[r-t] Reply to Richard of 11th January

[Robin Woolley]

Sorry for the delay, but, as Mark Davies is about to find out, sometimes
there are more interesting or important (or both) things than ringing.

Richard makes some good point of which, as is the way of things, some I
agree with, some I partially agree and others I totally disagree.

I don't think we disagree too much on what the decisions are for per se.
It's more a question the use to which they are put and how this use is
commensurate with reality.

If ringing could actually take place on very large numbers, n>>100 say, then
I would be quite happy to go along with 'indefinite' extension. However,
practical ringing is carried out for at most n=16. There have only been
about five peals rung on more than n=16 and the two surprise peals have used
the same three methods each. Since it's only three, I don't need to
look them up: York, Newgate and Feering - all indefinite extensions as it
happens from lower stages. Looking at Method 300, one can see why these have
been chosen. So, for practical ringing, anything more than 16 is, to a first
approximation, irrelevant.

Having done Latin at school, here's a simile for you. My wife once told me
she had to write a piece of code to convert a calendar date into a Julian
date (or vice versa). The Julian date is simply a number increasing by one
every day. 1st January this year was 2453370.5. This is an example of an
infinite counting system and is analogous to indefinite extension. The
corresponding finite system is to use January 1st 2005 and is analogous to
saying that indefinite extension is irrelevant in practical ringing.

This is not to say that, if ringing was an infinite-arithmetic situation, I
would be the first to agree with Richard's "I think that much of the time,
an indefinite extension series will have a more "natural" feel to it than a
single extension.". To see this, just look at the extensions of York, etc,
both in the collection and Method 300.

At their best, the decisions can be used like this., Suppose I ring a
next-stage extension of X. Person Q comes along and says that this is not a
true extension. I can then use the decisions to demonstrate that I've only
used the same way of extending as method Y was extended!

Richard uses the analogy of the C++ formal definition document of just 748
pages. This analogy is logically the wrong way round. In such a document, we
start from saying - 'we want a computer language to do this, this, this and
this, and we want to make sure that any such program we write is portable
from one machine to another'. I have a copy of the Shorter OED which runs to
over 2300 pages in two volumes. This is a better analogy for the extension
decisions in that a dictionary contains a list of words and they have been
used. New words and uses are added all the time. The decisions on extension
have changed from a list of the ways people have extended methods in the
past into a list of the only way to extend methods in the future. Remember,
the original rules were empirical. They were also asymmetric in that works
above and below were treated differently.

Graham John made some remarks about how we learn methods. Speaking from a
limited experience of this ages ago, I remember that it was much easier to
learn a set of down works for one up-work, than the other way round.

Ricahr asks for the 1989 Decision (D)E4(b). It says "Methods at different
stages in the same class which are uniquely related as in the Decision on
Method Extension shall have the same name, and where not uniquely related
one relationship shall have the same name; any relationship covering most
stages shall have preference over any others." The 1992 change is available
on the MC web-site and makes interesting reading (for the change).

Now for Richard's 'forbidding' statement, "..I do not believe it is the job
of the Decisions to make the subject approachable to everyone.  I feel this
should be left to other documents, whether written by the MC or by others."
Given the MC's record on not being able to explain its own decision w.r.t.
(D)E4(b), I wonder just who could write a revised one now. The Roker aspect
should have been made explicit in the guide. Richard remarks, echoing what I
have said before, "..people extended methods by taking what they considered
to be the essential properties of a method and trying to duplicate them on a
higher stage." I don't believe that they would have believed indefinite
extension as a sine qua non.

Something I have said before, but is worth repeating is that if indefinite
extension is correct, then the logical consequence of this has to be that no
new method can be allowed which is not indefinitely extendible.

Finally, to quote Tony Smith once more, "This new restriction (indefinite
extension) gives us the confidence to introduce several new constructions
without the risk that they will increase the number of unsatisfactory
extensions." Richard's work on the extension of Plain Minor methods has led
to the following place notation, x1x6x1x1x1; 2nds being a perfectly
acceptable extension of Plain Bob Minor to royal (4ABCD/5FG). For any one to
say "this is purely academic because PB Royal has been rung" is beneath
contempt. The fact that could be allowed at all is the point of interest.
For those who don't like this, try x4x23x25x5; 2nds as an (unrung) extension
of Buxton Bob (4AB/2EF). Both of these parents have plain hunting above the
treble whilst the allowed extensions do not. This gives a counter argument
to pabs w.r.t. Cromer Alliance. Most members of the list will probably find
the Anglian/Anglia problem counter-intuitive also. This is the same argument
as used in Anglia.

I feel that I have dealt with Richard's reply to mine of the 6th Jan.
previously. Much of this comes from the lack of explanation in the original
1953 report. Once again, we are individually having to re-invent the wheel.

Best wishes
Robin