Interesting, but not the focus of what I was trying to convey – that the Linked Data web has the potential to be a distributed database; in which case an entire database may be published and queryable without any infringement taking place.

The second precedent establishes that it is possible for a table of numbers to be a protectable expression, if they represent the result of creative effort. But PAF isn’t that either.

The commentary clearly indicates that a numbering system cannot be protected by copyright, even if it is original and non-obvious- that would be the province of patent law.

@Jonathan you’re right, there is a degree of speculation in my post about what would or would not be considered infringing. Without precedent it’s difficult to say, but I think it’s worth exploring the notion that Linked Data changes our perception of what ‘publishing’ data means – as a result of being able to rejoin all the parts to form a whole.

Depends on exactly how “creative” or “original” the numbering scheme is.

For instance, in the famous case that firmly expanded that “facts” (and thus databases) are not copyrightable in the US, Lexis’s page numbers were considered not copyrightable, as they were just, well, page numbers. Incremental integers assigned one after another to pages. No creativity or originality whatsoever.

Now, if a system has _some_ creativity or originality, does it have enough to be copyrightable? A judge would decide. And then it would be appealed and another judge would decide, who knows.

[But, incidentally, while I'm not a lawyer, I find OCLC numbers, incremental integers assigned in order of receipt of a record by OCLC, to be _quite_ analagous to page numbers, with the obvious implications. ]

That case settled several things. Firstly that the effort required to produce a database did not give it protection (no sweat-of-the-brow protection) and secondly that the intention to be complete made the database unoriginal in terms of Copyright as it had no editorial selection, only a mechanical process to collate it.

I wasn’t really thinking much about the OCLC vs The Library Hotel Dewey case, you’re right that that was a trademark case. OCLC claims Copyright over the labeling of the numbers, but not the numbering scheme itself. I can’t find any case law that upholds or rejects Copyright protection of a numbering scheme; anyone else?

As to the non-US nature, wikipedia cites the bill itself as saying that a database qualifies for protection if the producer is an individual living in the EU or a company registered in the EU or primarily doing business here. http://en.wikipedia.org/wiki/Database_Directive

The big question remains in my mind (as Ian Davis put it) – Can a thousand non-infringing releases be used as a whole?

Do you know of some precedent to establish that a “US database” can’t be protected by copyright in the UK if it is also published in the UK? I was not aware of that.

PAF is more than just a database, however, it’s an invented numbering scheme rather than factual data. That means it qualifies for Copyright protection I suspect (just as Dewey does) so would still be subject to Copyright in the US.

Though the jurisdictional differences would certainly make it harder to get it taken down.

It’s worth noting that the database right, i.e. the ability to copyright collections of facts, does not exist in the US. So a USA-based service that reconstituted the PAF database using the linked data techniques you describe ought to be immune from take-down, no matter how much of the database it accumulates. So ship us the data, we can let freedom ring!