Preservation

Open Licensing for Archives: Making a Collection Survivable

Licensing is usually filed under legal housekeeping. For a small archive it is closer to a survival strategy. A collection that anyone may lawfully copy can be mirrored, aggregated, harvested and — when its maintainer eventually stops — rescued. A collection with unclear rights cannot be any of those things, and when its server is switched off it simply ends.

Undocumented is worse than restricted

Aggregators, libraries and archiving initiatives all make the same decision when they encounter material with no rights statement: they leave it alone. The risk of using something whose status is unknown outweighs the benefit, so unclear material is passed over exactly as if it had been forbidden — while getting none of the protection an explicit restriction would give.

A statement of "rights status unknown, we welcome information" is therefore vastly better than silence. It is honest, it invites correction, and it signals that somebody has considered the question.

Two layers of rights, routinely confused

Digitised material has rights questions at two levels and they behave differently.

• The underlying work. The text, map or image as originally created. Its status depends on when it was made, when its creator died, and which country's law applies. Terms differ by jurisdiction, which is why blanket claims about what is public domain are unreliable.
• The digital reproduction. Whether photographing or scanning a public-domain original creates any new right of its own. In many jurisdictions a faithful reproduction of a flat work attracts no new copyright, because it lacks original authorship — but practice varies and some institutions assert rights regardless.

The position widely adopted by the cultural heritage sector is that faithful reproductions of public-domain works should be released as public domain. Several major institutions have adopted it explicitly, and it is the position that makes collections most useful. Whatever you decide, state which layer your statement applies to, because a licence on the scan says nothing about the text.

Choosing a licence

For material you hold rights in, use an established public licence rather than bespoke wording. Custom terms are unenforceable in practice, incomprehensible to tools, and disqualify the material from aggregation.

The Creative Commons suite is the standard vocabulary. Two practical notes. The non-commercial restriction is far more limiting than most people expect — its boundary is genuinely unclear, and the uncertainty makes cautious reusers, including some archives, decline entirely. And share-alike can prevent your material being combined with material under other open licences, which for a collection that ought to be aggregated is a real cost.

For works genuinely in the public domain, mark them as such rather than licensing them. Applying a licence to a public-domain work misrepresents its status and effectively claims a right you do not have. RightsStatements.org exists precisely to provide standardised statements for the awkward middle cases, including material whose status is genuinely undetermined.

Where the statement has to live

A rights page is not sufficient. The statement must travel with the material, because material gets separated from its website constantly — harvested, downloaded, mirrored, embedded.

Put the rights statement in the item metadata, so it appears in every export and every harvested record. Put it in the machine-readable markup on the page. And keep it identifiable in the file itself where the format allows. The test is simple: if somebody downloads one item and nothing else, can they tell what they may do with it?

Plan for your own absence

This is the argument that matters most and is made least. Small collections are almost always maintained by one or two people, and maintenance eventually stops — through loss of interest, funds, health or life. The question is what happens next.

Three things materially improve the odds, and all three are cheap while you are still here.

• An open licence, so a rescue is lawful without anyone needing to find and negotiate with you.
• Bulk export, so the collection can be taken in one piece rather than scraped page by page.
• Documentation of the data model, so what is taken remains intelligible.

A collection with those three properties can be picked up by anyone. Without them, even a willing rescuer faces a legal question they cannot answer and a data structure they cannot read — and in practice the collection is lost while still technically existing on a disk somewhere. Making the material legible to a stranger is the whole of the discipline, and it is the same argument as keeping addresses alive, applied to the collection as a whole.