Mikhail Tal And The Machines

There was a point, or pet theory at least, that I had intended to make in the last post. It involved quoting from a table I once saw out there in Internetland that showed the results of passing all the World Champions' games through a strong Chess engine and finding the percentage of coincidence between each champion's moves and the engine's first choice. For all the joys of Google, do you think I can find the damn thing when I want it?

From memory (I guess I'll update this if I stumble across the data I've misplaced...) pretty much all the champions in modern history played the "Fritz move" 80% +/-5% of the time. One interesting thing about this table that seemed to have a lot of people flumoxed was that Tal's coincident percentage was about the same as everyone else's - not the highest, but certainly not the lowest either. Tal by his own admission played many "unsound" sacrifices!

So the story goes, in the post mortem to one of his games, a young Garry Kasparov showed a sacrifice he had wanted to play but said he couldn't calculate all the consequences. "My boy, first you sacrifice, then you calculate!" was Tal's response.

As with any great magician, we become so dazzled by Tal's magic that we're distracted from noticing his consumate technique - the decoy tactical motif taken to a new level. It is not so often that Tal's opening knowledge and endgame technique are commented on, mundane as they are in the shadow of his middlegame wizardry. Yet noone becomes World Champion without total mastery of these mechanics of the game in addition to their stylistic expression.

One thing that the game in the last post shows - granted it's a very small data sample, but don't let that get in the way of a good story - is that it was Crafty trying to find Tal's moves, not Tal trying to find the machine's, and Crafty needed help! Crafty always came round to Tal's point of view eventually, but it had to be taken further down the game tree to realise that the wild attack and sacrifices were good for more than a draw.

"First I move my pieces to the centre of the board. Then I sacrifice them." - Mikhail Tal.

Let's interpret this statement in the context of Tal's high coincidence with computer moves. He begins the game with solid opening theory. Then he makes a sacrifice that, for the purposes of this argument we'll assume to be "unsound". His opponents can do one of two things. They can play sound moves and refute the sacrifice, or they can fail to find the defense - play unsound moves themselves - and get splattered. My 2700 rated version of Crafty falls into the second category - at least under tournament conditions with limited calculating time!

So how is Tal's "Fritz coincidence" so high? He only plays one unsound move per game, a very small minus percentage! The rest of his moves are perfectly sound, either more correct sacrifices in accordance with the position, or falling back on his World Champion class technique if the game steers towards an ending, or didn't permit even him to make sacrifices in the first place.