Counter-consensus on Kramnik vs. Deep Fritz

There are several assessments of the recent Kramnik vs. Deep Fritz match out today that suggest a growing consensus that machines have achieved complete dominance over humans at chess: see Stephen Moss at The Guardian ("Man v machine (and guess who won)"), Cyrus Farivar at Engadget ("Computer beats world chess champion, moving on to poker and go"), and George Dvosrsky at the Institute for Ethics and Emerging Technologies ("The future of chess"). I do not think this is a fair assessment of the match, which for me simply proved that "to err is human" and to forgive "does not compute." After all, Yasser Seirawan has shown quite convincingly that Kramnik had a clear win in game one. And most accounts of game two are that Kramnik had the better of a likely draw before "the blunder of the century." Games three to five were hard fought draws, where Kramnik proved himself at least the machine's equal. And in the final game, where Kramnik probably felt he had to go for a win as Black to tie the match, the World Champion tossed away at least equality with another blunder (the bizarre-looking 24...Rb6? according to Sakaev). It seems to me that, other than two blunders in games two and six, Kramnik proved himself at least equal to the computer. In fact, if we could just eliminate the problem of human error, we should be able to win next time.

Suggested solutions to the human error problem include giving the human more time, the ability to move pieces around on an analysis board (which the computer is basically allowed to do), or access to a database (which the computer has built in). All this points to the idea of pitting an unaided correspondence player against the computer. Though this would hardly be as interesting for the audience, it may be the best challenge for the computer programmers and something they are more likely to learn something from than the inevitable blunder-fest that staged matches have become.