Sedol's wedge in game 4 against AlphaGo

apetresc · #1

So, the general consensus appears to be that this was the winning move in game 4:

Click Here To Show Diagram Code: [go]$$Wc Moves 78 to 78 $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X O X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X . . . . X O O . . . . . | $$ | . . O . . O X . . O . . X O . . . . . | $$ | . . . . O X O . . . . X X O . . . . . | $$ | . . . . O X . . . X 1 X O X O O O . . | $$ | . O O O X . . X . a . . O X X X O . . | $$ | . . X X . X . . O . X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Black responded at 'a' and white's centre group eventually broke out, at which point AlphaGo basically melted down and threw away all the aji it had in the position.

I'd love to have a discussion about whether :w1:

actually works against optimal resistance; in particular, if black responds at L10 instead of K10, whether white can do any better than a ko to live (which he's unlikely to win because of all the ko threats black has around N8). Thoughts?

Uberdude · #2

I'll study the position more and review the commentaries later, but from the bits of the AGA commentary I saw Myungwan and Haylee couldn't find a way to make the wedge work if black answered correctly. But in the press conference it was mentioned Gu Li called it a divine move or so which I don't think he would if it didn't actually work against best resistance.

RobertJasiek · #3

Of course, because his well-timed prior moves were a preparation. The question remains: god-like or brilliant bluff?

gowan · #4

Uberdude wrote:

I'll study the position more and review the commentaries later, but from the bits of the AGA commentary I saw Myungwan and Haylee couldn't find a way to make the wedge work if black answered correctly. But in the press conference it was mentioned Gu Li called it a divine move or so which I don't think he would if it didn't actually work against best resistance.

If the wedge really doesn't work if black answers correctly it must be a real flaw in Alphago's program. That wedge is an ordinary basic type of tesuji (wariuchi) which Alphago should have "seen", no? We have four very strong commentators, Redmond, Myungwhan, Gu Li, and Alphago. Two human commentators said it was a brilliant tesuji and Alphago succumbed to it. Since Alphago doesn't feel anxiety or get nervous we have to assume it made a mistake that the program made a "misread".

Charlie · #5

When I read it out, during the broadcast, I saw Black K12 as a quite acceptable response.

Click Here To Show Diagram Code: [go]$$Wc Sequence 1 $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X O X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X . . . . X O O . . . . . | $$ | . . O . . O X . . O . . X O . . . . . | $$ | . . . . O X O . . 2 4 X X O . . . . . | $$ | . . . . O X . . . X 1 X O X O O O . . | $$ | . O O O X . . X . . 3 . O X X X O . . | $$ | . . X X . X . . O . X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$Wc Sequence 2, A is Black 10, B and C are options for later. $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X O X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X . . . . X O O . . . . . | $$ | . . O . . O X . . O 4 . X O . . . . . | $$ | . . . . O X O . . 2 3 X X O . . . . . | $$ | . . . . O X . . . X 1 X O X O O O . . | $$ | . O O O X . . X 8 7 5 6 O X X X O . . | $$ | . . X X . X . . O 9 X a O . . . . . . | $$ | . . . . . . . . . . . . b . . . . . . | $$ | . . . . . . . . . . c . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Either way, I can't see how this could fail to be miles better than what AlphaGo played in the game. Perhaps someone could enlighten me.

Charles Matthews · #6

gowan wrote:

That wedge is an ordinary basic type of tesuji (wariuchi) which Alphago should have "seen", no?

Not so average, when the M11 stone is there, though. The importance of White getting the K11 stone played, in relation with an external fight to net, is what makes it unusual.

Which is not to excuse AlphaGo. This was really the first time we saw a chink in its armour. Interesting that Redmond saw :b77:

as AlphaGo attempting a simpler way; and that as commentator he hadn't anticipated Lee's play.

So the "percentage play" failed here, and for very interesting reasons. In this particular fight, Lee's sight of the board may have been well ahead of AlphaGo.

