as regards the question of what do I (or users in general) expect to get:
if an actionable URI is based on a persistent identifier such as DOI/Handle or URN, multiple resolution is possible due to the functionality provided by PID resolvers such as Handle System. A user may get descriptive or other kind of metadata about the resource, or the resource itself, or whatever he/she wants and the resolver and the target system supports, as long as the thing he/she wants is available in the Web, and the resolver knows either its URL, or a URL that can be dereferenced to the current URL of the thing.
I don’t think that there is a single answer to the question of what I/we/the end users want, and therefore it is important to be flexible. Any model or technology forcing one solution only is a procrustean bed to be avoided.
Different PID systems have different ways of achieving multiple resolution, and in some cases (e.g. ARK) the solution currently supported is not satisfactory. But both PID specifications and resolvers will become stronger in this respect in the future. For instance, the National Library of Finland and the ISSN International centre are currently enriching the functionality provided by the library’s URN resolver, using URN R- and Q-components as specified in RFC 8141.
If the actionable URI is not PID-based, the additional functionality provided by PID resolvers is lost. It may be a weeny bit difficult to support resolution from one URL to many relevant URLs (multiple resolution) via HTTP dereferencing only. But that should not force us to specify any kind of default link behaviour or preference in Bibframe or elsewhere.
Can I bring the discussion back to the original question, whether a URI, supplied as the value (rdf:value) of bf:identifiedBy, should be encoded as a literal or an actionable URI.
I believe the question is probably irrelevant, because, I still believe, there is no practical purpose served by doing so.
Let me ask this: if it is to be an actionable URI, what do you expect to get upon dereferencing it?
· Do you expect to get a description of the resource? We’ve established that that’s not an appropriate use of an identifier.
· Do you expect to get a copy of the resource? Not only is that also not an appropriate use of an identifier, but there is an existing bibframe property to do that.
So what do you expect to get?