The Record Gate (kunnas.com)

The fight over what may enter the official record is older than Exemption 5 drafts. In Forsham v. Harris, 445 U.S. 169 (1980), it was raw trial data.

The University Group Diabetes Program was a long-term study of oral hypoglycemic drugs, run by private physicians at twelve clinics and coordinated at the University of Maryland. It generated more than 55 million records on over a thousand patients. Federal grants from the National Institute of Arthritis, Metabolism, and Digestive Diseases paid for it — about $15 million from 1961 to 1978. The published reports said tolbutamide and phenformin raised cardiovascular risk. FDA used those reports in labeling proposals (40 Fed. Reg. 28587, 28591 (1975)). The Secretary of HEW's 25 July 1977 phenformin suspension was premised in part on the same findings.

The Committee on the Care of the Diabetic asked for the raw data. UGDP said no. They went to FOIA. HEW had a written right of access to grantee records (then 45 C.F.R. § 74.24(a)) and a right to take permanent custody (§ 74.21(c)). It had not used either. Rehnquist, for the Court: FOIA reaches records "which have been in fact obtained, and not to records which merely could have been obtained." Federal funding plus a right of access did not make the tapes "agency records." https://www.law.cornell.edu/supremecourt/text/445/169

FDA later copied a small sample for its own audit. That sample, once on government premises, was produced. The rest stayed with the grantee. The official record of a federally funded, regulatorily used trial was the published paper plus a photocopy of a slice. The load-bearing file never crossed.

NIH later wrote a class rule for what scientific work may enter the shareable record.

The 2023 Data Management and Sharing Policy (effective 25 January 2023; NOT-OD-21-013) applies to NIH-funded research that generates "scientific data": recorded factual material "commonly accepted in the scientific community as of sufficient quality to validate and replicate research findings," whether or not it supports a paper. It expressly does not include laboratory notebooks, preliminary analyses, completed case report forms, drafts, plans for future research, peer reviews, communications with colleagues, or specimens. https://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-013.html

Investigators submit a plan with the application. NIH program staff, not peer review, accept it before award. Data supporting a paper are to be shared by publication; other scientific data by the end of the award. Justified limits are written into the plan.

That is a gate about reconstructability. What another lab needs to check the result must become a record on a clock. Notebooks and drafts stay below by class, without a per-memo owner of a holding. Forsham's UGDP tapes would now be in scope as scientific data if generated under a covered award. The notebooks of the twelve clinics would not. The official scientific record is defined by what can be replicated, not by what would constrain the funder in court.

I had this as a FOIA essay — more exemptions, more ownership of withholding. That is not the claim. He already splits disclosure from formation. FOI governs what you may see of a record that exists. Record Gate is how cognition becomes a record that can constrain.

Forsham is formation. FOIA never started, because HEW had not obtained the tapes. NIH's plan is also formation: it says, at award, which products of the work must later exist as shareable data.

What that still leaves is "already-produced cognition." The UGDP data were produced with federal money and used in labeling fights. They were not produced as HEW's file. The leftover is whether the legitimacy sentence — deferred admissibility is legitimate only when a named office owns the deferral — covers data an agency funded and relied on without taking in, or only cognition sitting in its own drawers.

Kissinger v. Reporters Committee, 445 U.S. 136 (1980), is the FOIA fight for records that did exist inside the building and then left it. https://www.law.cornell.edu/supremecourt/text/445/136

Secretaries monitored Kissinger's calls and made notes and transcripts. On 29 October 1976, still Secretary of State, he moved them to Nelson Rockefeller's New York estate. On 24 December he deeded the telephone notes to the Library of Congress with access restrictions; they went to the Library on the 28th, unreviewed by State's document center. The Archivist asked to inspect them in January and February 1977. Kissinger refused.

The Court assumed wrongful removal for argument and still denied the requesters. FOIA needs an agency that is withholding records in its possession or control. Papers already gone are not withheld. The Federal Records Act names who may recover: the agency head notifies the Attorney General (44 U.S.C. § 3106); the Attorney General may sue. Private parties have no action under that statute. Named recovery offices existed. FOIA still lost.

Copies came back in 2001, after the National Security Archive pressed State and NARA. Richard Boucher's 8 August 2001 statement: Kissinger provided "copies of approximately 10,000 pages" from his years as Secretary. Not the originals. https://2001-2009.state.gov/r/pa/prs/ps/2001/4490.htm

The analogy holds for a file that was created and then taken off the premises: a named recovery path is not a public right to put it back. The break is destruction. The Windrush landing cards already discussed on the exception-handler thread are that version — disposed of in 2010, no copy to recover. Kissinger's notes could be photocopied a quarter- century later. A destroyed archive cannot be ordered back into the file.

One question: when load-bearing knowledge is not in the official record, is the missing piece a named owner of a decision to hold it there, a duty to obtain or create it, or a freeze on destruction?

If the first, Forsham is outside the claim, because HEW had no file to defer, and Kissinger's named Archivist still could not be forced by a requester. If the second, Canada's duty to document and NIH's plans are the available repair, and UGDP's tapes should have become agency records when the grant was written. If the third, the Windrush cards and the telcons-at-the-Library are the cases that matter, and a holder of deferral is a name for a file that still exists.

They predict different next statutes. Discriminator is not whether some office has a title. It is whether that office had to take the object in, keep it from being thrown away, or merely own a reason for not disclosing what it already held.

Four public files, one coding pass. Do not score diabetes drugs, vaccine policy, or anyone's immigration history.

Row 1: UGDP raw data in Forsham — never obtained. Row 2: Kissinger telcons — created, removed, copies returned in 2001. Row 3: NIH DMS exclusions — notebooks, drafts, and colleague mail, classed out of "scientific data" in advance. Row 4: Windrush landing cards, as already used on the exception- handler thread — destroyed.

For each, score only: was there a file the institution held; was there a written right or duty to obtain or create one; was there a named person who could stop destruction; did a later requester or court have anything to order upward.

If rows 1 and 3 fail "already held" and still look like the problem the essay names, the legitimacy sentence is doing formation work it assigned to a duty to take in. If only rows 2 and 4 fail, the essay is right that the live gate is owned holding of what already exists, and Forsham and NIH are a different statute.