Hacker News Reader

Comment by badminton1 | Hacker News Reader

I don't think Ke Jie would win against Alpha Go either.

I agree. Lee lost a title 3-2 to Ke Je recently but he lost very convincingly against AlphaGo. Ke Je is not that stronger than Lee. He's going to lose too but he's 18 so I understand that he's feeling he can do anything.

My feeling is that AlphaGo could be a couple of stones stronger than the best pros. Considering that dans are capped at 9 but there are 9 dans that always win against some other 9 dans, the best pros could be 10 dan on their scale. That would also be 10 dan on the amateur scale (European, Americans are usually one stone gentler and Japanese much more than that). AlphaGo could be 12d amateur or, don't know, 14-15 dan pro? (pro levels are on a finer scale)

Do you know what the current limit is on a handicap that a professional player can overcome? Is there a known theoretical limit? That's a possible 'endgame' for go ai.

I'm not sure I'm understanding correctly but I believe some pros have done demonstrations against weaker amateurs (15k?) with something like 25 stones. Yes it'd be fun to see AlphaGo take pros with huge handicaps. And even inside handicaps there's alternatives. Like free placements vs just on star points. AlphaGo against a pro that starts with two stones in each corner could be neat. At some point I'd imagine the pro associations might balk at such games though.

There must be some limit because the number of the possible moves is finite. How many handicap stones the current professionals are from that limit... who knows. Apparently AlphaGo could set a better lower limit but it should play with many more pros, and start giving handicap if needed.

I wonder if Google will make it available to pros outside these official matches. It's not that Deep Blue played so much after winning with Kasparov in 1997, right? Kasparov asked a rematch and IBM refused. Deep Blue never played again.

Our computers are stronger than it now, mainly thanks to advances in the algorithms. Deep Blue used to crunch many more positions per second than the current chess engines do on standard PC. Still they're stronger. AlphaGo is running on 1920 CPUs and 280 GPUs, so either research trims down its algorithms or we wait for CPUs with 1024 cores. But at the current size they won't fit below the keyboard of a laptop :-)

This hardware setup is crazy. Given alone 280 GPUs is insane.

That's just for the training phase. In play mode it only uses 40. And the top supercomputer has 3 million cores, so, AlphaGo is not such a big computer either.

There's a single machine version, which beats the distributed version 30% of the time. That should be plenty strong to be useful.

I have read that a go world champion playing against god would need between 4 and 5 stones of handicap.

No, I'm skeptical. And most importantly, due to how AlphaGo seems to be raising the bar just enough to lead confidently, I think Ke Je is in no position to judge how he would fare in advance. I mean, we have absolutely no idea whatsoever as for AlphaGo's true potential.

Reply on news.ycombinator.com