liberty discussion (version 7)

2003 June 5

mAsterdam: A few days ago I made some remarks on liberty. This triggered a lot of interesting remarks from well respected deshi. Bill even started a re-WME.

At least one issue has not yet really been adressed. I stated "...this is not a two different meanings issue at all." and later : "...Maybe a better analogy is what happens to our understanding of numbers when we ar confronted with negative numbers for the first time in our lives. And later on with complex numbers it happens again. First it is confusing, one has to rethink the concept of numbers. But although there is (admittedly) new meaning to number, it is not so that suddenly there are two (or even three) meanings to the word. The new understanding encompasses the old. Not unlike here. Go players use liberty in the (extended) sense of 'move to capture' when discussing boardpositions, and dame only when they explain the basics to beginners, for instance BobMcGuigan: '... I note that when my friends and I are informally discussing these situations during a game review we usually just use the word "liberty", without distinguishing primary from secondary.' Though the liberty page still opens with "Liberty has two different, but related meanings in English", this puts way too much focus on what should be a side-issue. Furthermore, from what I learned from this discussion, I still think it's not a difference in meaning. So what type of difference is it?

I'll try to expand on the numbers analogy by building a small Go-knowledge pyramid.

At 30 kyu a player knows the ko-rule. He doesn't understand the implications, but the rule is clear enough. To understand this he must already know the rule of capture. The rough understanding of liberty as dame is good enough for this purpose. He learns about two-eyes and ladders, finds out that it is easier to make life when supported by the edge, and gets to 28 kyu. After losing much terrain he thought was his by not guarding his cutting points, he starts to do just that, sometimes with a hanging connection. On the road to 25 kyu he learns about the net tesuji, a bit of semeai - he starts counting -, nakade and damezumari.

Somehow the concept of liberty is refined already. When talking about board positions he uses the term liberty freely as moves to capture? without making this distinction explicit.

Now writing about liberty poses this dilemma: the deshi do not want to scare off the beginner (so they want to explain liberty as dame). But they also want to adhere to the (slightly advanced) practise, so use liberty as moves to capture?. At most places there is no problem at all. Only when defining liberty the problem surfaces.

The difference at hand is a difference in didactic opinion. Some try to escape the dilemma by blaming the supposed inadequacy of language: there are two meanings ... etc. Others try to refine with primary and secondary. This however leads to unexpected difficulties in other areas. How does a net affect the liberty count?

I would like to propose this: The explanation of liberty starts (maybe after a few examples) with a simplified definition for didactic purposes, and, when reaching the extended one, shows respect for those who adhere to the use of the simple definition only.

Summary: I still object to starting with "Liberty has two different, but related meanings in English". I propose to start simple, complicate slowly (as Bill does), and leave the difficulties to where they arise.

Moved here by Charles Matthews, from Messages to People Currently Present in the Library of recent days:

2003 June 4

BobMcGuigan: I see the problem of "secondary liberties" as being, yet again, a matter of context. I like best the idea of "moves to capture" because it emcompasses everything with one term and factors in the context. Given the bizarre shortage-of-liberty shapes that can occur near the edge it might be necessary to make a move 10 or more spaces away from a group in order to capture it. In other words this is a property of a position rather than a group. Some of the same issues come up in approach move kos. There, too, the key idea is number of moves needed to reach a certain state on the board.

Charles I have to say that in introductory writing on go, using 'group' sometimes to mean chain and sometimes not, is disastrous. Not so for most players. Isn't this the same kind of distinction?

A group  

If you make the liberties of the stones depend on those on the stones (which are in the same group, naturally), you have also to answer such questions as : in a capturing race, does White need to save the chain?

Bob's suggestion that we need position-context, which is quite true, makes a third level chain/group/position.

mAsterdam Is there a good reason to continue this discussion here instead of at secondary liberties discussion? or liberty discussion??

Another note: I feel a little guilt and a little pride at the same time for re-opening this can of worms. Fools can ask questions the wise cannot answer.

Charles Yes, tempo might be a step in the right direction ...

Bill: I have started a WME of the Liberty page, adding mAsterdam as an author. Thanks to him and to our subsequent discussion, I have realized that the discussion of the first sense of liberty (dame) confused it with the second sense (moves to capture). I have completely redone that section.

I also see that Charles Matthews is not using secondary liberty to mean move to capture, but only such moves as are not dame. That is a useful distinction, although I might prefer a more descriptive term, such as hidden liberty, or some such.

The edit is still under construction and I have things to do now. Please feel free to make comments, suggestions, or improvements. :-)

Charles The obvious definition is like liberties sense 2 - liberties sense 1. We have one problem: that this is like pears - apples. That is, sense 2 is a metric: it measures in time units how long it may take to capture a group. While sense 1 is a cardinality, attached to a set of adjacent intersections of a chain. In a number of cases we can get away with saying sense 2 comes from sense 1 by counting in some multiplicities; and in that case I think it makes sense to speak of secondary liberties lying at those points that need to be counted twice or more. But you might say that things are generally much worse than that. For example the saying 'a false eye liberty is -1 liberty' - the case where the opponent can force you to fill in a false eye in sente - refers to an effect like a negative secondary liberty,

Bill: I agree that it is like pears and apples, or at least like different varieties of apples. Therefore my revision has sharpened the distinction.

