Robert Jasiek makes a claim in another thread that he has discovered an all-singing, all-dancing New Semeai Formula. Maybe he has, but I was curious about his reference to a delta value. I have not seen his formula but his brief description of it made me wonder whether it is really just a cousin of a much older formula Delta >= F which I, in my memory associate with a chap called Lenz from Germany. Alternatively it may have come from Klaus Heine. Contrary to impressions given by RJ go theory research here has a long history, and the Koenigswinter Congress of 1979, for example, produced a large and mathematcally dense corpus of research. Can anyone else comment on the similarities?

The Lenz/Heine delta formula would certainly fall nicely into this period of 40-50 years ago. However, a version of it in plain English was given in the British Go Journal in 1970, and it was attributed to Segoe, so it may go back much further. I can say for certainty that the plain text version appeared in Japanese in Igo Sekai at least as early as 1953, and this is the first time also I have seen the distinction between shared liberties and exclusive liberties.

This Igo Sekai article did not claim any discovery, incidentally, and was attributed only to the editorial staff. In fact it was merely a response to a reader's letter. So, clearly, knowledge of how to fight semeais was much older and taken for granted.

MagicWand mentions that Asian players do not study semeais. Essentially that's right, even if put a little baldly. At least, what I have seen is that the simple plain-text version of the basic rule is taught first but then at appropriate times (e.g. in commentaries) this is suplemented with extra practical ad hoc advice in quasi-proverb form. E.g. among some I have seen (and hope I am quoting correctly) are: "fill in the from the outside", "two hanes on the edge give one extra liberty", "In an equal race to capture with one ko, take it last - but with two kos, settle one first and make yose ko", "In an unequal race to capture with ko it is yose-ko - play the ko first and use liberty filling as ko threats". These may not cover every conceivable case but they are easily digestible and useful. Overall, more useful than a formula to me.

There is a similar case with the 3-3, 4-5, 5-8, 6-12 formula. I know this is also in old Japanese literature but don't know quite how far back it goes. What I do know is that the underlying formula (0.5n(n-3), I think) has been mentioned many times but only by mathematicians. Everyone else just learns the 3-3, 4-5 sequence by heart and, as with semeais, grafts on little bits of useful extras as and when needed (e.g. corner square-4 is three liberties, cleaver-five is four).

In fact, this approach of giving a base rule and then modifying it on an ad hoc basis seems to be standard everywhere, east-west and outside of go, so I'm surprised it's being seriously questioned. Admittedly if you don't own Asian books and/or can't read them, the extra nuggets can be hard to find in the case of go.

But these nuggets do abound if you can read them. Here are a couple that especially impressed me which I haven't seen in the western literature yet.

(1) Given that you know the basics of fighting (e.g. five alive) there will come a time when you want to know how to change the direction of fighting. In that case the techniques to consider are, roughly in order: cut, nobi, tsuke, cap, inducing moves, hazama, hasamitsuke, sabaki, nozoki countermeasures.

(2) The order of play for large boundary plays (oyose): Moves affecting bases, 2nd line, thick endgame plays, moves that leave sente, gyaku yose, centre (but always assume smaller
than expected, though moves that simultaneously expand self and reduce opponent are good even in the centre.

I am not going to vouch for my accuracy in recording these, and I can't remember which pros said them. But I will assert that they are typical of a large category of titbits of advice and that they show that pros can make lists. I will also assert that represent the most natural and incremental way for most people to learn, so long as you do so via studying games and leave you brain's neural network to sort out the true hierarchy. Asian players have easy access to this fast food. Maybe that's a good part of the reason they are generally stronger.

For my semeai Classes 1+2, except for one seki case.



Simply speaking, the difference of both players' fighting liberties is compared with 0.



I call that the Old Semeai Formula. Delta refers to exclusive approach liberties. F is "forced liberties". It was nasty, requires exceptions and bad assumptions, and overlooked a few assumptions.

Later, Müller corrected part of it but introduced classes and new assumptions suitable only for computers but not for humans (some nakade are treated, other nakade are not treated).

I have stated classes useful for humans, made assumptions useful for humans, corrected assumptions, replaced the reference to two kinds of values (exclusive approach liberties and forced liberties) by only one kind of values (fighting liberties) and abandoned the nasty concept of forced liberties entirely.

I.e., I have reinvented the formula in the manner most suitable for us players.



It was Lenz. He does not recall the details of his DGoZ article in 4/1982, p. 20-21. It is possible that he was motivated by earlier Japanese sources. IIRC, he speaks Japanese or even was in Japan for some time.



I have not doubted the existence of Western seminars. There were some in the 50ies and others in Russia.



I guess I have a copy of that somewhere but cannot find it quickly. Was there anything about semeais in it?



Ok, so the history of the formula's development is longer. Without knowing the details (which liberties to count, when applicable), one cannot say whether it was meaningful or dubious.



Such proverb knowledge is weak; there can be important exceptions.



What do you do when you determine a semeai's winner? I hope that you compare some black with some white liberties. This is the task of the formula. It just does it correctly: counting the right liberties.



In fact, my book recommends this.



Everywhere where more careful researchers are not involved:)



Good lists are good - bad lists are bad.



No time to warm up this discussion at the moment.



There are several reasons why they are stronger, but weak knowledge is not a strong factor, IMO.