I played a long leisurely game today against someone from the local club -- almost 4 hours. I won playing white, but neither of us could figure out where black went wrong. I ran it through Lizzie to see if she could identify the issue, but she seems to think that white is at 60% by move 20. Was the pincer at C12 really a bad move? It seemed standard.
EDIT: Further review suggests that Lizzie thinks that black's prospects shift significantly between moves 49 and 55. I see how black's three moves are not ideal, but I'm honestly surprised that they are that decisive.
Even more strange is that Lizzie does not seem to like the standard jump and press joseki. (And I checked and it doesn't seem unique to this board position.) In this position, she strongly prefers A or B to the standard black reply. She thinks the result of the standard joseki is to give white close to a 60% winrate while either of these moves by black bring the winrate from the normal starting range to 50%. With A, she wants to push and cut. I let it play out for several thousand visits and the sequence had not settled down. She kept switching between various ugly fighting variations.
B lead to an interesting joseki-like sequence:
If someone with a fast GPU were inclined, I'd be interested to see what you get if you try to replicate this. I'm stuck with my CPU and so, I've hit the limit of what I think I can do.
Here is the winrate delta for white:
And here is the full game. Any thoughts would be appreciated!
| Attachments: |
2018-12-20 W AUS 4k H0.sgf [28.24 KiB]
Downloaded 425 times |
even before the AI era. By how much I am not sure.