Solomon · #7

The verdict from professionals seems to be that move 78 is the move that is most likely to cause AlphaGo to make a mistake, and that indeed with L10 there is nothing White can do. Most of us already saw in the AGA stream that Kim Myungwahn 9p found nothing for W if B plays L10. For some reason, L10 was not covered in the Korean stream, but after the game, in this game analysis by Chinese pro Zhang Dongyue 5p at 19:21, he tries to find something for W in response to L10 but he can't find anything either. In GoGameGuru's commentary by An Youngil 8p, he says that move 78 "was a brilliant move which seemed to provoke a strange miscalculation on AlphaGo’s part.". All of this seems to suggest that, like move 125 in game 3, Lee Sedol hoped for a mistake from B and this was the most subtle way of doing so (so maybe this is a brilliant hamete

).

gamesorry · #8

If black played L10, I think white could do the following:

Click Here To Show Diagram Code: [go]$$Bc Starting from move 79 $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X O X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X 4 . . . X O O . . . . . | $$ | . . O . . O X 5 . O . . X O . . . . . | $$ | . . . . O X O . . 8 . X X O . . . . . | $$ | . . . . O X 6 . 2 X O X O X O O O . . | $$ | . O O O X 7 . X 3 . 1 . O X X X O . . | $$ | . . X X . X . . O . X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

After that:

Click Here To Show Diagram Code: [go]$$Bc (Cont'd) Variation 1 (black fails) $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X O X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X O 2 . . X O O . . . . . | $$ | . . O . . O X X . O . . X O . . . . . | $$ | . . . . O X O 4 3 O 1 X X O . . . . . | $$ | . . . . O X O . O X O X O X O O O . . | $$ | . O O O X X . X X . X . O X X X O . . | $$ | . . X X . X . . O . X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$Bc (Cont'd) Variation 2 (white breaks out with ko) $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X O X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X 7 5 . . X O O . . . . . | $$ | . . O . . O X X 6 O . . X O . . . . . | $$ | . . . . O X O 4 . O . X X O . . . . . | $$ | . . . . O X O 1 O X O X O X O O O . . | $$ | . O O O X X 8 X X 2 X . O X X X O . . | $$ | . . X X . X . . O 3 X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$Bc (Cont'd) Variation 3 (white breaks out) $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X O X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X O 5 . . X O O . . . . . | $$ | . . O . . O X X . O . . X O . . . . . | $$ | . . . . O X O 1 . O . X X O . . . . . | $$ | . . . . O X O 4 O X O X O X O O O . . | $$ | . O O O X X 6 X X 2 X . O X X X O . . | $$ | . . X X . X . . O 3 X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Edit: after looking at the video posted by Solomon I found I overlooked the following move:

Click Here To Show Diagram Code: [go]$$Bc (Cont'd) Variation 4 (white is still in hard battle) $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X O X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X O . . . X O O . . . . . | $$ | . . O . . O X X . O . . X O . . . . . | $$ | . . . . O X O . . O . X X O . . . . . | $$ | . . . . O X O . O X O X O X O O O . . | $$ | . O O O X X 1 X X 2 X . O X X X O . . | $$ | . . X X . X . . O 3 X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Also, not exchanging H14-H13 might be better.

Kirby · #9

I've read various articles and commentary on the game.

The feeling I get is that the wedge is not really working* to give as good of a result as what was in the game. In the press conference, someone asked Lee Sedol about this move, and he responded:

Lee Sedol wrote:

더 쉽게 수가 나올줄 알았는데, 어려웠다. 그래서 또 지는게 아닌가 싶었다. 78번째 수는 그 수 밖에 없었다. 다른 수는 찾으려해도 찾을 수 없었다. 그런데 많은 칭찬을 받아서 어리둥절하다.

I thought that it would be easier to gain profit, but it was difficult (more than expected). So because of that, I was thinking that I might lose again. Other than move 78, there was no other move. No matter how hard I searched, I couldn't find another move. So I'm kind of bewildered/puzzled that I'm getting all of this praise for the move.

---

My feeling is that Lee Sedol was in a losing position - AlphaGo played well to that point, and it came to the point where Lee Sedol had to make something happen in the center or lose the game.

