World #1 Go Player Ke Jie accepts Google Alpha Go Match..

[Laskos]

No, if you follow "Nature" paper, 60-70% (especially tactically, MCTS) of gain comes from CPU.

Absolutely wrong. Why is it so hard to see the obvious from the paper???
It is enough just to look at Extended Data Table 10 (Scalability study).
2 threads vs 1 thread 70% winning chance.
2 GPUs vs 1 GPU 95% winning chance.

Not that wrong, when I see the table, improvements from CPU and GPU come pretty equal, about 700 ELO points each. GPU improvement dampens faster. So, I stand corrected in the sense that 50% comes from CPU, 50% from GPU, in their reasonable range (1->40 threads, 1->8 GPUs). First doubling is indeed large for GPU, that's why I have 2 of them for a 4 core machine.

Not really, at least not in case valid for smartphones. 50% comes only when system is well balanced and has a lot of GPU power (bear in mind GPUs in Lee Sedol match were those custom extremely powerful TPUs).
40 threads 1 GPUs is only 18% vs. 4 threads, 8 GPUs. So 10 times more cores is much, much weaker (in terms of Elo particularly since 1% advantage is much more Elo than in chess) than 8 times more GPU.
When GPU power is low, the difference in favour of GPU strength is really pronounced. And this is particularly strong point in case of smartphones which have pretty low GPU power compared to CPU power.

Want to bet that by 2020 we will have top world level Go on Smartphone?

[Laskos]

Wow, already a pro level Go program for pc, and remarkably it's free! People were saying maybe we'd have a pro level Go engine for pc in a year or two after this match, but it already happened! I wonder if any of the techniques used in these go programs would help with chess?

The new development started one year ago, with the first Alpha Go appearance and paper. Leela is the first for public to reach Pro level, and it's free. I guess in a year more we will have top world class Pro level on strong PC (with GPU), and in 3 years the same on smartphones. Leela is already a good analysis tool, its picture about the whole board is very accurate, so much that it helped me to understand the games better than top Pro humans commenting.

DeepMind team published 10 self-games of Alpha Go for top players to acquire new ideas. I analyzed with Leela these games, and Leela sees that they proceeded differently from Alpha Go - Ke Jia matches. A typical game:


A for applicability in chess, there are threads in Programming Forum here. The general opinion is that as of now there is little applicability for top engine level chess, and this is my opinion too.

[Laskos]

Ke Jie after match statement at press conference, really interesting, from the strongest human player:

