[r-t] What is a method? (long message, sorry)

[Don Morrison]

What is a method?

It appears to be a term we currently use with at least two, different
meanings.

1) A specific kind of process for generating rows, decomposable down
into sub-sequences of changes that are applied in order. Typically we
make small modifications to these sub-sequences we call "calls". The
current Decisions spend a lot of words both categorizing and limiting
what this use of "method" applies to. The Norwich Axioms are a
different set of categorizations and limitations. Philip's proposal
was in large part a third set of categorizations and limitations. When
saying "method" a lot of folks on this list are implicitly meaning
using it in this manner. Constructions such as Dixon's are not methods
in this meaning of the word.

2) A different meaning of "method" is as any sort of process for
generating rows. It might be a "strict MBD kind of method"; or a rule
based construction such as a Dixonoid; or something none of us have
yet imagined.

Today the Decisions appear to require recognized change ringing to
consist only of methods in the sense of (1). I believe most of us
contributing to this discussion think things like Dixonoids should be
perfectly legal in peals and so on; please feel free to correct me if
I'm mistaken about this.

That seems to leave us with two possibilities:

a) We use "method" in sense (1), and allow peals and so on to consist
of things other than just methods.

or

b) We use "method" in sense (2), further defining some subset sort of
"method in the restricted sense" (which almost certainly deserves a
decent name, though in the following I'll just keep talking about
"type (1) methods").

Why would we care which of these possibilities we pick?

One of the big things we do with methods is name them. For whatever
reason, we clearly care a lot about this. The amount of intellectual
effort that has gone into worrying about extension, which is all about
naming, is sufficient evidence of this. So is the amount of effort
that goes into the various databases of methods that are maintained.
And the frequency with which these databases are consulted.

If we follow, as many seem to be implicitly doing, path (a) above, we
shoot ourselves in the foot. We are condemning anything that does not
meet the restricted use of "method" to a different name space. For
example, there is nothing today to prevent me from ringing a new
principle, completely unrelated to what we all know of as Dixon's, and
calling it Dixon's. That name would stick, and the non-method method
Dixon's would, if it's ever made legal, have to find itself another
name. While I'm amused by the notion that Dixon's would end up not
being a Dixonoid, and that the canonical instance of a Dixonoid would
not be Dixon's, it does seem unfortunate.

The use of "method" in just the type (1) sense mistakenly leads many
ringers to think that change ringing must be composed of just these
kinds of things. That is certainly result when the current Decisions
are used to force fit link methods into categories in which they do
not fit, with strange lead ends and bobs baring no relationship to
what the ringers actually rang.

But change ringing can certainly exist without type (1) methods. It
existed before such things were invented. I am no expert in the
history of ringing, but as I understand it in the beginning each way
of generating an extent was considered a distinct thing. There were no
methods and calls, in the modern definition of the type (1) sense of
"method". Those ways of generating an extent that fit conveniently
into our modern notion of methods and calls we have recast into those
terms (in the case of Grandsire, to Eddie Martin's distress). Those
that do not fit conveniently have withered and are little noticed
these days.

A further consideration is that using "method" in just the type (1)
sense devalues anything else. Whether we intend it or not, today
Dixonoids are somehow "not as good" as "proper methods" (that is, in
the type (1) sense). Which seems a shame. Any band that can
successfully ring an extent of Dixon's has accomplished something
requiring more diligence and concentration that just ringing an extent
of, say, London, despite Dixon's being so simple to describe and entirely
right place.

I think we would be far better served adopting course (b) above. Let
us use "method" to refer to any means of generating rows. And when a
band uses a new one, they get to name it. If it meets various
criteria, exactly how it must be named may be circumscribed, just as
today. But if it fits no where into our current scheme do not deny it
a name at all, and pretend it doesn't exist and is anathema to True
Change Ringing.

I'm not saying that our current notion of method, in the type (1)
sense, is a bad tool. On the contrary, it has served us well, and such
methods are good things, our daily bread and butter. But we must not
fool ourselves and pretend they are the only thing possible. They are
an example, and by far the most important one, of kinds of methods (in
the type 2 sense), but still only one example.

