MoGo is not 5d KGS -- there are some bots that are 4d/5d KGS, using the approach given in the article. The two best (that I am aware of) are Zen19 -- closed source -- and pachi -- available here: http://pachi.or.cz/

The author of Pachi has a paper detailing the major algorithm advancement beyond Monte-Carlo game simulations, which is sharing information across simulations in a way that is totally nonintuitive to me.

For me, a 3d/4d KGS go player, this is all very interesting stuff!

Zen beat Takemiya Masaki a few weeks ago in a non-blitz two-game match at five and four stones. It didn't even use an excessive amount of hardware, just a couple of boxes:

http://gogameguru.com/zen-computer-go-program-beats-takemiya...

Actually, the author considers them to be a blitz game.. http://gogameguru.com/zen-computer-go-program-beats-takemiya...

Fair enough. I'd call it a "quick" game, analogous to quick games in chess. It was 25 minutes of main time and 5x30 sec byo-yomi periods, IIRC.

Not the kind of blitz people play on KGS, at least.

Go often takes a lot more moves (~300+) to finish than chess. 25min per player / 150 moves = 10 seconds a move per player.

Compared to ~40 in a chess game would be 2 min per side / 20 moves = 10 seconds a move. (http://www.chessgames.com/chessstats.html)

A game with 5x30 sec byo-yomi could take quite a bit longer than the 25 minutes free time, so that's not a good comparison.

https://en.wikipedia.org/wiki/Byoyomi

After the 25 is up, each player can make as many moves in 30 seconds as they'd like, the 5 means that they can overstep that bound 5 times (by 30 second increments).