Computers reach 5d on KGS

[ethanb]

The other interesting thing (to me) is that when you push the blitz settings past a certain value, the bots lose again. Intuition counts for a LOT in this game.

Now, I don't have 24 cores available at home, but I know that on my computer shrinking the time per move below 10 seconds gives me huge advantages against MoGo or GnuGo. Probably more parallelization would push the threshold down to a level where the advantage of intuition is outweighed by inability of the human to click fast enough, so this won't be universally applicable.

[hyperpape]

Interesting tidbit: when Myungwan Kim played Mogo, giving it a 9 stone handicap, he used only 11 minutes of his time, and said that he would not have benefited from using more.

[Kirby]

Interesting tidbit: when Myungwan Kim played Mogo, giving it a 9 stone handicap, he used only 11 minutes of his time, and said that he would not have benefited from using more.

That is very interesting. I wonder why. Maybe further time spent reading ahead would be spent reading variations that Mogo wouldn't likely come up with? I wonder what he meant.

[hyperpape]

It's ambiguous: Mogo uses Kim's thinking time. So perhaps he reasoned that it would help the bot too much to have more time, even if he could play slightly better with more time.

But there was also a professional who said that when he played an amateurs, only his fingers got tired not his brain. That would imply that professional basic instinct is what defeats amateurs, in which case more time is almost irrelevant.

I suspect both factors matter a bit.