Computers and Position
I was using Fritz 5.32 (w/o an opening book) to analyze a game of mine the other day and noticed something curious--I was playing with a focus on position, working to constrict my opponent's pieces. By around move 15, I had really restricted his mobility and he hadn't developed his pieces very well at all, yet the computer thinks I had only a fractional advantage until he made a few mistakes, at which point my advantage shot up and stayed high for the remainder of the game. It did, however, drop at one point, when I elected to castle to free up a rook and make my king a bit safer. Speaking generally, then, Fritz seemed to want me to play more aggressively, and to see positional moves as inferior. I realize, of course, that I'm not a great player or analyst by any means, so it's possible that the computer was right in this case, but it does make me wonder more generally: do computers tend to favor a more tactical approach? It would seem that if they're using a ply tree to look a few moves ahead, then they'd only be able to play good positionally with the short term in mind, because long term positional strategies would be buried in the exponential curve of possibilities.
Anyway, here's the game I was talking about, with annotations by Fritz (unfortunately, its snarky comments seem to have vanished). Analysis was done at a 60s per move, though my computer seems to take a lot longer than that to run it. Blunder threshold is thirty centipawns.