[r-t] Extension

[Mark Davies]

OK, so I am definitely in favour of "Extension" as an idea. I like the 
fact that quite a wide range of methods seem to have children at 
infinite numbers of higher stages which, in some sense, look like their 
parent. It's a happy part of our ringing culture.

Now, I am most certainly with Philip Earis in that I don't believe there 
is a single, fixed way of generating extensions. But I'm not with him, 
Peter and Alan in thinking this means we shouldn't try and find good 
algorithms for generating extensions.

As I've said before, if someone comes up with an extension to a method, 
and can demonstrate a good way of deriving it, that is excellent. 
Changeringing is richer if we can find these connections between stages. 
However I'm not so happy if someone says "I want this method to be an 
extension, but I have no justification for that". There needs to be a 
rationale, and it would definitely smell funny if there was a serious 
place-based bit of work in the extension that didn't exist in the parent.

The current regime, as encoded in the Decisions, is actually a pretty 
good way of generating extensions, although (a) it is codified in 
impenetrable wording, (b) there seems no public software to implement 
it, and (c) as I've said above, it ought to allow for alternatives.

To follow up my previous example, if Lower Snotscommon gets augmented to 
ten, in an ideal world there would be an automated way of finding a 
suitable Lower Snotscommon Delight Royal that the local band could ring. 
It would be even better if there were multiple algorithms available, and 
the potential to create new ones, as long as we accept that, no matter 
what extension scheme we devise, ultimately there might not be anything 
suitable. But use whatever creativity you can bring to bear on the 
subject to find one.

MBD