They continue the same mistake of setting amateur 9d below professional 1p. If AlphaGo v.18 were much stronger than human 9p, it would not have lost any game against Lee Sedol.

Agree, this is my version of the graph, I don't claim it's 100% accurate though cause professional dan work in different way than kyu - amateur dan.

Gosh, beating Nature/Fan Hui AlphaGo with 4 stones is impressive, but I reserve my judgement about the meaning of that in actual strength given bots may have trouble playing with/against handicap. A better test would be could/did it give 4 stones to Fan Hui? If AlphaGo v18 were 4000 Elo and Lee Sedol 3500 then it should win a single game 95% of the time and the match 5-0 with 78% chance so it doesn't look 4000 let alone 4500. Having said that it only lost game 4 due to one move far below the usual high standard of its moves (and subsequent monte carlo nonsense which can sort of be dismissed, though not entirely as the ability to play sensibly and well when behind is important, though perhaps not so much if AlphaGo fixes that bug and is then never behind in any game against a human ever!) so if they could fix that bug/blunder/weakness then such high Elos might be plausible. However, I expect humans to improve how they do against AlphaGo with practice if Deepmind make it more widely available, though of course AlphaGo will also likely keep improving. We can already see than in just 5 games Lee Sedol was able to learn and adapt how to play against AlphaGo with some success, whilst AlphaGo v13 needed 60 million positions and games to get to its level (plus scores of clever developers/researchers) and AlphaGo v18 probably used far more, so humans are still pretty good at this learning thing.

P.S. ajahuang was on KGS yesterday testing the gtp interface for an AlphaGoBot to play on KGS, subject to Google's permission. Said it would likely first play in some arranged matches against some pros / 9ds before (if) being more widely available. Exciting!

I was also there, very exciting time even it's a false positive

I think that graph is believable is you treat "9d" as "average 9d". In that case, Lee Sedol himself would not be 3500 but probably 3800+.

Remember that these scales are arbitrary anyway.

Well I can certainly see why they were "quietly confident" going into the Lee Sedol match!

BTW it played 7 games on Tygem today and won all 7.

What nickname is it on tygem?

The "deepmind" account actually has something like 40 wins 20 loses.

Yes the deepmind account. I just meant it's most recent 7 games.

But in last 7 games it won 2 and lost 5 or was it won 3 lost 4

I'm at work now so I can't check with the client, I'm going off this reddit post https://www.reddit.com/r/baduk/comments ... ns_so_far/ which has this image linked: http://postimg.org/image/rvxwwof93/

There is has seven red circles in a row at the end, and below you can see those games played on March 25.

Given that Aja Huang said he didn't get Google approval to play on KGS yet, I doubt that the deepmind account on Tygem is AlphaGo, at least it won't be the latest version.
Why would they give approval for Tygem, but not for KGS?

_________________
Sorin - 361points.com

Didn't Aja Huang also say that 'deepmind' on Tygem was his account [1]? Is it possible that 'deepmind' on Tygem is not "pure AlphaGo", but rather Aja Huang + AlphaGo - maybe experimenting a bit, or just playing around?

Just a theory, but it could explain things.

---
[1] https://www.reddit.com/r/baduk/comments ... _on_tygem/

_________________
be immersed

They aren't exactly straightforward about it. Yes earlier Aja Hwang had said that deepmind was his own personal account. But later there was a Korean article with a picture of Demis Hassibis and the CEO of Tygem. The title was something like "Thanks Tygem!". Even in that article it's sorta like Demis is saying it's Aja's account, but he sorta "indirectly" confirmed it. It's like they are saying it's Aja's personal account, but they are no longer directly denying that he used it for AlphaGo anymore.

I want to be careful about this too but at this point is just seems overwhelmingly likely that AlphaGo is playing on that account, just for some reason Deepmind just doesn't want to publicize this. That's my take at least.

Maybe this article?

http://www.tygem.com/news/news/viewpage ... &findword=

_________________
be immersed

I think you're probably right, looking at that article (I assume it's the same one). Probably Aja Hwang created the account, and they are playing AlphaGo's moves on it (at least) now.

I'd guess they are probably not going all out to win every game on Tygem, but probably, they are using it as a testing ground to get new ideas from humans, and to iron out bugs (like AlphaGo's response to Lee Sedol's move 78 in Game 4, for example).

_________________
be immersed

I assume that the scale they are using is the one of goratings.org, where Lee Sedol is 3518.

I also checked goratings before posting, but as far as I know there is no indication that they are using it. I just offered a possible explanation.

They're using it, it stated in nature paper, and the nature AlphaGo rating in nature paper is the same as this AlphaGo v13 in this graph.

If you see a slide that says "CAUTION", it's good to take that into account.

The word covered by the speaker's head is "self". Bot results in self-play are always massively exaggerated. It's not uncommon to see a 75% self-play winrate in selfplay to translate to 52% winrate against a third-party reference opponent. c.f. fig 7&8 in http://pasky.or.cz/go/pachi-tr.pdf . Intuitively, I'd expect the effect to be less pronounced with very strong programs, but we don't know anything about the mechanics here and experiments are difficult.

It's no doubt today's AlphaGo is much stronger than the Nature version. But how much? We'll have a better idea when they pit it in more matches with humans, and ideally when other programs catch up further. Without knowing more (like the rest of the slides or a statement by someone from Deepmind), I wouldn't read much into this graph.

_________________
Go programmer and researcher: http://pasky.or.cz/~pasky/go/
EGF 1921, KGS ~1d and getting weaker