Few pieces of research paperwork are as universally disliked, and as consequential, as the current and pending support disclosure. It is the document on which a US federally funded investigator declares all the support — financial, in-kind, and resource — that they currently hold or have proposed elsewhere. Since the implementation of NSPM-33, getting it wrong has moved from an administrative nuisance to a research-security and integrity matter with real teeth. And yet, for most researchers, the disclosure is still assembled by hand each time it is required, from memory and scattered records. This is precisely the kind of problem that machine-readable, identifier-anchored data is meant to solve, and it is why funders are pushing hard in that direction. The relevant vocabulary lives in the funding and finance domain.
What the disclosure actually is
Current and pending support (CPS) — the term used by the NSF and, in harmonised federal guidance, across agencies — is the investigator’s declaration of all ongoing and proposed support for their research effort. The NIH equivalent is called Other Support (OS), and while the two have converged considerably, they are not identical in scope or format. Both ask, in essence: what else are you funded to do, by whom, for how much, and for what fraction of your time? The companion document is the biosketch, the funder-prescribed biographical sketch that accompanies a proposal.
The purpose is twofold. The traditional purpose is fiscal and scientific overlap: a funder wants to know that it is not paying twice for the same effort, and that the investigator is not committing more than 100% of their time across all their awards. The newer purpose, sharpened by research-security policy, is transparency about foreign and undisclosed support: an investigator must declare in-kind contributions and appointments that previous generations of the form let slip through. Under NSPM-33, a foreign component or an undisclosed in-kind contribution from a foreign source is exactly what the disclosure exists to surface.
Why the manual version fails
The problem with a hand-assembled disclosure is not that researchers are careless. It is that the information is genuinely hard to compile accurately from memory: the precise start and end dates of every award, the person-months committed to each, the value of in-kind support, the exact legal name of every sponsoring organisation. An honest investigator can make an inadvertent error on any of these, and under the current regime an inadvertent error can be treated as a disclosure failure. The form asks for facts the investigator’s institution already holds in its grants system, but which the investigator must nevertheless re-enter by hand.
This is the signature of a metadata problem. The same facts — award, sponsor, dates, effort, value — exist in the CRIS, in the grants office, and in the funder’s own records, but there is no structured, authoritative representation that the disclosure can simply draw from. So it is reconstructed, manually, every time.
What machine-readable disclosure looks like
The direction of travel is toward disclosures that are generated from authoritative structured records rather than typed afresh. The clearest existing example is SciENcv, the federal system that produces NSF and NIH biosketches and current-and-pending / other-support documents from a researcher’s stored profile, with the ability to pull data from an ORCID record. The principle is exactly right: the researcher maintains the underlying facts once, in a structured profile, and the funder-specific document is rendered from it on demand.
For that to work well, the underlying facts need persistent identifiers. An award identified by a Crossref grant ID is unambiguous in a way that a typed-in award number is not. A sponsoring organisation identified by a ROR ID resolves the endless variation in how institution names are written. The investigator identified by an ORCID iD ties the disclosure to a profile that other systems can recognise. The disclosure becomes a structured assertion over identified entities, which a funder system can validate — checking, for instance, that declared effort across all awards does not exceed full-time — rather than a free-text document that a human reviewer must read and cross-check.
Why funders want this, specifically
Funders are not pushing machine-readable disclosure to make researchers’ lives easier, though it would. They want it because it makes the disclosure checkable. A structured CPS can be validated automatically against the funder’s own award records and against the investigator’s other declarations. Overcommitment of effort can be detected. Inconsistencies between a biosketch, a current-and-pending statement, and a project’s other-support document can be flagged before they become an integrity case. In a research-security environment where the cost of an undetected non-disclosure is high, the ability to validate disclosures programmatically is worth a great deal.
There is also a fairness argument. A regime that punishes inadvertent disclosure errors but forces researchers to assemble disclosures by hand from imperfect records is structurally unfair: it manufactures the very errors it then penalises. Generating disclosures from authoritative, identifier-anchored data reduces the rate of honest mistakes, which is in everyone’s interest.
The pieces that need to connect
None of this requires new identifier infrastructure — it requires connecting the infrastructure that already exists. ORCID for the person, ROR for the organisations, Crossref grant IDs for the awards, and a structured profile (SciENcv-style) that holds the effort, dates, and values. What is still missing is a shared vocabulary for the disclosure’s own concepts — what counts as in-kind, how effort is expressed, what distinguishes current from pending — so that disclosures generated by different systems mean the same thing. That definitional layer is the contribution a convening body like CASRAI can make, federating to the authoritative funder and research-security guidance rather than inventing normative content.
What to do now
For researchers: maintain your underlying facts once, in ORCID and a structured profile, and generate funder documents from them rather than re-typing each time. For institutions: ensure your grants system records awards with their Crossref grant IDs and sponsors with ROR IDs, so that disclosures can be assembled from authoritative data. For funders and standards bodies: specify disclosures as structured, validatable assertions over identified entities, and reduce the manual burden that produces inadvertent errors.