Myungwan Kim was beaten by the MoGo computer system by 1.5 points in a nine stone game.

“It played really well,” said Kim, who estimated MoGo’s current strength at “two or maybe three dan,” though he noted that the program – which used 800 processors, at 4.7 Ghz, 15 Teraflops on borrowed supercomputers – “made some 5-dan moves,”

Read the horrific details here.

Comments

  1. #1 Brian X
    August 11, 2008

    Well, okay, yeah. Throw enough computer cycles at any game, it’ll crack. But based on the known limitation of Moore’s Law, is this sort of power ever likely to trickle down to consumer level, or will such a system always remain a pathological case available only to research labs?

    (Not so worried about the 4.7 GHz processors themselves — those will probably be commonplace by 2010.)

  2. #2 Duae Quartunciae
    August 11, 2008

    Now I can no longer tell my friends that computers have been unable to play at Dan strength. Sigh.

    Although… the computer was given a huge advantage, with 9 stones. Has a computer ever beaten a Dan-rated player without the handicap?

  3. #3 student_b
    August 11, 2008

    @Duae

    AFAIK no, a computer has never beaten a professional player without a handicap.

    Though I would wager that no computer would ever be able to defeat Hikaru. :D

  4. #4 Ian
    August 11, 2008

    Does this mean you have to have some serious stones to take on a computer these days…?

  5. #5 Dunc
    August 11, 2008

    Well, okay, yeah. Throw enough computer cycles at any game, it’ll crack.

    It’s not necessarily that simple. Go isn’t really much like chess – it’s much less susceptible to brute force calculation.

  6. #6 lylebot
    August 11, 2008

    There’s always much rending of garments and gnashing of teeth when a computer becomes significantly better at some game, but it seems to me that the reason these games are interesting to us is that they require unnatural computational abilities, the kind of skills that very few people possess and that are very difficult to teach. It shouldn’t be surprising that computers can beat people at them, any more so than a cheetah outsprinting the fastest human or a chimpanzee outlifting the strongest human. It shouldn’t take away from the accomplishment of people who are good at them. What’s really surprising to me is that it’s taken so long to come up with programs that are good at them, but that just speaks even more to the ability and accomplishment of people that are good at them.

  7. #7 rpenner
    August 11, 2008

    Has Conway quantified “some 5-dan moves” ?

Current ye@r *