[Open-access] NISO Recommendations on Open Access Metadata
cn at cameronneylon.net
Tue Jan 7 12:07:25 UTC 2014
On 7 Jan 2014, at 10:42, Graham Triggs <grahamtriggs at gmail.com> wrote:
> I'm only following this on the periphery at the moment, so forgive me if I misstep a bit here. But I don't see why there is a theoretical problem to machine readable standard for rights - create a model for the rights that you wish to be able to check for, then the licenses can be expressed in those terms.
Sorry you're right. Rhetorical over-reach. The "theoretical" issue is that metadata is an assertion, whereas licensing is an offer of terms for acceptance. My sense having gone over this is that the two are effectively incompatible or at least not equivalent in their underlying logic of responsibilities and guarantees. When you use a license you are bound by an agreement. When you use metadata there is no such binding. The set of responsibilities and who holds them are different. That said, I have not done a rigorous philosophical analysis, just a proof by exhaustion :-)
You can agree a subset of rights that are expressed through metadata, but then the metadata elements must in effect refer to defined underlying legal language, which means that they are effectively a license, just one expressed in a way which isn't obvious. And you still have to agree that subset of rights and what they mean...which is as we've said not really feasible.
In terms of a registry - I realise where Mike was coming from now. You're absolutely right that a whitelist of defined URLs would be better but this wasn't politically tractable. What was tractable was a requirement that the URL resolve, and that the publisher agree they are bound by the terms at that URL (ie the URL must resolve to an actual license). My assumption is that those requirements make it clear that in those cases where a CC license is used you would choose to point at the CC URL. I guess in principle someone could put a copy of a CC license on their own server but why would you bother?
What we can do is two-fold. Firstly work on best practice guidelines for linking to licenses and license expression to drive consistency (part of my work for this year) and secondly use existing registries eg the OKFN license server to build up that list of URLs in the wild and some subset of rights that matter to us. But my acceptance that this sits outside the recommendation was driven in large part by the same issue as Mike's concerns. A standards process will get bogged down whereas our community can just get on and build those tools up which will in the end be used by default.
More information about the open-access