[Computer-go] Congratulations to AlphaGo

[Lukas van de Wiel]

Whoa, what a fight! Well fought, and well won!

Lukas

[David Fotland]

Tremendous games by AlphaGo.  Congratulations!

[Jim O'Flaherty]

It was exhilerating to witness history being made! Awesome!

_______________________________________________
Computer-go mailing list
Compu...@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

[Aja Huang]

Thanks all. AlphaGo has won the match against Lee Sedol. But there are still 2 games to play.

Aja

[Hideki Kato]

Congratulations to DeepMind team! What an excellent full-board
perspective.

Hideki

Lukas van de Wiel: <CAK4iS=hY6CZ+JjOmc1uj1H_9GTqu...@mail.gmail.com>:

>Whoa, what a fight! Well fought, and well won!
>
>Lukas

>---- inline file

>_______________________________________________
>Computer-go mailing list
>Compu...@computer-go.org
>http://computer-go.org/mailman/listinfo/computer-go

--
Hideki Kato <mailto:hideki...@ybb.ne.jp>

[Darren Cook]

Well done, Aja and all the DeepMind team (including all the "backroom
boys" who've given the reliability on the hardware side).

BTW, I've gained great pleasure seeing you sitting there with the union
jack, representing queen and country; you'll probably receive a
knighthood. :-)

> Thanks all. AlphaGo has won the match against Lee Sedol. But there
> are still 2 games to play.

I love your focus! The mainstream media might start to lose interest
now, but at least the people on this list appreciate the implications of
the difference between 5-0 and 3-2. Best of luck in the last two games!

(And just when you thought you were almost back to a nice quiet studious
life, I heard rumour (*) that a Ke Jie match is coming soon.)

Darren

*: I say rumour, as the source was an interview with Demis Hassabis, but
only published in Chinese.

[Erik van der Werf]

Congratulations Aja & Deepmind team!

Now that the victory is clear, perhaps you can say a bit more on the latest developments? Any major scientific breakthroughs beyond what we already know from the Nature paper?

Enjoy the moments!

Erik

[Olivier Teytaud]

And please tell us how many stones AlphaGo can give to Aja or

to other strong players :-)

(well, I understand you have the two last games to take care of :-) )

--

=========================================================
Olivier Teytaud, olivier...@inria.fr, TAO, LRI, UMR 8623(CNRS - Univ. Paris-Sud),
bat 490 Univ. Paris-Sud F-91405 Orsay Cedex France http://www.slideshare.net/teytaud

["Ingo Althöfer"]

Dear Aja,

congratulations to you and the whole AlphaGo team for this
awesome performance of your bot!

I fully understand that you take also the last two games seriously.
But, please, do it in such a way that Lee Sedol will be willing
to support you in further matches to come.

Enjoy the day,
Ingo.

PS. Someone else mentioned it already: Your dress code was really
impressive. Maybe you can come to the Computer Games conference
in Leiden (June 30 - July 1) in exactly this jacket. (We will
organize a Union Jack for you :-).

PS-2. I spoke with an outsider. He thought that "Alpha"Go was
meant in the spirit of pre"Beta"version and asked: how strong will
the beta be and how strong the true engine?

["Ingo Althöfer"]

Hi Hideki,

in a sense this is also a great day for your Zen bot.
Some German guy who is fluent in Chinese (Prof. Dr. Marc
Oliver Rieger) followed the game today in a Chinese transmission.
He wrote that the Chinese pro player who was commenting
used commercial Zen to get life evaluations for the positions
of the game and informed his audience about the %-values!

Best regards, Ingo.

PS. On KGS two users (including me) used Crazy Stone to give
%-evaluations from time to time, and one more user (ruby) who
mentioned Zen-% a few times. Our numbers were accepted as
serious indicators of what was going on.

[Hideki Kato]

Dear Ingo ,

Thank you for the info. I guess these three games were so silent
that both programs estimated almost correct rate and score (for
the earlier stages, CS could be better than Zen, though :).

Best regards, Hideki

Ingo Althofer: <trinity-42173b52-29e5-4cb7-a151-6fc92503c167-1457788483634@3capp-gmx-bs67>:
>Hi Hideki,

>

>

>Best regards, Ingo.

>

>_______________________________________________

>Computer-go mailing list

>Compu...@computer-go.org

>http://computer-go.org/mailman/listinfo/computer-go

--
Hideki Kato <mailto:hideki...@ybb.ne.jp>

[Hideki Kato]

Congratulations again, Aja!

Once you told me (or I told you?) that "Go is fighting!". Now
which do you think, fighting or (whole-board) perspective?

