Lessons in the Fundamentals of Go, Page 171 has this problem and recommended moves:


"If White strikes at the key point with 1 (2 in my diagram), he can penetrate deeply and finish with sente while Black barely manages to make two eyes in what had seemed a large territory."

But what if white cuts at A? Surely it's better for black to play A than to descend at 7 in my diagram? Then white will escape out underneath and black and save to playing at 1,2. It's a difference of 5 points or so, isn't it?

Please help me understand this problem. I've come back to it twice, ran it through MFOGv12's "Computer Solves Problems" (which has sorta confusing output) and I don't get why black would respond this way.

It probably doesn't affect the analysis, but the original problem has an extra stone at R12 (W's shape much thicker.)
And we probably think of more as a connect than a descend.

I think it really depends on what stage of the game this happens in. Those two black stone are big in endgame terms but earlier in the game it would likely be a wasted move for white to capture them, so black might as well leave it as a ko threat.

In the description of the problem, he says "It would be a pity for white to hane at a (P19) in the endgame."

What I don't get is why black plays S16 instead of Q17, or if unconcerned about the two stones, just play somewhere else instead of S16. S16 leaves those two black stones vulnerable if white takes Q17. Whereas if black takes Q17, white has to go crawl out at the top immediately and black is safe. And as far as I've read, if black doesn't respond to white's 3rd stone (S19), then black is OK and can tenuki. Or am I missing how S16 prevents white from a major kill?

And yes I forgot a white stone at R12.

Tenuki is not an option in this case. If white were to play S16 the whole black group dies.

As to the question why not play Q17 before - it looks like a case of Aji Keshi to me. The actual two black stones are not that important, especialy if this occurs during the early stages of the game. For the same reason, white will probably not play Q17 himself which means these two stones can still be saved by black in the future.
If played directly white will answer Q17 with P19 and then black still has to play S16 to live. If this is left open later black may play Q17 to save these two stones, but white may not directly respond with P19. In that case O19 becomes a big endgame move for black, an option that would have not been there if the exchange was played to early.
Last but not least, Q17 and the threat to capture the two white stones may be used by black as a ko threat.

All in all there a number of good reasons to not play Q17 directly and just go straight for the important S16.

_________________
Magics way up the hill (personal study journal)

As others said white cutting is gote so black may wish to save it for a ko threat rather than saving the stones in sente before living. Which is best depends on the stage of the game and how big other moves are and the ko situation. It's quite a subtle issue.

However the much bigger mistake is your thinking that t18 lives with more points. It's good that you want to live with maximum points but unfortunately it doesn't live. Can you see how white can kill now?

Yes now I see the problem - how embarrassing to have spent so much time and written a post without noticing :\. And now I also understand how it may be more valuable to keep sente than kill those two stones. Thank you everyone.

http://www.go4go.net/go/games/sgfview/41747

Have a look when black plays e6. This is like your connect in that it saves some stones in sente where the opponent could capture them in gote. The threat if ignored is bigger though so it can be used as a ko threat for a bigger ko.

P.S. it seems macelee now requires registration just to view on go4go, an unfortunate step that means I will link to his site less as a source for games.