I was unsatisfied with the official interpretation of Ke Jie's comments so I took it upon myself to do a better translation. (any suggestions are welcome!): [Edited]
It’s a great honour to be able to participate in these matches. Many thanks to everyone’s support, Zhejiang Go Association, to such a great opponent Alphago, and thanks to everyone’s hard work for making this event possible.
Actually these matches, for me, are more meaningful than any other competitions I’ve ever been in. In today’s game, I originally thought I could have played a bit better. But unexpectedly, even in the opening I played some unforgivably bad hands, causing the situation to be very difficult from the start. I knew that when playing Alphago, if you start out behind, trying eke out a win is pretty much hopeless. In fact even continuing to fight was very difficult. It is really too perfect. It has no weakness, and no emotional lability. So I’ve been blaming myself for not playing a bit better. (visibly upset) (applause)
I hoped I can play a bit better. Unfortunately I wasn’t able to do that today. I don’t want to call this is a win-win situation, because it was a competition. So it’s really quite painful. Maybe I feel differently than others, because I really feel my performance was terrible. And I probably won’t have another chance like this again, so I think it’s regrettable. I hope I will continue to discover truths about Go. Through these games, especially the last one, I’ve discovered that I have too many shortcomings, and too many emotional liabilities. There are definitely problems with my understanding of Go. I feel like many other players could have done better than me in this match, so I don’t feel worthy of your praise. I’m very sorry, I’ve lost. I wish I will be better in the future. (applause)
Actually today I was flustered, not my usual self, because it was really too perfect. In the first half of the game, I already knew that I was in a disadvantaged position. First of all, whenever I feared Alphago would play a move I had predicted, it would definitely play it. Sometimes it would play a move I had not considered, and after thinking about it for a while, I’d realize that it had come up with a much better move. I can only predict half of its moves. The other half I cannot even guess. This reflects a huge gap in our abilities. There’s no way around it, I have to admit that. The gap is huge.
I’m very thankful to have an opponent such as Alphago, to allow me to know that I still have so far to go. I really wish I can improve in the future, and continue to work hard, and be able to shrink that gap a bit. But this is just a wish. I believe that the gap will become bigger and bigger in the future. It is already so perfect. Again I feel regretful about my performance today, and I’m happy for the AlphaGo team, that they’ve created such an amazing, frightening, calm, such a perfect go player. It is an incredible accomplishment to create such a perfect player. Thank you.
In response to the first reporter’s question:
I don’t know. I don’t know how I will change in the future. Actually I’ve been trying to change myself continuously. But in contrast to a great company such as DeepMind, I can only make some small transformations, limited to myself, but they will change the whole world. So I will just say that I want to make myself better, I will continue to work hard, play go, and help bring the joy of go to more people. Actually I think that playing go should be a very happy thing, just like my friends who are go players (looking at them). Playing with them is a very fun thing to do, so I guess I’ve failed to show this to the audience today. Sometimes I care about winning too much, and I have a hard time trying to control this desire. The last couple nights I didn’t sleep well, because I knew this was the last game, so I was very nervous. I was thinking about what kind of strategies I should use against AlphaGo. I was just worrying excessively causing more trouble for myself. I really believe playing go should be fun. I will try to reflect the fun nature of go better in the future. But sometimes without the challenge of win or loss, go can be more boring. For me, winning IS fun. I guess this is different from AlphaGo. I can enjoy happy moments playing go more with human players. But playing with AlphaGo, because it is too perfect, playing against me, it has no flaws, so it’s been painful for me. Before the match, I’ve said before that this is probably the last time I will play against AI. The reason should be obvious, because playing against it is really quite painful. I don’t wish to bring such pain to myself, but it is just so cool and detached, it does not give you any hope of winning. It doesn’t give you any hope at all.
Today I thought it would play me until the end of yose, and win by a little bit as usual, like by 1.5 or 2.5 moku. That would save some face for me. But it ended up taking my stones (laughs), I didn’t think it would be so cruel. Why doesn’t it let me save some face? (laughs)
It has so many qualities that amaze me. I don’t think I could possibly surpass it in my lifetime. But I can improve myself, overcome myself. This I believe I can do. As the reporter mentioned, I hope I will continue to be the Ke Jie that is flamboyant and self-confident. I hope I can also let more people know that Go is not so such an esoteric, mysterious, and unapproachable. I think it is actually a very grounded, approachable, and easy game. Thank you.

[Laskos]

Ke Jie after match statement at press conference, really interesting, from the strongest human player:

