For researchers

Identifiers and citation

Why every record gets a permanent URI but not its own DOI.

Identity and citation get conflated constantly, and they are two different jobs with two different costs.

Identity: every record, from day one

Every record has a permanent, resolvable URI:

https://mofplus.org/net/pcu
https://mofplus.org/bb/benzene
https://mofplus.org/framework/…

Records are keyed by unique human-readable names, so the identifier is derivable rather than looked up. The same address serves a human page and a machine-readable representation — send Accept: application/json, or append .json.

Old addresses keep resolving. /nets/net/pcu redirects permanently to /net/pcu; the same holds for building blocks and frameworks. A URI that has been cited must not stop working, and a permanent redirect is what tells a crawler or a reference manager that the target has moved for good.

Identity costs nothing per record beyond the discipline of not changing names.

Citation: per dataset release, not per record

DOIs here are minted for versioned dataset releases and curated collections, not for individual records.

This is a maintenance decision, taken deliberately. A DOI is a standing obligation: it must resolve forever, its metadata must be kept accurate, and somebody must answer for it. Minting one per record means tens of thousands of standing obligations — a workload that a two-person part-time team cannot honour, and an unhonoured DOI is worse than no DOI, because it is a promise that turned out to be false.

A release DOI is a small number of obligations that can actually be kept. It also matches what a reader needs: the state of the data at the time you used it, which is what makes your work reproducible. A per-record DOI without a version tells them nothing about what the record said when you read it.

Registration goes through the university library's DataCite membership rather than a commercial account — cheaper, and institutionally durable in a way a group subscription is not.

What to put in a paper

Both. The citation block on every record page gives you the sentence:

MOF++ contributors (2026). pcu [net]. MOF++ dataset release v. DOI: … Record: https://mofplus.org/net/pcu

The release DOI says which data. The record URI says which record. Together they are reproducible; either alone is not.

Until the first dataset release is registered, the block names the release version without a DOI. That is honest — a DOI that has not been registered is not a DOI, and inventing one would be worse than saying so.

Licences

The licence is stored per record and is never empty. The default for new deposits is CC BY 4.0.

Per-record rather than site-wide, because external contributors may need other terms — an institution or funder can require a particular licence, and a single site-wide statement would either exclude those deposits or misrepresent them. The licence you see on a record is the licence that record was deposited under.

CC0 is arguably the better fit for purely factual metadata, and the argument is real. CC BY across the board was chosen because one licence is simpler to administer than two, and at this team size administrative simplicity is worth more than the theoretical gain.