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