Why is it impossible? If they're paying for a copyright extension, shouldn't it be clear which works they are paying for? Anything not in the registry would fall under the shorter term.
Consider a short story published in a literary journal last week. Allow a copyright system to exist that covers each eligible work for 30 years by default, and allows multiple extensions of 5 years at a cost of $1000 per extension per work, up to a total copyright term of 100 years.
Someone in 2047 discovers a copy of this particular journal issue, which has now become obscure. They identify that this short story, though never being given acclaim, has had a subtle yet profound cultural influence. They decide to republish the story in their own commercial magazine, but want to check if the copyright has been extended. If it has not, the story is in the public domain. If is has been extended, the story is still covered by copyright.
They can search the central registry for the title of the short story and the name of the author. But they also have to search for the names of the editors of the journal, because it isn't clear whether or not the copyright stayed with the author or was transferred to the journal as part of the submission process. In fact, the copyright might have been granted to the owner of the journal, which may be an individual or may be an institution. The entry for the copyright might be under the title of the short story, or under the title for the journal, or under the title of that particular issue of the journal.
Furthermore, the author of the story, if they retained the copyright, may have submitted the story under a pen name, and/or may have registered the copyright of the short story under a different title than the title they used to submit to the journal, or the title might have been changed by the editors.
Or, the author, after having the short story accepted by the journal, may have sold it to a publisher who published it in a short fiction anthology, the copyright for which is registered completely separately and is essentially invisible to our hypothetical future researcher.
Any of these agents might have extended the copyright. It would be extremely difficult to prove to a high degree of certainty that the story was or was not covered by copyright.
It would also be basically impossible for anything more intensive than text.
Violation of a registered copyright only carries punishment if "the" UUID of the work is registered and is contained in the work (byline, front page, whatever as appropriate). Doesn't actually matter if the UUID has any relationship to the work, although it would be nice if it did.
Example, This HN post from 2014 is registered copyright UUID 12345-23-242342 by VLM.
In 2097 simply check https://copyright.example/12345-23-242342 for current status. Instead of copyright.example, I hope they can think of a more intelligent international (interplanetary?) domain name, but the general idea holds.
Don't want copyright enforced on a work created in 2015, simply neglect to list the UUID you got at registration. I see no reason a registration should be expensive other than the usual crooks trying to eliminate competition. No technological reason.
False claim (intentional random numbers instead of valid registration) equals instant public domain.
Essentially this large integer is the title - registration ID - type of concept for this copyrighted media.
Personally I think it would be awesome if the title generator also stored licensing info. My panoramic landscape photo is not only registered to me, but is CC:NC. If you messed with the system you could force all UUID beginning with 0 to be all rights reserved, all UUID beginning with 9 to be CC:SA. Just keep on generating UUIDs until you get ASCII codes for C, C, S, and A as the first 4 bytes for a CC:SA UUID. Or the first 3 bytes being ASCII G, P, and L. Better be a long/wide UUID system.
And it's worth considering that every tweet, every photo on Flickr, every video on YouTube, every status update on FB, and so on, constitutes a new copyrightable work. Not everyone will care about the copyrights of their tweets, but some will. How do you copyright something for which you have no control of the identifiers? I assume you don't intend for people to hit up the service for a new UUID before tweeting and embed the token in their tweet?
(Obviously this problem would go away if e.g. Twitter itself signed onto this scheme, automatically registering and UUIDing your tweets on your behalf.)
Is taken care of by
"Violation of a registered copyright only carries punishment"
Or if thats a non-starter just reduce all the penalties by 1e6 or perhaps 1e3 for violation of an untagged work.
Retweeting being a copyright violation, but if a REGEX doesn't find a UUID in there, the legal cost would be a micropayment transaction, so probably ignored.
I would have to re-read Berne to determine if this sneaks by on a technicality, or if Berne is simply culturally and technologically obsolete. Its a copyright treaty, not an economic suicide pact... can be worked around.
> you might have a 35-year-old work and have no idea if its copyright has been extended because it was not originally printed with a number
No... if protection is contingent on publication of the number along with the work, then a work printed without a number is unprotected. There is no concept of "extending copyright" on a work that isn't even copyrighted.