Tracker · US science policy
LiveOpened 21 July 2026 · last updated 24 October 2026
The question this tracker follows
Will the US “Golden Age” science strategy translate into funded reality, and can metascience survive being adopted by one government?
The OSTP report reads as a metascience document wrapped in a national-security frame. In October it moved from paper to practice: agencies launched their first funding experiments, and the US took the agenda abroad in the Kyoto Vision. The money is still contested, and the deeper stake remains whether the reform agenda gets branded as partisan and discarded.
The coverage falls into clear camps: reform insiders giving a bounded endorsement; critics reading it as an alibi for cutting universities; delivery sceptics; the American right (endorsement and “not far enough”); and European voices on openness, academic freedom and plural perspectives. The revealing cut is how each treats the word “metascience”, and the finding that almost no one treats it as a distributed, system-wide capability.
The OSTP report Science: A New Golden Age (Kratsios, 21 July 2026) drew reactions from reform insiders, critics, funders, unions, European policy circles and the American right. Read across them, the responses disagree less about the report than about what “metascience” even means.
The 123-page report to the president, with the operative FY2028 R&D Priorities Memorandum (Kratsios + OMB's Vought) as its annex, directing agencies over $3bn to report in 90 days how they will reprioritise toward named security and technology fields. See also the release.
Broadly supportive, but with a boundary.
Endorses the mechanisms but draws the line: metascience innovations are “game-changing additions to America's scientific core — not a replacement for it,” and “ambitious new models depend on stable support for science overall, university & academic research, independent peer review, and recruiting top scientific talent from anywhere in the world.”
A “timely, constructive contribution… on the role of metascience”; the vision is serious, delivery is everything, and the OMB rule cuts against it.
Reform-community engagement with the report's metascience agenda.
The reform vocabulary is legitimating cuts and control.
Cole Donovan (Stand Up for Science): a “bad faith argument,” “does not seem like a document that is realistic”; the reallocation benefits industry and AI firms. See the org statement.
Neal Lane calls the report “disappointing”; the administration's own policies make it a missed opportunity. See also the “calcified system” explainer and the budget “slasher sequel.”
A split verdict across the research community.
Concede “our research enterprise faces serious challenges,” but the plan “robs universities of research money and funnels it to artificial intelligence and private companies”; “reforms must strengthen scientific independence — not weaken it.” Also: it “does nothing to make anyone healthier or safer.”
Reform as cover for defunding the university.
Good-faith fix or smokescreen; carries the worry the metascience movement is used to justify punishing universities, with Jean-Claude Burgelman (“bypass the power of universities”) and EU science counsellor Florent Bernard (“solely serves private interests”).
A third independent attestation (with Tech Policy Press and AAUP/AFT) of the university-to-AI reallocation. See also European researchers' alarm over cooperation.
Announced is not the same as enacted.
The research universities welcome the 8 October initiatives but note much of the money is state or private, and that funding cuts, conditions and slowed grants have cut PhD admissions while visa and H-1B policy deters foreign talent.
A bipartisan Senate stopgap moves to block the proposed political control over grants.
Republicans join the measure stripping political appointees' proposed veto over awards.
Of ~497,000 public comments on the Uniform Guidance rule, ~95% opposed, ~1% supported. Context: SciLine explainer.
“I approve of Science: A New Golden Age and endorse it.” Reads it through the reproducibility / “gold standard science” lens (the irreproducibility crisis, AI fraud-detection, unwinding the “grant-getting” culture), but wants “getting the science right” before treating government as “an efficient technology venture capitalist.”
Welcomes the diagnosis but argues the report will fail like Bush's 1945 report because it keeps the “scientific abundance gospel”; his fix is to “close the NSF and shutter the NIH's extramural programs.”
A real golden age needs open borders and proper funding for public health, social science and the humanities, alongside technology and engineering.
Accept the report's diagnosis that science has become strategic; reject its prescription. Europe should build the capacity to translate discovery without importing the politicisation.
The competition reframed as innovation-system vs innovation-system; what the US can and cannot borrow from China.
Places the Golden Age inside a wave of national strategies (US, Germany, Korea, Japan) launched in three months because “the old ways of delivering STI programming no longer works,” and reads metascience as a tool to “get the most bang for the buck.”
The agenda goes international and the first experiments start. The responses turn to co-optation, academic freedom and what the binding documents leave out.
James Wilsdon warns of a “convenient hook” and a “political land grab”; others argue researchers should report what they find regardless of who uses it.
The US and 16 countries endorse a one-page statement valuing “a deliberate portfolio of funding mechanisms” and “dedicated metascience capabilities in our governments”. No Nordic country signs. Germany’s research ministry stresses it is non-binding and that research freedom is non-negotiable.
Germany’s DFG and Rectors’ Conference fear “new institutional models” could serve an agenda that calls universities “legacy institutions”.
Names the first experiments: golden tickets and distributed review at NSF’s new Office of Metascience, double-blind review at NASA, golden tickets at DOE and a $174m NIH reproducibility initiative (fact sheet). NSF and UKRI plan an international network of metascience units.
No money or projects are named yet, and early evidence on golden tickets is mixed.
The OSTP–OMB budget memo matters more than the report because OMB controls the money. It carries long grants, prizes, new institutions and metascience offices into action plans, and never mentions replication or reproducibility.
The Kyoto Vision is innocuous, the report behind it nostalgic and dismissive of universities. He calls the German Rectors’ reply tin-eared and points to Sarah de Rijcke’s six principles of academic freedom and to plural perspectives.
Analysis
The same word is doing very different work across the responses. Seven readings:
The finding: almost no one treats metascience as a distributed systemic capability: the science system's capacity to keep reforming how it funds, organises and evaluates itself. That definition is the unoccupied position, and it reconciles the rest: mechanisms (A) and units (B) are instruments of the capability, co-optation (C) is what concentrating it does, the label fight (D) and the rigour-vs-design split (G) are symptoms of no shared definition, and the macro blindness (E) is why the debate talks past itself. Marblestone reaches the same conclusion without the vocabulary: additions, not a replacement. The October responses sharpen it: Asterisk shows the binding memo keeps the mechanisms and drops reproducibility, and Curry asks for plural perspectives. A capability spread across the whole system would hold both.
David Hilmer Rex
The US now runs real funding experiments, and they deserve to be watched closely. Experiments are useful. The test is whether they sit inside work on the system as a whole, with secure basic funding, research careers, reliable findings and academic freedom, or stand in for it while budgets shrink.
Metascience is only “co-opted” if the field lets one office define it. You can capture an office. A capability spread across every university and funder is much harder to capture.