Some may point out that such a broad definition of "method" opens us
up to ambiguities and abuses. They may wish to have some way of
mapping from a performance uniquely to a method. But that, too, is a
human-centered process, and one we are failing at today. Is it Single
Court Minor or Original? Spliced Cambridge and St Edward's, or just
Cambridge with half-lead bobs? Cambridge Minor or Original with four
different kinds of calls? The ambiguities will always be there. We
just make things worse by pretending we can make them go away if we
just try hard enough to force fit things into a framework completely
different than that which the ringers think of them as being.

Related to this dichotomy of use of the word "method" is the long
standing angst over extension.

The root of this problem, I believe, is that extension is not a
technical concept, but a human-based, psychological one. Trying to
solve such problems strictly technically nearly always fails. The
practice of Law really is different than the practice of Mathematics.

I have tremendous respect for the folks that have worried about
extension, and what they've achieved. I have no double that Philip
Saddleton will find (or perhaps has already found) a way to make
things far better than they were. But I don't think it will ever reach
a point where it is good enough to be the real measure of merit in all
cases.

This does not mean that there is no value in trying to make methods
with the same name somehow like one another. There is great value in
this. That any one of us, if we were suddenly capable of ringing on
1,000 bells, could without a moment's thought ring Plain Bob 1,000, or
Little Bob 1,000, or Grandsire 999, or Stedman 999, or even Cambridge
1000 is lovely. I think most of you could ring Bristol 1,000, though
that would probably be beyond me without spending some time working it
all out first. And, frankly, New Grandsire 999, too! And it's not just
the value of being able to walk into a tower anywhere in the world and
being able to grab hold and ring the right thing with folks you've
never met before. It's part of how we learn to ring, and helps us
pedagogically. If as baby ringers we didn't see and weren't taught the
way Plain Bob Major relates to Plain Bob Minor the advancement of our
skills would have been even more painful than it already was.

Extension may be difficult, but it is valuable. But as it is really a
political, or human perception, problem, and not a technical one, it
is not as amendable as we would like to a strictly technical solution.

The history here tends to run along a path something like

- there is a lot of commonality in what people ring; lets codify that
  in our definition of what a method is

- we have some good examples of how methods at different stages are
  somehow the same

- by being clever, we can figure out some rules that codify our
  examples

- oops; people ring more different things than we considered, perhaps
  even than they ever did before

- fuss and fiddle and argue, and eventually rearrange the rules,
  probably adding assorted special cases, until the result fits the
  new needs

- rinse, repeat, ad infinitum

I doubt this process, if allowed to continue, will ever reach closure.

If we, I think sensibly, instead said "a method is a process for
generating rows" then it would be clear that there is no single
technical definition of what should be allowed to be a valid extension
of any method.

What Philip Earis, and I think others, have suggested is the right
response to this: a more human-centric rule on what is a suitable
extension. It needs to seem "the same, just at a different stage" to
ringers. Most of the time the band ringing it first can be expected to
make that judgment sensibly. In those few cases where, either through
willfulness or carelessness, that fails, you have some backup mechanism
for correcting things. Philip proposed the Council, which seems
reasonable.

You can even take this further, though I personally don't see any need
to do so. Once you have established that a method is any process for
generating rows, and attach names to them no matter what their nature,
we can also, if we choose, establish further rules. These rules to
limit the naming choices, by various rules of extension, for certain
subclasses of methods.

For example, we might say "If a method can be looked at as a single
hunt, non-little plain method, it just extends as follows to get the
same name" or whatever. We can codify the places that we can apply
tight rules to with some success, but limit such rules only to the
places where they work. Such rules would have to be crafted carefully
to avoid conflicting, overlapping cases, but that's easily done with
careful wording setting priorities.

This is in some ways similar to Richard's, I think sensible,
suggestion that we have as broad a set of definitions as possible, and
then, if we choose, restrict what we value by further, explicit rules
of what is or is not acceptable. That does seem a very sensible way to
break out of the current "Whatever isn't required is prohibited"
mindset.

That said, I don't personally see the need for such restrictions. But
if others do, it is certainly a better solution than our current
pretending that one size fits all. Which is, I contend, a result of
our pretending that all change ringing consists only of methods in the
type (1) sense.



-- 
Don Morrison <dfm at ringing.org>
"Anyone who is capable of getting themselves made President should on
no account be allowed to do the job."                  -- Douglas Adams