Black is toast after 24...Qxd5?

24...c4 seems to be the right move.

24...c4 25. Bxc4 Rc8 26. Bd3 Rxc1+ 27. Qxc1 Kh8=

click for larger view

How does Black continue after 24...c4 25. Be4?>

Be4 is much better. White avoids the trade of rooks after Bxc4 to keep the attack going. In fact, I'm finding white winning advantages after Be4. But the lines are long and it's slow going.

So far it's fair to say that black's problems started move 24.

<Anderssen kündigte hier "ohne Erbarmen" Matt in fünf Zügen an.>

Anderssen vs Paulsen, 1862

And here we see Steinitz missing 15....f6 in the Paulsen defense

First it was nice to see that by now, the top players actually played really decent openings. It's amazing how much progress top-level chess had made in the generation since the La Bourdonnais - McDonnell match, which wasn't really that hot, but at least fueled international communication and the beginnings of chess theory as we know it. Stockfish doesn't really like the Evans Gambit nor <10.Bb2>, but up to the 15th move, both players played really well.

Stockfish also doesn't like how Anderssen prepares his kingside attack -- but the thing is that he actually *prepares* it --, but even after Zukertort's equally well-prepared <17...b5>, it's still just -1, i.e. the Evans-gambited pawn, and after the somewhat premature <18...b4?> and <19...Bb6?!> (D), the positions are again equal. Personally, I thought that especially the latter move is a significant mistake -- the bishop and the a5-knight are effectively out of play, and as the game will show, they will remain this way until the bitter end and be the main reason for Zukertort's quick loss.

click for larger view

With the well-timed <20.g4!> (Stockfish would have played it even a move earlier), Anderssen finally starts his kingside attack, but it's only after <23...Bxf5?> that Stockfish begins to show a significant advantage for White. Instead it would have been absolutely essential to play <23...c4!> here to get those queenside pieces back into the game and generate some counterplay. Same goes for his next two moves, which further reduce the evaluation by a couple of pawns each -- each time, Stockfish and I agree that <...c4> would have been best.

<25...c4!> (D) is especially interesting, because White now has one and only one way to keep his advantage: <26.Bxc4! Qxd2 27.Bxf7+ Kxf7 28. Nxd2 Kxf6 29.Rf3>. Every other move leads to equality or an advantage for black. (It's nearly a swindle, although I agree that the winning move is pretty much forced.)

click for larger view

But as it is, Zukertort doesn't have any kind of counterplay, so Anderssen can peacefully build up his attack, and after <26...Kh8??>, Stockfish says it's +16 -- in other words, mate is only a question of time (and not blundering horribly). Zukertort seems to hope that he can use the enemy g-pawn as a shield -- it's one of my favourite tactics and often works wonderfully (I can literally hear my opponents scream in impotent rage XD), but I have already learned that it doesn't work not when the enemy g-pawn is on the 7th rank (and even less if half your army is chilling on the opposite wing).

And so it's over pretty quick; after <28...Qd6>, Anderssen announced a really nice mate in 5 which goes <29.Qxh7+! Kxh7 30.f6+! Kg7 31.Bh7+! Kxh7 32.Rh3+ Kg7 33.Rh8#>.

I found it very instructive to see that Zukertort's defeat essentially hinged on one single pawn push that would have mobilized his pieces which he kept not playing. I'm slowly beginning to think that piece mobility is the top strategic priority, as pretty much every other strategic guideline can be reduced to it. If I could ever be motivated to write another crappy chess engine (it's not that hard actually), I'll experiment with one that uses piece mobility as the only evaluation criterium (maybe with a little pawn structure evaluation on top), I'm pretty sure it will be quite decent.