Best regrads, Hideki

Aja Huang: <CAJbO_wGM5__ekRRK9jTX45kU...@mail.gmail.com>:

>Thanks all. AlphaGo has won the match against Lee Sedol. But there are
>still 2 games to play.
>
>Aja
>
>On Sat, Mar 12, 2016 at 5:49 PM, Jim O'Flaherty <jim.ofla...@gmail.com>
>wrote:
>
>> It was exhilerating to witness history being made! Awesome!
>>
>> On Sat, Mar 12, 2016 at 2:17 AM, David Fotland <fot...@smart-games.com>
>> wrote:
>>
>>> Tremendous games by AlphaGo. Congratulations!
>>>
>>>
>>>

>>> *From:* Computer-go [mailto:computer-...@computer-go.org] *On
>>> Behalf Of *Lukas van de Wiel
>>> *Sent:* Saturday, March 12, 2016 12:14 AM
>>> *To:* compu...@computer-go.org
>>> *Subject:* [Computer-go] Congratulations to AlphaGo

>>>
>>>
>>>
>>> Whoa, what a fight! Well fought, and well won!
>>>
>>> Lukas
>>>
>>> _______________________________________________
>>> Computer-go mailing list
>>> Compu...@computer-go.org
>>> http://computer-go.org/mailman/listinfo/computer-go
>>>
>>
>>
>> _______________________________________________
>> Computer-go mailing list
>> Compu...@computer-go.org
>> http://computer-go.org/mailman/listinfo/computer-go
>>

>---- inline file

>_______________________________________________
>Computer-go mailing list
>Compu...@computer-go.org
>http://computer-go.org/mailman/listinfo/computer-go

--
Hideki Kato <mailto:hideki...@ybb.ne.jp>

[Thomas Wolf]

A suggestion for possible future games to be arranged between AlphaGo and
strong players:

Whoever lost shall be given 1 stone or the equivalent of 1/2 stone handcap in the
next game. Games should continue until each side has won at least once. This
way AlphaGo will be forced to demonstrate its full strength over a whole game
which we are all too curious to see.

Thomas

[Jim O'Flaherty]

I LOVE this suggestion!

["Ingo Althöfer"]

Hi Thomas,

Von: "Thomas Wolf" <tw...@brocku.ca>
> A suggestion for possible future games to be arranged between AlphaGo and
> strong players:
>
> Whoever lost shall be given 1 stone or the equivalent of 1/2 stone handcap in the
> next game. Games should continue until each side has won at least once. This
> way AlphaGo will be forced to demonstrate its full strength over a whole game
> which we are all too curious to see.

That is one interesting proposal. I have another one:
You are the master of computer tsume go.
Give DeepMind a set of your tsume go compositions (from easy
to really difficult) and let them test which of the problems
AlphaGo can solve.

Cheers, Ingo.

[Michael Alford]

Give it the hatsuyo-ron :)

[Thomas Wolf]

Hi Ingo,

I have the manuscript of 2 books with each 100 computer generated problems
which 1st class insei Yutae Seo (Korea) picked out of 20,000 computer
generated problems, working for 5 months on this selection. Many have a tricky
ko status. I am happy to provide them.

Simpler even and more revealing would be to write down semeai problems using,
for example, the work of Teigo Nakamura. These problems can be evaluated in no
time once one understood the math but which take arbitrarily long to solve if
a brute force search would be applied. Simple pattern matching should not help
there.

Finally, there are seki problems which I showed several professional players,
including famous 9p who could not tell whether the game was over or not.

Lot's of fun tests one could do.

Cheers, Thomas.

[Sorin Gherman]

It is fascinating indeed to try to find how much stronger is AlphaGo compared to top humans.

Given the fact that it is hard to find the reason why Lee Sedol lost, and that AlphaGo seems to get mysteriously ahead without a clear reason, tells me that the difference is definitely more than one stone handicap, maybe 2+ stones, as crazy as it may sound given Lee Sedol's level.

I am pretty sure he will not accept to play with handicap against AlphaGo though. Maybe "younger wolves" like Ke Jie will though and we will find out.

[Thomas Wolf]

Chris,

Prompted from a discussion on the computer go email list
(and my last email today) :

We currently have no measure at all to judge how safe a winor loss is at any
stage of the game. The measure applied currently of counting territory does
only apply if both players try to maximize territory but not if at least one
player maximizes the chance of winning. (I know, it was mentioned already).

But really, comments like "Player ... is catching up" are pretty meaningless
and are only valid if one explicitly mentions points or territorry, and adds
that this has nothing to do with winning probabilities.

Even the winning percentages provided by the computer programs themselves are
no real indicator for winninig chances. They are tools to find the best move
and are a statistical measure over several playout sequences based on selfplay
not based on play against that opponent. Equally, winning percentages worked
out by other computer programs are also not adequate (although they are at
least unbiased) because they do also not use the real opponents to play out
the sequences.