I was unsatisfied with the official interpretation of Ke Jie's comments so I took it upon myself to do a better translation. (any suggestions are welcome!): [Edited]
It’s a great honour to be able to participate in these matches. Many thanks to everyone’s support, Zhejiang Go Association, to such a great opponent Alphago, and thanks to everyone’s hard work for making this event possible.
Actually these matches, for me, are more meaningful than any other competitions I’ve ever been in. In today’s game, I originally thought I could have played a bit better. But unexpectedly, even in the opening I played some unforgivably bad hands, causing the situation to be very difficult from the start. I knew that when playing Alphago, if you start out behind, trying eke out a win is pretty much hopeless. In fact even continuing to fight was very difficult. It is really too perfect. It has no weakness, and no emotional lability. So I’ve been blaming myself for not playing a bit better. (visibly upset) (applause)
I hoped I can play a bit better. Unfortunately I wasn’t able to do that today. I don’t want to call this is a win-win situation, because it was a competition. So it’s really quite painful. Maybe I feel differently than others, because I really feel my performance was terrible. And I probably won’t have another chance like this again, so I think it’s regrettable. I hope I will continue to discover truths about Go. Through these games, especially the last one, I’ve discovered that I have too many shortcomings, and too many emotional liabilities. There are definitely problems with my understanding of Go. I feel like many other players could have done better than me in this match, so I don’t feel worthy of your praise. I’m very sorry, I’ve lost. I wish I will be better in the future. (applause)
Actually today I was flustered, not my usual self, because it was really too perfect. In the first half of the game, I already knew that I was in a disadvantaged position. First of all, whenever I feared Alphago would play a move I had predicted, it would definitely play it. Sometimes it would play a move I had not considered, and after thinking about it for a while, I’d realize that it had come up with a much better move. I can only predict half of its moves. The other half I cannot even guess. This reflects a huge gap in our abilities. There’s no way around it, I have to admit that. The gap is huge.
I’m very thankful to have an opponent such as Alphago, to allow me to know that I still have so far to go. I really wish I can improve in the future, and continue to work hard, and be able to shrink that gap a bit. But this is just a wish. I believe that the gap will become bigger and bigger in the future. It is already so perfect. Again I feel regretful about my performance today, and I’m happy for the AlphaGo team, that they’ve created such an amazing, frightening, calm, such a perfect go player. It is an incredible accomplishment to create such a perfect player. Thank you.
In response to the first reporter’s question:
I don’t know. I don’t know how I will change in the future. Actually I’ve been trying to change myself continuously. But in contrast to a great company such as DeepMind, I can only make some small transformations, limited to myself, but they will change the whole world. So I will just say that I want to make myself better, I will continue to work hard, play go, and help bring the joy of go to more people. Actually I think that playing go should be a very happy thing, just like my friends who are go players (looking at them). Playing with them is a very fun thing to do, so I guess I’ve failed to show this to the audience today. Sometimes I care about winning too much, and I have a hard time trying to control this desire. The last couple nights I didn’t sleep well, because I knew this was the last game, so I was very nervous. I was thinking about what kind of strategies I should use against AlphaGo. I was just worrying excessively causing more trouble for myself. I really believe playing go should be fun. I will try to reflect the fun nature of go better in the future. But sometimes without the challenge of win or loss, go can be more boring. For me, winning IS fun. I guess this is different from AlphaGo. I can enjoy happy moments playing go more with human players. But playing with AlphaGo, because it is too perfect, playing against me, it has no flaws, so it’s been painful for me. Before the match, I’ve said before that this is probably the last time I will play against AI. The reason should be obvious, because playing against it is really quite painful. I don’t wish to bring such pain to myself, but it is just so cool and detached, it does not give you any hope of winning. It doesn’t give you any hope at all.
Today I thought it would play me until the end of yose, and win by a little bit as usual, like by 1.5 or 2.5 moku. That would save some face for me. But it ended up taking my stones (laughs), I didn’t think it would be so cruel. Why doesn’t it let me save some face? (laughs)
It has so many qualities that amaze me. I don’t think I could possibly surpass it in my lifetime. But I can improve myself, overcome myself. This I believe I can do. As the reporter mentioned, I hope I will continue to be the Ke Jie that is flamboyant and self-confident. I hope I can also let more people know that Go is not so such an esoteric, mysterious, and unapproachable. I think it is actually a very grounded, approachable, and easy game. Thank you.

I felt that beyond this facade of politeness and humbleness of DeepMind team and Hassabis, a rude attitude emerged as a cultural misunderstanding. What they basically said is that once Go was conquered, they will move to harder and more important things like Starcraft. This is ridiculous even to me. The level of Mastery of top Pro level player in Go is probably above even that level in Chess. Go is a game of 400 years of competitive history, with literature, tutoring, learning effort higher than Chess. Let's call it 3000 ELO level Chess mastery. To become a top expert in Starcraft an imbecile needs 2 weeks of hard play and learning, if tutored. Let's call this 1600 level of Chess. The DeepMind team is sufficiently rude to dismiss the human skill level, and the simple fact that Starcraft uses similar techniques and is even harder for AI would legitimize dismissal of an entire culture.

So much for fake politeness, to which I pointed earlier.

