Defining tradition (or folk, or folk music, or folklore) is hopeless, but not pointless. That is, you never get to a consensus definition, but we discover all kinds of interesting things along the way that probably wouldn't have been brought forward by any other means (that would be my defense of all the BS threads that understandably annoy some whose core interest is music, somewhat strictly defined).

On a subject very nearly related to the question at hand, the most broadly accepted definition of folklore at present is Alan Dundes', which basically says it's any cultural product shared by a folk, in turn defined as two or more people. It has the virtue of maximum inclusiveness, but fails the test of a good definition, excluding what doesn't belong (it would allow herpes, I suppose). The fashion is away from the grand theoretical systems with ambitions to universality (for example, Sir James Fraser) and towards specificity: "Motif-structure in Coming-out narratives of Ex-Amish Lesbians of Dinwiddie County during the postvietnam era"). But it's not very satisfying.

Maybe some progress can be made by trying to define what is perceived as acceptably traditional within a given community (the emic rather than etic approach), and by recognizing there are concentric zones of things more and less "folk." This has been one response to the canon-wars in literary studies -- seeing that it's not so much a matter of what's "in" or "out" but of estimating, for a given time, relative centrality (Shakespeare) or remoteness (not touching that one!)

Also useful is generating a list of descriptors, perhaps ranked or weighted, some combination of which seems to confer a measure of "folkitude" or traditionality. In oral studies, it's clear that the audiences of traditional narrative have a standard of decorum or fittingness, and Parry found that his Serbian informants would accept as traditional epic treatments of some very recent innovations (soccer) but not others (collectivizaion and five-year plans). He concluded that the subject matter had to be adapted/adaptable to existing story-patterns. A system like I'm suggesting can lead one into the illusion of precision that numerically-based analyses always court, and also to some very squishy descriptors (how do you quantify the audience' sense of "ownership" of a musical form?), but at least it breaks a generalized intuition down into more manageable segments.

Blathering. Stop. Anything useful here?

Adam