The only valid strength indicator would be to gradually increase handicap
stones or komi for the previous loser in a series of games.

Regards,
Thomas

[Brian Sheppard]

Playing on KGS will give a good sense. Players have sustained ranks between 9 and 10 dan there, but never over 10 dan. If AlphaGo can sustain over 10 dan, then there is clear separation.

[Jim O'Flaherty]

I think you're correct, Thomas. The challenge is going to be getting ANY professional to be the one who "takes handicap stones" for the first time in years. The possible "shame" of doing so is what will make it messy.

Once someone does take that step, though, I think it is only a matter of time before the rating of humans will be made a subordinate rating relative to the "objective" rating of the AIs, AlphaGo just being the first. And that has its own psychological challenges as the Go world has many decades of handling ELOs and rankings for humans. So, I don't think change in this area is going to be welcomed anytime soon.

[Brian Sheppard]

Play on KGS. Pros can be anonymous, and test themselves and AlphaGo at the same time. :-)

 

From: Computer-go [mailto:computer-...@computer-go.org] On Behalf Of Jim O'Flaherty
Sent: Saturday, March 12, 2016 4:56 PM
To: compu...@computer-go.org
Subject: Re: [Computer-go] Congratulations to AlphaGo

 

I think you're correct, Thomas. The challenge is going to be getting ANY professional to be the one who "takes handicap stones" for the first time in years. The possible "shame" of doing so is what will make it messy.

[Lukas van de Wiel]

It makes me think that is a very good idea that they did not select a Japanese professional to play AlphaGo, They tend to be very passionate about shame (even though there is of course no shame in losing against AlphaGo) and it might have been... messy...

[Thomas Wolf]

Having AlphaGo playing exclusively on KGS would be such a boost to KGS!

[Clark B. Wierda]

On Sat, Mar 12, 2016 at 5:05 PM, Thomas Wolf <tw...@brocku.ca> wrote:

Having AlphaGo playing exclusively on KGS would be such a boost to KGS!

For sure.

The other Go servers might have their own opinion on that.

Clark 

[Lukas van de Wiel]

And the hardware available for this tournament was tremendous. It remains to be seen whether the hardware and the people maintaining it would be available for a longer period. The costs of this are not to be underestimated. Who would pay it?

Lukas

[Thomas Wolf]

On Sat, 12 Mar 2016, Lukas van de Wiel wrote:

> And the hardware available for this tournament was tremendous. It remains to be seen whether the hardware and the people
> maintaining it would be available for a longer period. The costs of this are not to be underestimated. Who would pay it?

The AlphaGo team would get feedback from testing by players with very
different ideas/strengths who they would otherwise never get in contact with.

For example, Michael Redmond mentioned repeatedly in the last 3 reviews that
he would love to play AlphaGo to study Go, to find new openings,...

[Lukas van de Wiel]

Oh, I did not say that it would not be beneficial, to AlphaGo, and to the people playing it, and to the Go community as a whole, but still, it will have to come from somewhere. Just the electricity bill alone would be hair-raising.

And the big-scale benefits in prestige and marketing are over, with this victory.

It would be cool to build on the works of AlphaGo, and I would like to see it as much as the next enthusiast, but I doubt the feasibility...

[Gonçalo Mendes Ferreira]

It does seem unlikely for DeepMind not to move on to "bigger" things,
but maybe the Go community can make some kind of fundraiser to keep an
instance of AlphaGo playing 24/7? I think there are some websites for
this kind of thing. Someone would be in charge of scheduling time for it
to play pros, other programs, and maybe play online on breaks. Just an
idea, oh Google overlords that watch all communications.

Gonçalo

[Lukas van de Wiel]

This would be many thousands of dollars per day. A single game would be more than a thousand dollars in total costs.

I do not think a kickstarter project or so would be successful, as the go community is simply not *that* big...

[Igor Polyakov]

At this point I don't doubt that the single machine version is professional strength which is enough to be used as a tool to analyze games...

[Jim O'Flaherty]

I'd expect this achievement by AlphaGo is very similar to when the first human ran a 4 minute mile. No one had done it prior. However, right after Roger Bannister did it, suddenly there were people all over the planet doing it. Roger Bannister essentially made the possibility real, and then the psychology changed and lots of others made it over the hurdle. AlphaGo turned the possibility of an AI becoming a 9d into reality.

AlphaGo may have made it to 9d first. However, I expect we will now begin seeing lots of different successful attempts to accomplish the same thing, and relatively soon, too. We've already seen several different comments on this email list of people working furiously to make the same leaps the AlphaGo team have described creating. The risk on these investments has been substantially reduced by AlphaGo's unambiguous success. If the chess world is any sort of guide to how Go AI is going to continue to develop, then we will see plenty of progress over the next 36 months. I wouldn't be surprised to hear the Facebook team working on their AI ends up coming in second to the +9d AI club.

