Join 3,559 readers in helping fund MetaFilter (Hide)

REMIND ME AGAIN HOW THE LITTLE HORSE-SHAPED ONES MOVE
April 4, 2012 12:46 AM   Subscribe

The King's Gambit, a once-popular chess opening, has now been solved to an extremely high degree of confidence. The author of Rybka, a high end commercial chess engine, modified it to be able to distribute work to a cluster of IBM servers similar to that used to power Watson and have spent 10,750,000 CPU-hours exploring the space of possibilities.

The claimed solution is that White has exactly one move which allows it to force a draw and everything else is a loss for White. This is not a complete solution as it proceeds by pruning out any positions it thinks are untenable or guaranteed wins (if the score is more extreme than +/-5.2 it considers it a guaranteed win/loss) but is believed to be extremely likely.

Rybka, the software in question, is the subject of some controversy. In 2011 it was disqualified from competing due to claims of plagiarizing two other Chess engines, Crafty and Fruit. Nevertheless, the software continues to be sold, reasonably convincing arguments have been put forth claiming this judgement was unwarranted. Either way Rybka remains a highly competitive piece of software and the validity of the Kings Gambit solution is not affected by its provenance. (Previously on Metafilter)
posted by DRMacIver (4 comments total)

This post was deleted for the following reason: Poster's request. -- taz



 
This is really interesting, but my faith in the result (and one needs a little faith, since the search space has been heavily pruned) is a little hurt by remarks like this one :

"Our algorithm works in an iterative manner – it first forms a hypothesis, and then it confirms or alters that hypothesis over a number of passes using a non-deterministic Turing Machine program running across the clusters"

Really? You've got 3000 cores, and what you're doing is using them to
a) simulate a machine that writes symbols on an infinite tape (Turing machine)
b) simulate many of these machines in parallel, but never allow them to communicate with each other? (The nondeterministic bit)

The Turing machine line seems calculated to impress people who know nothing about Computer Science. That makes me nervous about the rest of the claims made.
posted by Omission at 1:08 AM on April 4, 2012


Apparently I've been belatedly April fooled. Sigh. I hate internet stupidity day. Can a mod remove this?
posted by DRMacIver at 1:10 AM on April 4, 2012


Bottom right corner DRMacIver to "contact" someone.
posted by lee at 1:44 AM on April 4, 2012


Got it.
posted by taz at 1:52 AM on April 4, 2012


« Older The Seattle Times has just published a largely unf...  |  Borg Cube vs Death Star... Newer »


This thread has been archived and is closed to new comments