[Dirt]

Ke Jie after match statement at press conference, really interesting, from the strongest human player:
...

That's a much better translation than the on the fly one they did live. Thank you for posting it.

I was wondering after the Lee Sedol match if Google would just declare a success and go home. I'm glad they didn't. I also agree with Ke Jie now, it's over. Just as in chess it's no fun for the top players to keep losing to the machines.

[Vinvin]

"I've suffered enough defeats from it"

https://www.youtube.com/watch?v=OCevCII1zo0

[Milos]

What they basically said is that once Go was conquered, they will move to harder and more important things like Starcraft. This is ridiculous even to me. The level of Mastery of top Pro level player in Go is probably above even that level in Chess. Go is a game of 400 years of competitive history, with literature, tutoring, learning effort higher than Chess. Let's call it 3000 ELO level Chess mastery. To become a top expert in Starcraft an imbecile needs 2 weeks of hard play and learning, if tutored. Let's call this 1600 level of Chess. The DeepMind team is sufficiently rude to dismiss the human skill level, and the simple fact that Starcraft uses similar techniques and is even harder for AI would legitimize dismissal of an entire culture.

History means nothing compared to science and technology. Most of traditional martial arts were developing for hundreds if not thousands of years and still they are laughable in usefulness and powerfulness compared to modern MMA combo that has been developing for hardly 20 years.
The whole history of human chess opening before computers is laughably weak in comparison to a plain Brainfish book.

[whereagles]

do note that an API has been available for StarCraft: Brood War that accomplishes much the same thing for around 7 years now, receiving some academic attention. Despite very early on achieving superhuman feats of micromanagement (many videos on YouTube boasting 40K actions per minute), the level of these AIs still hovers around that of a very mediocre amateur player. At some point they may find a better way to bring that power to bear, but it hasn't happened yet.

Usually strategy games AI resort to cheating to beat humans (cheat = producing 2-3 times more with same resources).

[Leo]

Ke Jie after match statement at press conference, really interesting, from the strongest human player:

I was unsatisfied with the official interpretation of Ke Jie's comments so I took it upon myself to do a better translation. (any suggestions are welcome!): [Edited]
It’s a great honour to be able to participate in these matches. Many thanks to everyone’s support, Zhejiang Go Association, to such a great opponent Alphago, and thanks to everyone’s hard work for making this event possible.
Actually these matches, for me, are more meaningful than any other competitions I’ve ever been in. In today’s game, I originally thought I could have played a bit better. But unexpectedly, even in the opening I played some unforgivably bad hands, causing the situation to be very difficult from the start. I knew that when playing Alphago, if you start out behind, trying eke out a win is pretty much hopeless. In fact even continuing to fight was very difficult. It is really too perfect. It has no weakness, and no emotional lability. So I’ve been blaming myself for not playing a bit better. (visibly upset) (applause)
I hoped I can play a bit better. Unfortunately I wasn’t able to do that today. I don’t want to call this is a win-win situation, because it was a competition. So it’s really quite painful. Maybe I feel differently than others, because I really feel my performance was terrible. And I probably won’t have another chance like this again, so I think it’s regrettable. I hope I will continue to discover truths about Go. Through these games, especially the last one, I’ve discovered that I have too many shortcomings, and too many emotional liabilities. There are definitely problems with my understanding of Go. I feel like many other players could have done better than me in this match, so I don’t feel worthy of your praise. I’m very sorry, I’ve lost. I wish I will be better in the future. (applause)
Actually today I was flustered, not my usual self, because it was really too perfect. In the first half of the game, I already knew that I was in a disadvantaged position. First of all, whenever I feared Alphago would play a move I had predicted, it would definitely play it. Sometimes it would play a move I had not considered, and after thinking about it for a while, I’d realize that it had come up with a much better move. I can only predict half of its moves. The other half I cannot even guess. This reflects a huge gap in our abilities. There’s no way around it, I have to admit that. The gap is huge.
I’m very thankful to have an opponent such as Alphago, to allow me to know that I still have so far to go. I really wish I can improve in the future, and continue to work hard, and be able to shrink that gap a bit. But this is just a wish. I believe that the gap will become bigger and bigger in the future. It is already so perfect. Again I feel regretful about my performance today, and I’m happy for the AlphaGo team, that they’ve created such an amazing, frightening, calm, such a perfect go player. It is an incredible accomplishment to create such a perfect player. Thank you.
In response to the first reporter’s question:
I don’t know. I don’t know how I will change in the future. Actually I’ve been trying to change myself continuously. But in contrast to a great company such as DeepMind, I can only make some small transformations, limited to myself, but they will change the whole world. So I will just say that I want to make myself better, I will continue to work hard, play go, and help bring the joy of go to more people. Actually I think that playing go should be a very happy thing, just like my friends who are go players (looking at them). Playing with them is a very fun thing to do, so I guess I’ve failed to show this to the audience today. Sometimes I care about winning too much, and I have a hard time trying to control this desire. The last couple nights I didn’t sleep well, because I knew this was the last game, so I was very nervous. I was thinking about what kind of strategies I should use against AlphaGo. I was just worrying excessively causing more trouble for myself. I really believe playing go should be fun. I will try to reflect the fun nature of go better in the future. But sometimes without the challenge of win or loss, go can be more boring. For me, winning IS fun. I guess this is different from AlphaGo. I can enjoy happy moments playing go more with human players. But playing with AlphaGo, because it is too perfect, playing against me, it has no flaws, so it’s been painful for me. Before the match, I’ve said before that this is probably the last time I will play against AI. The reason should be obvious, because playing against it is really quite painful. I don’t wish to bring such pain to myself, but it is just so cool and detached, it does not give you any hope of winning. It doesn’t give you any hope at all.
Today I thought it would play me until the end of yose, and win by a little bit as usual, like by 1.5 or 2.5 moku. That would save some face for me. But it ended up taking my stones (laughs), I didn’t think it would be so cruel. Why doesn’t it let me save some face? (laughs)
It has so many qualities that amaze me. I don’t think I could possibly surpass it in my lifetime. But I can improve myself, overcome myself. This I believe I can do. As the reporter mentioned, I hope I will continue to be the Ke Jie that is flamboyant and self-confident. I hope I can also let more people know that Go is not so such an esoteric, mysterious, and unapproachable. I think it is actually a very grounded, approachable, and easy game. Thank you.

He sounds suicidal. Some of these Asian guys are taught to take the "honorable" way out.

[Uri Blass]

Ke Jie after match statement at press conference, really interesting, from the strongest human player:

I was unsatisfied with the official interpretation of Ke Jie's comments so I took it upon myself to do a better translation. (any suggestions are welcome!): [Edited]
It’s a great honour to be able to participate in these matches. Many thanks to everyone’s support, Zhejiang Go Association, to such a great opponent Alphago, and thanks to everyone’s hard work for making this event possible.
Actually these matches, for me, are more meaningful than any other competitions I’ve ever been in. In today’s game, I originally thought I could have played a bit better. But unexpectedly, even in the opening I played some unforgivably bad hands, causing the situation to be very difficult from the start. I knew that when playing Alphago, if you start out behind, trying eke out a win is pretty much hopeless. In fact even continuing to fight was very difficult. It is really too perfect. It has no weakness, and no emotional lability. So I’ve been blaming myself for not playing a bit better. (visibly upset) (applause)
I hoped I can play a bit better. Unfortunately I wasn’t able to do that today. I don’t want to call this is a win-win situation, because it was a competition. So it’s really quite painful. Maybe I feel differently than others, because I really feel my performance was terrible. And I probably won’t have another chance like this again, so I think it’s regrettable. I hope I will continue to discover truths about Go. Through these games, especially the last one, I’ve discovered that I have too many shortcomings, and too many emotional liabilities. There are definitely problems with my understanding of Go. I feel like many other players could have done better than me in this match, so I don’t feel worthy of your praise. I’m very sorry, I’ve lost. I wish I will be better in the future. (applause)
Actually today I was flustered, not my usual self, because it was really too perfect. In the first half of the game, I already knew that I was in a disadvantaged position. First of all, whenever I feared Alphago would play a move I had predicted, it would definitely play it. Sometimes it would play a move I had not considered, and after thinking about it for a while, I’d realize that it had come up with a much better move. I can only predict half of its moves. The other half I cannot even guess. This reflects a huge gap in our abilities. There’s no way around it, I have to admit that. The gap is huge.
I’m very thankful to have an opponent such as Alphago, to allow me to know that I still have so far to go. I really wish I can improve in the future, and continue to work hard, and be able to shrink that gap a bit. But this is just a wish. I believe that the gap will become bigger and bigger in the future. It is already so perfect. Again I feel regretful about my performance today, and I’m happy for the AlphaGo team, that they’ve created such an amazing, frightening, calm, such a perfect go player. It is an incredible accomplishment to create such a perfect player. Thank you.
In response to the first reporter’s question:
I don’t know. I don’t know how I will change in the future. Actually I’ve been trying to change myself continuously. But in contrast to a great company such as DeepMind, I can only make some small transformations, limited to myself, but they will change the whole world. So I will just say that I want to make myself better, I will continue to work hard, play go, and help bring the joy of go to more people. Actually I think that playing go should be a very happy thing, just like my friends who are go players (looking at them). Playing with them is a very fun thing to do, so I guess I’ve failed to show this to the audience today. Sometimes I care about winning too much, and I have a hard time trying to control this desire. The last couple nights I didn’t sleep well, because I knew this was the last game, so I was very nervous. I was thinking about what kind of strategies I should use against AlphaGo. I was just worrying excessively causing more trouble for myself. I really believe playing go should be fun. I will try to reflect the fun nature of go better in the future. But sometimes without the challenge of win or loss, go can be more boring. For me, winning IS fun. I guess this is different from AlphaGo. I can enjoy happy moments playing go more with human players. But playing with AlphaGo, because it is too perfect, playing against me, it has no flaws, so it’s been painful for me. Before the match, I’ve said before that this is probably the last time I will play against AI. The reason should be obvious, because playing against it is really quite painful. I don’t wish to bring such pain to myself, but it is just so cool and detached, it does not give you any hope of winning. It doesn’t give you any hope at all.
Today I thought it would play me until the end of yose, and win by a little bit as usual, like by 1.5 or 2.5 moku. That would save some face for me. But it ended up taking my stones (laughs), I didn’t think it would be so cruel. Why doesn’t it let me save some face? (laughs)
It has so many qualities that amaze me. I don’t think I could possibly surpass it in my lifetime. But I can improve myself, overcome myself. This I believe I can do. As the reporter mentioned, I hope I will continue to be the Ke Jie that is flamboyant and self-confident. I hope I can also let more people know that Go is not so such an esoteric, mysterious, and unapproachable. I think it is actually a very grounded, approachable, and easy game. Thank you.

I felt that beyond this facade of politeness and humbleness of DeepMind team and Hassabis, a rude attitude emerged as a cultural misunderstanding. What they basically said is that once Go was conquered, they will move to harder and more important things like Starcraft. This is ridiculous even to me. The level of Mastery of top Pro level player in Go is probably above even that level in Chess. Go is a game of 400 years of competitive history, with literature, tutoring, learning effort higher than Chess. Let's call it 3000 ELO level Chess mastery. To become a top expert in Starcraft an imbecile needs 2 weeks of hard play and learning, if tutored. Let's call this 1600 level of Chess. The DeepMind team is sufficiently rude to dismiss the human skill level, and the simple fact that Starcraft uses similar techniques and is even harder for AI would legitimize dismissal of an entire culture.

So much for fake politeness, to which I pointed earlier.

I do not know how you compare Chess and go to say that the level of top players in go is higher than the level of top players in chess but based on the age of Ke Jie I think that it is the opposite.

No chess player became number 1 in chess in an official list or world champion before age 18.

If some thinking sport is really harder than chess then I expect the number 1 in the world to be at least 30 years old because at age 20 you have not enough time to learn to become number 1.