It looks like Lee Sedol thought he might lose again, and spent a lot of time trying to find a way to use the aji. In the end, he selected move 78, which while apparently not really working that well, made things complicated - all other simpler paths didn't seem to work, and he thought he'd lose. The computer misread the situation in that complexity, and played a mistake. Lee Sedol took advantage of the mistake and won the game. I think that's about how it goes.

Lee Sedol won because of AlphaGo's mistake, even though AlphaGo was in a winning position - but I guess that's how a lot of games are won - by the mistakes of our opponents.

As a side note, during commentary on BadukTV, there were more than one instances where the commentators noted that Lee Sedol was playing "differently" than he'd play against a human professional player. I guess he was trying to find a weakness in AlphaGo.

In that sense, it appears that he succeeded.

* I should add that it's also possible that white does have a way after :w78:

, even if black responds in the proposed way - I don't know for sure. I don't know what I would do. If white does have a way, then I would say that it's really a tesuji. But I can't see a way, myself.

Kirby · **#10**

I've been really curious about this move, since I can't read it that well. I found a site here providing their opinion.

Click Here To Show Diagram Code: [go]$$Bc $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . 7 . 0 . 9 . . . . X . . . . . | $$ | . . . X . X 6 X 8 . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X 2 4 . . X O O . . . . . | $$ | . . O . . O X 3 . O . . X O . . . . . | $$ | . . . . O X O 5 . . . X X O . . . . . | $$ | . . . . O X . . . X O X O X O O O . . | $$ | . O O O X . . X . . 1 . O X X X O . . | $$ | . . X X . X . . O . X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$Bcm11 $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . X . O 1 X . . . . X . . . . . | $$ | . . . X . X O X O 3 . O O X . . . . . | $$ | . . X , X O 2 X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X O O . . X O O . . . . . | $$ | . . O . . O X X . O . . X O . . . . . | $$ | . . . . O X O X . . . X X O . . . . . | $$ | . . . . O X . . . X O X O X O O O . . | $$ | . O O O X . . X . . X . O X X X O . . | $$ | . . X X . X . . O . X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . 4 . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Their opinion is that in this variation, after reducing black by about 10 points on top in sente, and gaining something like 15 points on the right, it'd be somewhat difficult, but close enough that the match could continue.

---

This situation is complicated for me, so I have no idea if this guy is correct. But if this variation is close, maybe it's possible that AlphaGo ruled it out?

I'm not quite sure, but what AlphaGo opted for in the game seems like a worse result.

Either way, I guess Lee Sedol's move made things complicated enough that something didn't compute well with AlphaGo.

Solomon · **#11**

In the comments thread on GoGameGuru, An Youngil 8p posted a variation that he believes is the strongest response for Black in response to 78:

Click Here To Show Diagram Code: [go]$$Bc Black 7 is the critical move to protect against the cut at 'a'. $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X . X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . . . X O . . . . . | $$ | . . . . . O X a . . . X O O . . . . . | $$ | . . O . . O X . . O . . X O . . . . . | $$ | . . . . O X O . . 6 . X X O . . . . . | $$ | . . . . O X 4 . 2 X O X O X O O O . . | $$ | . O O O X 5 7 X 3 . 1 . O X X X O . . | $$ | . . X X . X . . O . X . O . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$Wc $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . X . . . . . | $$ | . . . X . X . X . . . O O X . . . . . | $$ | . . X , X O . X O , . . X X . X . . . | $$ | . . . . . . O X . . 4 . X O . . . . . | $$ | . . . . . O X . . . . X O O . . . . . | $$ | . . O . . O X . . O . . X O . . . . . | $$ | . . . . O X O . . O . X X O . . . . . | $$ | . . . . O X O . O . O X O X O O O . . | $$ | . O O O X X X X X 1 X 3 O X X X O . . | $$ | . . X X . X . . O 2 X . O . . . . . . | $$ | . . . . . . . . 5 . . . . . . . . . . | $$ | . . . . . . . . . . 6 . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . O . . . . . . . . . . O . . . | $$ | . O . O X O . . . , . . X . X , O . . | $$ | . . . . . O . . X . . . . X O O . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

This was also covered by Kim Myungwahn 9p in the AGA stream, who also concluded that this is not good enough for W.

