A new graph deserves a clear commons: why OpenRS defaults generated BIBFRAME to CC0

Libraries are again investing significant energy in resource sharing, shared catalogues and metadata record supply. Much of that energy comes from a genuine and justified discontent with what has happened to library data.

Libraries believed they were contributing records to a commons: a shared body of description made more useful by collective effort. Somewhere along the way, the norms and expectations around that contribution fell out of step with the pace of technology. Data supplied for cooperation could become enclosed inside a shared service. The community had contributed the raw material, but its practical ability to retrieve, reuse or build upon the resulting aggregation was uncertain, constrained or commercially mediated.

The current desire to address that history is welcome. Several new resource-sharing and shared-index initiatives draw power from the community’s dissatisfaction with enclosure. But changing the name above the door is not the same as changing the structure.

“Trust us” is not an architecture

The model still too often looks like this: libraries contribute data to a central aggregation and trust its operator to use that data as intended.

Sometimes the reassurance is simply: “Trust us; we are not like those other organisations.” Sometimes it is: “Trust us; this service is open source.” But open-source software can still operate a walled garden for library data. The licence on the application does not automatically determine the licence on the records held by it, nor the practical freedoms available to the libraries that supplied them.

That observation does not undermine the legitimacy of shared services. Aggregations require real investment, stewardship, infrastructure and sustainable operating models. Commercial services can create enormous value. The issue is narrower: every time a library contributes data to an aggregation, it makes an act of trust. It trusts that today’s policy will survive tomorrow’s leadership, ownership, funding and commercial pressures. It trusts that data contributed for one purpose will not acquire unintended restrictions or consequences later.

Trust matters, but it is not a substitute for structure. If our new systems reproduce the same producer–aggregator relationship as the old systems, with only a new promise from a new operator, what have we actually changed?

BIBFRAME gives us a new starting point

BIBFRAME 2.0 gives libraries more than a different record syntax. Its Work, Instance and Item model lets us create a new linked-data representation: a bounded intellectual resource assembled from descriptive facts inferred and mapped from source records.

That distinction creates an opportunity. The generated BIBFRAME graph is not the book, journal or digital object it describes. It is not a claim over the source record from which facts were derived. It is a new representation created by a defined mapping process. We can state clearly what may be done with that representation.

Without a corresponding decision about data rights, however, BIBFRAME alone changes the shape of the container rather than the social and economic relationship around its contents. We risk recreating the old problem in JSON-LD: a technically modern graph whose terms of reuse remain implicit, provider-specific or dependent on goodwill.

That is why ORS-Appliance now defaults newly generated BIBFRAME descriptions to CC0 1.0.

Put the statement in the record

The important detail is not only that CC0 is the default. The assertion is carried inside the BIBFRAME graph as administrative metadata:

{
  "@id": "urn:openrs:resource-description:{id}#metadata",
  "@type": "bf:AdminMetadata",
  "dcterms:license": {
    "@id": "https://creativecommons.org/publicdomain/zero/1.0/"
  },
  "dcterms:rights": "The descriptive metadata in this resource description is made available under CC0 1.0. This statement does not assert rights over the resource described."
}

The statement travels with the graph. A consumer does not need to remember which website once described a provider’s policy or infer a licence from the software used to produce the record. The machine-readable URI and the human-readable scope are present at the point of use.

This does not prevent somebody charging for a valuable service built around CC0 data. Nor does it make every enrichment added later automatically CC0. What it does is preserve a reusable commons at the level of the generated description. A service may build a wall around its interface or its own additional value, but it cannot turn the permissions already granted for that CC0 description into an exclusive right. Another service, library or community project remains free to retain and build on the same graph.

A default, not a hidden assumption

CC0 is on by default because defaults express intent. If open metadata is merely possible but every tenant must discover and enable it, the practical norm will remain ambiguity. OpenRS instead starts from a clear expectation that newly generated descriptive metadata should be available to the commons.

This is still a consequential choice. Creative Commons describes CC0 as a public-domain dedication that waives copyright and related rights to the extent allowed by law, with a public-licence fallback. An organisation should select it only when it has authority to make that dedication.

The KI Console activation screen therefore does three things explicitly:

• defaults Generated metadata licence to CC0 1.0;
• offers Unspecified as a deliberate alternative;
• warns that the activating organisation must have authority to dedicate its generated descriptive metadata to the public domain.

The choice becomes a tenant Application Setting, so an ORS-Appliance administrator can change it later. Changes affect records created or canonicalised after that point; there is no retrospective backfill. If incoming BIBFRAME already contains a licence or rights statement, ORS preserves that source assertion rather than overwriting or mixing it with generated CC0 terms.

The implementation is deliberately centralised in the canonical BIBFRAME path. MARC ingestion, direct record creation and other bibliographic sources therefore reach the same policy boundary. UNSPECIFIED means exactly that: ORS adds neither a licence URI nor a rights statement. It does not claim that the metadata is closed; it declines to manufacture an assertion the tenant has chosen not to make.

Open source is necessary, but data freedom must be explicit

Open-source infrastructure is important. It makes systems inspectable, adaptable and less dependent on one supplier. But code openness and data openness solve different problems. We should resist using the first as a proxy for the second.

The next generation of library infrastructure should not depend solely on finding more trustworthy operators for structurally unchanged aggregations. It should reduce the amount of trust the structure requires. Portable identifiers, open protocols, exportable graphs and explicit machine-readable rights all help move authority back towards the contributing community.

This CC0 default is a small feature in software terms. In policy terms it is a statement about what BIBFRAME can be for: not simply a new way to feed the next shared index, but an opportunity to create a descriptive commons that remains available beyond any one service.

If we talk about BIBFRAME without acting on the licensing of the graphs we create, we risk rebuilding the problems of the past with better technology. OpenRS is choosing a different default—and putting that choice in the record where everyone can see it.