[r-t] Consultation

[King, Peter R]

The problem seems to be a mixing of different concepts. On the one hand is the argument that place notation is at the heart of method construction (I don't dispute that) and should be the basis for method extension. I could live with that, but then there are the caveats. E.g. providing it doesn't introduce "new work". What exactly does that mean. As I have said before a piece of work is not a well defined concept. Is 4 blows behind new work or is it just two lots of 2 blows behind. We have no objection to doubling up on dodges (turning single dodges into double) but we do object to doubling up places. Also we do seem to allow new work if we argue that the parent method is an example of zero occurrences of that piece of work (as in Bristol).
There are other ad hoc "rules", if the parent method has a regular lead end then so must the offspring. Again Why? There are no requirements irregular lead end methods to fall into particular families. Or the rule about indefinite extension. What is wrong with a family of methods that only exist in finite sets (possibly with only one method - those methods that really don't extend).
I would be perfectly happy with method extension if it really was "algebraic". Take this place notation, do this and only this to it and then live with the consequences of whatever method it produces. It is the arbitrariness of the additional constraints that I would take issue with. Especially since they seem to rely on poorly defined concepts, like "piece of work".
I also dislike concepts like a "decent extension". There are all sorts of methods at different stages that I personally think are not "decent" but that's my opinion, I don't have to ring them, I don't see why I should inflict my views on the rest of the community. If there is a straightforward, unambiguous implementation of method extension then use it but don't introduce subjectivity, leave that to bands and what they choose to ring.