Computer could not for awhile, and then when it did, move it chose for awhile longer was also mistake, wow! Think Tal. Think counter-intuitive. Continuing plan logically as Karpov did with 20...exd4 was first real error in game, 20...h6 playing on K-side and weakening K position computes as best. Houdini picked 20...Qb6 for awhile but this is also mistake. After 21.h5! continuing plan 21...Qb6 computes as best, and 21...gxh5 was decisive mistake where, as someone pointed out above, 22.Nf6!! instantly wins because black does not have time to save R and also prevent mate, this is extremely tricky move very unlikely to be found OTB in normal time control. With 22.Qxd5 chosen, 22...Rf8 was forced, same for 23.Bc2 Qe5 and for 24.Ng3 Qxe1+ with finally some choice on move 25 but very sharp attack Karpov missed elusive 25...Bh6 move and 25...h6 to try and kick B lost.

Notice how much material Tal went down 24.Ng3!? gave away second rook. For someone to do this intuitively goes beyond magician into genie, especially since 24...Qxe1+ is check giving black chance to breathe after K moves. Amazing game for normal time control, fact that it was blitz makes it even more awesome.

12 ... Be6 looks so ugly, Karpov must have expected 13 Rxe6 - for Tal it would have almost been a reflex. So he surely thought he could defend it. Despite the apparant imbalance the engine confirms the game to be remarkably even, right up to 21 ... gxh5?? I can only think he overlooked 22 Nf6! and "correctly" expected 23 Qxh5, only to go wrong again.

Even for Karpov, there are limits to what he could see in a blitz game.

In terms of 22.Nf6 maybe I did not make clear that, once one starts looking at it, calculating what happens is not difficult, this is not where "trickiness" was, I sometimes (often?) have difficulty finding proper words. It is just that 22.Qxh5 is so much clearly "thing to do" (at least to me, and evidently to Tal also) that tricky part is to NOT do this, step back, and find better move, especially in blitz. Like avoiding error of recapture 24...cxd4 in Kasparov vs Topalov, 1999. In both cases, computer does not care it can compute a zillion things as long as it has time and, since it has no intuition (or vision for that matter), it does not get tricked by what it sees.

25. Nf6 mate

Well, seeing your comments here, which were directed at YOU, to whit, that you we either a genius or a wanker, it is clear you are, sadly, a resident of the latter category.

The sac on e6 is a well known motif. I admire many of Tal's games (I have done since 1961) but this is, while ingenious, not one of his best.

Karpov played some wonderful attacking games when he needed to. There is no need to belabour this game...study some of Karpov's great games.