Following Iyama Yuta (no world ranking discussions)

[Uberdude]

There is some English text and video commentary on the Meijin title match Iyama vs Cho from Michael Redmond here: http://www.asahi.com/special/timeline/4 ... meijinsen/. A big ko trade from a 3-3 invasion joseki to begin.

[Uberdude]

Quick Elf v1 commentary:
- pretty much every white choice in the lower left joseki was a mistake (block wrong side, jump should be hane at head of 2 and double hane or extend, hane corner should be extend, cut inside should be double hane outside) so after the squeeze Iyama was 68%. A good case study for Gomoto.
- Elf not fan of e5 hane though.
- wanted e2 as c8 ponnuki (it's sente or else a seki in corner) and then take g4 ponuki and sacrifice corner. I think hard for a human to sac like that given you don't get a seal and black outside groups not alive yet either.
- Cho's large ko threat creation at to left lost too much
- Cho should answer m18 invasion with m17 to keep connected instead of make ko direct
- Iyama answer Cho's threat slightly better, but taking stones not a big mistake (86 vs 84%). Black's next threat (ignored) would be l16 which makes a big upper left area seeing as top left threat was answered.
- Elf really didn't like the whole sequence of Iyama starting ko lower right. Preferred q13, seem to be because it's happy to take right side and thinks it can sabaki in white's lower side later starting with k9 reduction after white connects along P line.
- f16 ko threat lost a lot as it wanted to save those 2 stones with empty triangle later (it's painful to extract them but once you have white's cutting stones are floating in a black sea.
- p8/o7 not valuable and white is winning at move 115 (should have been peep not block).

Edit: game record:

[Uberdude]

Here's a playlist of Michael Redmond's commentaries and adventures with a parasol:

https://www.youtube.com/playlist?list=P ... SkTVURQQ0S

[Bill Spight]

About the ko in the bottom left corner. Once White makes it a regular ko, it is big. Each ko play gains almost 16 pts., which is bigger than the first play in the game. However, as an approach ko its size depends upon who, if anyone, is komaster. If Black is komaster, White can capture one stone with sente, which takes away one potential point. So it's a 2 pt. sente. If White is komaster we divide by 4 instead of 3, so each ko play gains a bit less than 12 pts. (According to John Fairbairn, pros know how to calculate the value of this approach ko when White is komaster, so we may assume that Cho has made the calculation. A 12 pt. play is probably worth playing at this point in the game, or at least the loss is slight.) However, White is not komaster. Probably neither player is. White certainly ignored a ko threat to convert the ko to a regular ko. If neither player is komaster, then the value of a ko move lies between 2 pts. and 12 pts. How to evaluate the ko depends upon the rest of the board. A reasonable guess is that each ko play is worth only around 9 or 10 pts. Which means that White should not have started the ko at this point in the game, and White should certainly not have taken a local loss in the top left corner to make a ko threat or two. It might have been worth losing a couple of points in that corner if White could have become komaster.

[Uberdude]

Cho U won the 3rd Meijin game by 4.5. Iyama sacrificed some stones for a wall which Elf and LZ thought was too much and there was quite a lot of ups and downs and kos and trades but Cho held on for the win.

[Uberdude]

Some more international participation from Iyama: he just won in the round of 16 of Samsung cup, beating Li Xuanhao (a rising Chinese pro you might not have heard of but he's top 30 in world now). Tomorrow he plays in the quarterfinals, will be a tough game, against Xie Erhao (who just knocked out Park Junghwan).

I was surprised Li never moved out with the n15 cutting stone (I've seen that in other games and gives black some future problems, Elf thought he should at various times, p13 at p14 seems more common recently and then black doesn't feel need to add n13 to prevent o13) so when Iyama captured that I liked his position. Interestingly Elf identified s6 as a point white should have invaded the right side moyo that just became a territory without the lower left getting in too much trouble.

Thanks mace for quickly adding to go4go!

[TheCannyOnion]

Some more international participation from Iyama: he just won in the round of 16 of Samsung cup, beating Li Xuanhao (a rising Chinese pro you might not have heard of but he's top 30 in world now). Tomorrow he plays in the quarterfinals, will be a tough game, against Xie Erhao (who just knocked out Park Junghwan).

I've been following Iyama's performance at Samsung cup, and I must say that he lucked into an easy path. His opponents were relatively weak compared to what Korean and Chinese pros had to face. Anyway, Iyama's luck has just run out; he's been eliminated.

[Uberdude]

4th Miejin game is on, here's the first day's play. Some complex fighting which ended up with Cho U getting a load of territory and Iyama a wall. I had a quick look with Elf and LZ 181 but with these kind of fighting with ladders involved I don't trust them too much without doing a lot of exploration and validation. Anyway, after the dust settled both AIs think it's essentially even (Elf slightly more leaning to white but the bots have 7.5 komi).

[dsatkas]

Wow, this is an even position?!

[Kirby]

Funny, because I preferred black. I guess the left side has a lot of potential for white.

[Uberdude]

Iyama won: the centre potential ended up too much. When Cho jumped at o9 Iyama poked the shape at q8 and Cho never connected with q9 (I suppose he thought it submissive and didn't like white sabaki with q4 next) and Iyama ended up cutting off o9 whilst Cho solified the right side and corner territory but Elf says o9 was too big. Cho's reduction/invasion of the left side ended up living in combination with the h13 cut but Iyama got more than enough in the centre.

[Uberdude]

5th Meijin game is in progress today: opening showed the influence of Elf with lots of ignoring approach to 4-4 to approach another 4-4.

Update: lots of big trades in the late middlegame meant Cho U won by 9.5 to stay in the match with Iyama 3 wins, Cho 2.

[sorin]

lots of big trades in the late middlegame meant Cho U won by 9.5 to stay in the match with Iyama 3 wins, Cho 2.

It was quite strange to follow live using LeelaZero: before Cho U connected his upper and lower groups, LZ seemed confident that an attack on them will result in Iyama's win. However Iyama chose to play away, let Cho U play some moves in a row in the center to connect his groups.

I wonder if Iyama misjudged the position (he thought he would win by playing peacefully), or is it that LZ miscalculated something in the fight in the center (for the attacking variations that it had "in mind").

[hyperpape]

I wonder if Iyama misjudged the position (he thought he would win by playing peacefully), or is it that LZ miscalculated something in the fight in the center (for the attacking variations that it had "in mind").

9.5 seems to imply a huge miscalculation, for that far along in the game.

[Uberdude]

I wonder if Iyama misjudged the position (he thought he would win by playing peacefully), or is it that LZ miscalculated something in the fight in the center (for the attacking variations that it had "in mind").

9.5 seems to imply a huge miscalculation, for that far along in the game.

Compared to that peaceful position where Iyama allowed Cho to sort-of connect through the middle the following changed by the time they got to counting:
- Iyama's bottom ~60 point group died (so Cho's big bottom group didn't)
- Cho's top ~35 point group died (so Iyama's top left didn't)
- Cho's right ~15 point group died, rescuing Iyama's 4 stones, leading to
- Cho's top right ~15 point corner died

I was impressed Cho found 202 as a clever response to Iyama's clever 201 allowing him to engineer the favourable trades.