The main problems in this discussion appear to me to be:
1. The inherent conflict between the concept of copyright and that of common property. As Stephen Sedley, (now Lord Justice) said in an article in Ewan McColl and Karl (Fred) Dallas' magazine'Folk Music', the law of copyright does not adequately handle the concept that a song can have common ownership;
2. The concept of Public Domain is an American legal one. As far as I know, it does not exist under European Law;
3. The usual reason for considering a song with a known recent composer to be traditional is ignorance of its origin;
4. It is however, a misconception that a song which has a known author cannot be considered traditional. Here are the names of three songs demonstably composed before 1831 by a mid-Antrim (Ireland) poet named Hugh McWilliams - The trip over the Mountain; When a Man's in Love; The Lass among the Heather (look in the Bodleian under Blooming Heather) and what about Laurence Price, Martin Parker John Morgan etc.?

In my view, this business of 'folk' or even 'traditional'is a matter of misapplied concepts. It is all about what people sing - if enough people sing it intensively enough or for long enough, it takes on the appearance of traditionality. What matters is what people do. I know this is not neat, not amenable to facile definitions. But what interests us is singing. Why not remake the definitions according to ideas that derive from singing rather than from whether a song is old or recent, has an author, or a claim to authorship, is common or rare. In other words, singing engages our subjectivity but our concepts are objective. 'The tradition' if such a thing exists is made up of the subjective choices of myriads of singers. Why not judge it in a similar subjective fashion.

I may seem to be out on a limb but I suggest it's a more substantial limb that the one that most others are sitting on.