Book Club. Tesuji by James Davies, Chapter 5 When Liberties Count
The material in this chapter challenges my reading. OK, it is an insuperable challenge to my reading. And I think I need a different kind of explanation than this text offers. In this case, an explanation that tells me how to count the liberties and virtual liberties. Clearly that’s a direction my study should take, soon.
I’ll highlight a few heuristics that the chapter pointed out.
EYES - in a capture race, don’t let your opponent form an eye. especially not a big eye. for you, form an eye!
Safety Play- what makes something a safety play? He doesn’t say. Sensei’s Net has some help for this: Safety Play at Sensei's Library.
“James Davies calls a “Safety Play” something entirely different. In his usage, a safety play is a tesuji that eliminates all possibility of resistance …”
So the common element to these is … that they work really well?
Two Hanes Makes One Liberty - now that’s a helpful heuristic!
I guess the benefit of the book is any updates, typo fixes, or a revised presentation etc. I haven’t really compared the two properly.
In short the idea is to count how many moves you actually have to play to capture the stones and take them off the board, that’s the real idea behind liberties, virtual or otherwise.
There’s lots of examples where things that look like liberties aren’t liberties, like if you can’t afford to give up stones connected by a bamboo joint, or if you can squeeze liberties with a throw in and atari etc.
There’s other examples on the first line or in corners, or shortage of liberty type situations where one side has to connect stones before they can take away a liberty and so those moves count as “liberties” in the above sense.
Similarly there can be “shared” liberties between groups, but in the case of one eye vs no eye capturing races, what happens is that one side can have a liberty which has to be filled last (the eye) and so it turns out that shared inside liberties are kind of one sided in a sense.
All the above can be qualified and clarified with examples, but I might do that later - in any case much of this might be in the articles I linked
I’ll start reading Chapter 5 and see if I can understand what is meant by safety play
I strongly identify with both this statement and the first few sentences of the part 1/the book chapter 1 above
Reading out capturing races is easy. It’s usually just a matter of comparing two single-digit numbers. However, in practice, even dan players have trouble with numbers greater than three.
The bad news: the magic bullet you’re looking for doesn’t exist. Capturing races are hard. You can invent formulae that work some of the time (and Richard Hunter’s book does a good job there). But you always need to check for hidden traps. It’s a combination of reading plus seeing enough examples that you’ll develop an instinct for when something fishy is going on.
The good news: it’s not just you. We all make mistakes on this one. There’s a reason why chapter 5 is the longest chapter of the book!
Just checking that you realise: this James Davies book (and also his Life and Death book) is not in order of difficulty! Some of the problems are beginner level, and some of them are challenging even for dan players. The level goes up and down fairly randomly as you turn the pages. It’s more of a sampler pack of techniques rather than a graded course. A book you’ll keep re-reading over the years, and each time you come back you’ll get a little more out of it.
I haven’t finished the chapter yet, but I was pretty excited to learn that a four-space eye takes five moves to fill and that a five-space eye takes eight. Smaller eyes just take n moves, so we have
En = 1, 2, 3, 5, 8, …
That looks familiar. Is it the Fibonacci sequence?
Turns out it’s not, a six-space eye takes 12 moves to fill, not 13. (Assuming the other player doesn’t form two eyes.)
I suppose it’s because the sequence is given by
En = En-1 + n - 2.
And you can conveniently sum up the terms for large n as
En = (n - 1)(n - 2)/2 + 2.
I can’t imagine how this will ever be useful, though. I’d love to see capturing race problem where you can use all 38 liberties of a ten-space eye.
I guess the problem is the combination of big eye and capturing race.
There’s probably various extremes. The small cases 5,6 and below there are killable shapes with nakade so one should be able to cook up capturing races that are close for those.
Around 7 or so, a lot of the eyespaces turn into seki or outright life if you try to fill in from the inside. So in a lot of instances if it’s a proper capturing race, one side will just be alive and the other dead, or if it’s kind of symmetric and both are alive then it’s not really a capturing race.
There’s still a few special shapes and cases, but I’m not sure they strictly involve something similar like a nakade.