Kirby · **#12**

Yeah, that doesn't look great for white to me. I actually like the other one better - just split the top in sente, take right territory, and be patient.

Game is not over, as it's clear alphago can make mistakes.

swannod · **#13**

I've heard that Gu Li and Ke Jie both found variations that seemed favorable for Lee Sedol. Is this true and if so does anyone know what these are? If not, do they agree with Kim Myungwan's analysis?

seigenblues · **#14**

I've heard from Hajin that the wedge works if done BEFORE the cut... need to get five minutes to put together to see the sequence though. Anyone manage to track that down?

Solomon · **#15**

Cho Hyeyeon 9p also said on her FB that "it didn't work even though several bad ajis were combined.". I also noticed that in Michael Redmond 9p's 15-minute game summary, he does not cover the strongest variation for Black in response to L10. At this point, I am pretty sure that 78 was a divine and the strongest bluff.

Solomon · **#16**

Actually, I think I'm being too disrespectful to Lee Sedol here to suggest that the move is a bluff or a trick of some sort. I think everyone agrees that it is, at least, the best move. When one plays hamete or trick plays in the early game, they are not seeking to play the best move, but simply trying to trick the opponent. This 78 does not give off such an intention. It just happened to be that this move is the one that is most complicated and most likely to cause AlphaGo to make a mistake. At the same time, it is really the best path for Lee Sedol anyways, so I think that explains why the move can be considered "divine". But I still stand by the fact that L10 is stronger than K10, and that if AlphaGo played L10 instead, the result would have been worse for Lee Sedol.

Solomon · **#17**

swannod wrote:

I've heard that Gu Li and Ke Jie both found variations that seemed favorable for Lee Sedol. Is this true and if so does anyone know what these are? If not, do they agree with Kim Myungwan's analysis?

I don't think this is true; here are video snippets / marks on both of their streams where they go over the move but don't go over L10:

Gu Li: https://www.youtube.com/watch?v=xgQddtHhrJk
Ke Jie (2h 42m 3s mark): https://youtu.be/3InZ3j0MgPo?t=2h42m3s

Kirby · **#18**

Solomon wrote:

But I still stand by the fact that L10 is stronger than K10, and that if AlphaGo played L10 instead, the result would have been worse for Lee Sedol.

I agree. But I think that this is how Go works, sometimes - even with an AI as strong as AlphaGo: You win the game by your opponent's mistakes.

I don't think Lee Sedol was trying to play a trick move as much as that he was trying to play the best he could come up with in the position that he was in. And in this case, it was good enough to lead to a blunder by AlphaGo.

If AlphaGo had played at L10 in response, it would have probably been a much harder game for Lee Sedol. But you can say the same for any mistakes that happened earlier in the game by Lee Sedol - had Lee Sedol played differently earlier, perhaps he would have been in a better situation before this point in the game. It would be somewhat more satisfying if there weren't a stronger refutation to Lee Sedol's move, but if that were the case, there would be an earlier point in time we could identify where the computer had made a mistake.

The satisfaction I get from this situation is that it shows that Go is a complicated game. And even if you have a distributed network of computers that have been trained against 30 million moves, having trained itself by playing against itself thousands of times... it's still not immune to mistakes that humans can identify.

Somehow, that's a bit comforting to me - at least until they come out with the next version of the software :-)

hyperpape · **#19**

Worth posting on here that An stated explicitly that he didn't see L10 while observing. So it seems that it's a hard position for professionals to read.

Cassandra · **#20**

Kirby wrote:

The satisfaction I get from this situation is that it shows that Go is a complicated game. And even if you have a distributed network of computers that have been trained against 30 million moves, having trained itself by playing against itself thousands of times... it's still not immune to mistakes that humans can identify.

Both (!) sides did what had been usual in the Four Go Houses during the Edo period in Japan:

Study the game in secret, in order to find a "new" move, which should work according to your own judgment, and which your opponent has never seen before.

Sedol's wedge in game 4 against AlphaGo

Who is online