[Gonçalo Mendes Ferreira]

The single machine version and the 2000 machine version is apparently a
difference in 150 ELO, so maybe a 32 core instance in Amazon would be
close enough, costing about 1200$ a month. Maybe it's doable, or for a
Tokyo U. or BGA or AGA to set up something like this with Google permission.

[Petr Baudis]

On Sun, Mar 13, 2016 at 12:33:55AM +0000, Gonçalo Mendes Ferreira wrote:
> The single machine version and the 2000 machine version is apparently a
> difference in 150 ELO, so maybe a 32 core instance in Amazon would be
> close enough, costing about 1200$ a month. Maybe it's doable, or for a
> Tokyo U. or BGA or AGA to set up something like this with Google permission.

It would be great if Google was willing to do this, but I'm not sure
this is realistic at all, as it might be all built around internal
Google frameworks and keeping it running at Google would mean someone
has to take care of it. So I suspect the Go world has to be a little
more patient until Deep Zen, Aya, oakfoam, possibly
https://github.com/Rochester-NRT/AlphaGo and others catch up. Save the
money for the openly available reinforcement learning iterations!

Just to clarify, 4xGPU, 32xCPU instance is $1800/month, while AlphaGo
single machine version uses 8xGPU, 48xCPU (and not sure what GPU
generation). So that's another computation power halving.

--
Petr Baudis
If you have good ideas, good data and fast computers,
you can do almost anything. -- Geoffrey Hinton

[Robert Jasiek]

On 12.03.2016 22:03, Thomas Wolf wrote:
> We currently have no measure at all to judge how safe a winor loss is at
> any stage of the game.

We have my theory according to my books for assessing the territorial
and the dynamic aspects (development directions, neutral stones,
statuses (incl. those of potential invasion groups), options, aji,
invasions, reductions, (local) potential, influence, thickness, fights)
of every position. The theory does not provide a single number (such as
a one-dimensional probability) but judgement need not be one-dimensional
and can depend on reading to assess particular aspects, such as a status.

> The only valid strength indicator would be to gradually increase handicap
> stones or komi for the previous loser in a series of games.

Altering komi is much better than altering handicap because komi can be
adjusted in finer steps and does not artificially restrict strategy a lot.

--
robert jasiek

[Dmitry Kamenetsky]

Congratulations to AlphaGo and its team! You have done what many of us could only dream to do and in such short time I may add. This is a truly historical moment and an amazing achievement for AI research!

I hope this is not the end of Go and only sparks more interest in this beautiful game. What an exciting time we live in and I can't wait to see what the future holds. 

Regards,

Dmitry Kamenetsky

[David Fotland]

Smart-games.com is getting a big increase in traffic, so there is certainly more interest in the game now.  I hope it holds up for the long term.

 

David

[David Doshay]

The SiliValley Go club is getting requests to join our email notifications at about 5 times the normal rate since the AlphaGo paper was published. So far everyone has had some prior knowledge of the game, and several have not played in a while. Some are beginners, but so far no people who do not yet know how to play.

Cheers,
David G Doshay

ddo...@mac.com

["Ingo Althöfer"]

Helo,

popular culture is growing around AlphaGo's win. Some pieces are
nice: others are, let's say, "special":

Here is the link to a nice Youtube video with an A capella hymnus
(31 seconds) on AlphaGo, performed on 9 GPU ;-)

https://www.youtube.com/watch?v=dh_mfGo183Y

Ingo.

[David Doshay]

Here is a tinyURL link to a panel discussion of things AlphaGo that included:

• Oliver Roeder: Senior writer at FiveThirtyEight. All too human.
• David Doshay: Archivist for the American Go Association, co-creator of SlugGo, a Go-playing computer program.
• Matt Ginsberg: Businessman, astrophysicist, creator of a former computer bridge champion called GIB and an expert-level AI crossword puzzle solver called Dr. Fill. FiveThirtyEight wrote about Matt and his new basketball prediction technology in October.
• Andy Okun: President of the American Go Association and a 1 dan Go player. He attended the match in Seoul.
• Jonathan Schaeffer: Computer science professor at the University of Alberta and the man who solved checkers.
This conversation has been edited for length and clarity.

http://tinyurl.com/jx7ctaw

As with all editing, each of us might have done it differently … at one point I am going on about multi-cpu and multitasking without context because they edited out the comment I was responding to. But I am pleased with the article in total, particularly the headline they chose which I think will resonate with the people on this list.

Cheers,
David G Doshay

ddo...@mac.com