Life In 19x19

Does anyone know if this match will be available via some stream in the west with decent commentary?

Kim Yoonyoung is gonna stream it on Twitch tonight.

Didn’t see the game live, but checked cyberoro and it looks like Lee Sedol won the first match.

Lee won in under 100 moves, he captured Handol in a net (in which some surrounding stones couldn't be captured in a ladder, which might have confused the bot). But it still seems a kinda dumb way for the bot to lose. For example LZ can see the net after just a few thousand playouts (I would expect at least 100 thousand for a match like this), and for example wanted to play m4 at l7 instead which would have avoided the shortage of libs leading to the ladder/net of the game. Nevertheless, Lee's move 78 (again the magic number!) was a nice tesuji. Does anyone know how strong Handol is? The first I'd heard of it was this match. There was another Korean bot DolBaram which has been a regular of bot competitions which I think is weaker than LZ.

Well, in 'Yoon's Baduk Cafe' there is a video of a review of Handol vs Leela Zero where Handol won

https://www.youtube.com/watch?v=4U6199vSWZY

Handol got 3rd place in this 2019 AI championship

Uberdude wrote:Does anyone know how strong Handol is? The first I'd heard of it was this match. There was another Korean bot DolBaram which has been a regular of bot competitions which I think is weaker than LZ.

Lee Sedol, if I recall correctly!, did say a month ago that he expected to lose all the games against HanDol. So my expectations of HanDol were higher.

Lee Sedol, if I recall correctly!, did say a month ago that he expected to lose all the games against HanDol.

Possibilities include:

• He wasn't sufficiently informed ( re: AlphaGo ) ;
• He was being diplomatic ;
• He was managing expectations ;
• He was being truthful ;

Yeonwoo already has a video on the game:

Lee Sedol reached super-AI strength.

Lee Sedol = John Connor

at L5 is a blindspot of Katago.

But even without this blindspot, Katago believe black played nicely, losing very little of his initial advantage

Some of the Korean articles are noting that move 78 was a clever move; I guess it makes a headline due to the same move number for gave 4 of the alphago series.

Regarding strength of HanDol, some articles are saying it's above ELO 4500? (e.g. https://news.joins.com/article/23660254) Says it's stronger than AlphaGo Lee and the master AI that Ke Jie played.

Not sure how they measured that, or if it's accurate, though.

4,5XX is still lower than 4,858 of AlphaGo Master, but they measure in different pool so...