Jan: Charles' mention of multiplicities set me thinking about Four is Five and Five is Eight and Six is Twelve. I wonder where those 1, 3 or 6 extra liberties are: for example in a square 4, does each of the real liberties count for a quarter virtual liberty? The vital point in a rabbity six would have a quite high virtual liberty count.

Just an idle thought...

Charles They are there within the group - but unless you are in a case where there is an essentially unique way to fill in internally, you might have to 'average' them over ways to fill in. Messy ...

Bill: If you keep to the idea of number of moves rather than points, it seems to me that you do not have the same question, or it does not have the same significance. Different moves may achieve the same effect, no problem. Exactly where they are does not matter, unless they have other effects that depend on placement.

2003 June 3

HolIgor: Perhaps, because I come from the country where chess is very popular, I always understood and called what as named here as "secondary liberty" as tempi. White group has two liberties but three tempi to capture.

2 liberties 3 tempi  

On white's turn white wins semeai. On black's turn black can form a group with 3 liberties but 4 tempi to capture, because white cannot attack from the corner. Her group has only two liberties. Damezumari.

jvt About secondary liberty, is it commonly used terminology? I only saw secondary (and ternary and nth order) liberties mentioned in the computer go literature, with a different meaning, i.e. empty points within a Manhattan distance of n from a chain.

Charles I suppose all the discussion is because standard terminology has in the past not supported the distinction. I'm pretty sure that Richard Hunter does not use it, because of his attitude to how people should count liberties in practice.

I'm not sure that tactics for gaining liberties (either sense or both) are discussed anywhere on SL. Certainly not systematically, which would be an addition to world literature, I think.

jvt Yes indeed, but I'm a bit surprised by the use of the word "secondary" in this sense, thus my question. Does it not imply there can be ternary, etc.? What is "secondary" in an approach move? I would prefer "virtual" liberties or some other non numeric adjective.

Charles But they aren't 'liberties' anyway, they are 'moves' in a fight. By saying this I'm in danger of re-opening a discussion that I have backed out of once already. My point of view is roughly this: I would never mention 'secondary' liberties to complete beginners; then you certainly need the shortage of liberties concept to study tactics at all seriously, and the approach move is a good example at this point. For players around 10 kyu, perhaps, thinking of 'extra' liberties caused in certain ways should be clarifying. So here, with the pedagogic purposes we have on certain pages, we can talk about secondary liberties as a kind of working concept - not very computable or at least not trivially computable. Everyone hates 'virtual'; well, I exaggerate but there have been comments before, about virtual ko threats.

I have to say that it is groups as a whole that have these 'secondary' liberties, not chains (as far as I can see). Or at least for a given chain you have to label adjacent chains that are or might be connected in some way ...

Richard Hunter I do not use the term secondary liberty, although you are welcome to use any terms you like. I had never seen this term when I started writing about liberties. I use physical liberties for the actual points adjacent to stones and then add approach moves to get the liberty count, which is how many moves it takes to capture the stones. I intend to post some definitions of the terms I use here on SL soon (in the next few weeks). My new book Counting Liberties and Winning Capturing Races has been published and should be arriving at the usual distributors in Europe before long.

mAsterdam: It is at het Paard. I'll buy it next saturday.

Bill: As Charles says, they are moves. I propose to call them that. Certainly not "secondary" liberties, moi.

(Later note): I thought the term "secondary liberties" was being applied to all moves to capture. I do not object to a more specialized meaning, as indicated above.

Dieter: I don't recall how it all started but I do remember I was so thrilled by Richard's writings in the BGJ that I became an ardent advocate of using "liberties" in its semeai sense. By actively proliferating this usage throughout relevant pages here at SL, I may have foregone the problems it brings to beginners.

mAsterdam to Dieter: (In het Engels, dan kunnen de anderen meegenieten): Judging from the older pages this has been going on a while. I have re-awakened an old unresolved issue when I failed to find (or make from what I found) a definition of a term. This issue had a fake resolution by dealing with it as a language thang - two meanings of a word - instead of using the power of language to deal with it - admittedly partly an ambiguity thingy.

I am just a beginner at go (I did not play 19x19 until february), and senseis is an important source to me. Sometimes when there is something I do not understand I get angry. Unlike with books (which I also like), here at senseis this anger can transform into into sensible texts for me and other novices because of (just to name a few) Charles, Bill, unkx80, Andre Engels, Arno and your kind self. You really help me understand. Thank you.

Bill: Dieter, you are hardly to blame for calling moves to capture "liberties". I was caught up short over 20 years ago when I used the word in that sense to a friend who had learned go from a book, and only knew the dame sense of the term. I could not recall when I started using liberties for moves to capture. That's a sign of normal language learning. I picked it up somewhere. The same linguistic ambiguity exists in Chinese, as unkx80 explained, though Japanese uses two different terms.

Are we justified in calling such moves liberties? Yes, indeed. Consider a dame liberty. As long as a chain of stones has at least 1 of them it stays on the board during play. Why do we count more than 1? Very simply, to capture the stones unless they are added to, the opponent must fill each of them. The dame count tells us the minimum number of moves he must make to capture them. The reason for counting liberties has to do with that fact. So the underlying idea in talking about liberties (plural) rather than simply a condition of being captured vs. not has to do with how many plays it will take for capture. Such plays are rightly called liberties.