TalkChess.com

Just for the fun of it, let me display the guesses that go players on KGS (a go server) made on the outcome of the match.:
Lee Se-Dol vs Alphago match's guess:
BotHater36: 4-1.
Warfreak2: 0-5.
Isaac: 2-3.
rsun: 0-5.
bonobo03: lee wins.
naphthalin: 4-1.
Azal: 5-0.
dohduhdah : 0-5.
weirdchess: 3-2.
Falcon89: 3-2.
yoyoma: 4-1.
marathon: 0-5.
jv: 5-0.
Nick23: 0-5.
Falcon89 (changed his mind): 4-1.
Entity: 4-1.
Chris: 0-5.
heather: 4-1.
profan: 5-0.
USSRMan: 0 to 4 - 1 to 5.
superpal1: 5-0.
eucalyptu: 5-0.
bolbi: 5-0.
flobby: either 5-0 or 0-5.


So it's not that much of a surprise for many go players that Alphago won although Lee seems to had the edge. I am the one who guessed a 3-2 win for Alphago.

Dirt wrote:

Henk wrote:These poor programmers or researchers that switched from Chess to GO after it appeared that Chess had been 'solved' where should they go to now ?

Apparently StarCraft. I would be surprised if that's even as hard, but I know nothing about it.

StarCraft is an RTS with 3 assymetric but extremely well-balanced sets of units. It involves a lot of strategy, economic planning and supply management, scouting (fog of war), unit tactics, timing attacks, micromanagement of build orders and battles between units, and much more. There is a complex metagame of knowing what combinations of units and what build orders work best to counter certain other unit combos and build orders; this changes from time to time as players discover new strategies, sort of like human theory in chess or go. Skilled players have an almost eerie ability to synthesize correct guesses about what their opponent is up to, from obscure clues revealed in their scouting, and adapt their own plans to compensate. Bots are not really able to do this yet.

Watching tournament games on youtube with commentary from skilled players is a good way to get a feel for what Starcraft is about. (Mostly humans play Starcraft 2, but its similar to the original game in most ways. An example I like is the SHOUTcraft Invitational tournaments run by totalbiscuit.)

Also, check out the AIIDE competition they hold each year, to see the state-of-the-art in computer play. Spoiler: much higher APM (actions per minute) than humans, so they can execute dumb micro-heavy stategies like mutalisk harassment almost perfectly, but even the best bots are still far below the skill level of the strongest human players at strategy and adapting to the opponent.