Quote:
Originally Posted by Dire
Oddly enough it seems a computer has already defeated an 8 dan pro at Go on a 9 stone handicap: http://www.cs.unimaas.nl/g.chaslot/muyungwan-mogo/ The success was declared to be due to the development of: "revolutionary algorithms that were invented in the period 2006-2008". Hyperbole or not, Go doesn't have anywhere near the history of computer involvement that chess does and the computers aren't going to be getting any weaker.
I'm aware of that game. The algorithms that Mogo (and MogoTitian) uses, is very ironically, the Monte Carlo method. It essentially plays random moves (for both sides), very quickly, and chooses the one that leads to the best result. Seems hardly a advancement, but it works.
For the record, they played several rematches, and MogoTitian was crushed, making mistakes that a 15 kyu wouldn't make.
Quote:
but right now - computers just show that humans are bad at chess, not that computers are particularly good at it.
Hypothetical: Get the top 10 GMs, the top 10 correspondence GMs, throw them in a room, and have them play a no-time limit game vs whatever computer is best now. What's the result?