• The Problems Are Baked Into the Incentives
• We Have the Parts to Rebuild This
• A New Layer: Signed, Portable Review
• Forbidden Questions #1 (Draft): How Publishing Broke, Who Profits, and How to Patch It
• Call to Action

Publishing underlies everything above it in the stack — policy, medicine, engineering, public trust. When that control plane is brittle, captured, or running decade-old firmware, the whole system fails. We’re fighting an epistemological information war with logistics built for 1950 and metrics tuned for vibes.

§The Problems Are Baked Into the Incentives

The distortions aren’t mysterious — they’re structural:

Oligopoly pricing and APC tolls turn access into a luxury good Goodharted metrics (Impact Factor, h-index) reward splash over substance Novelty bias plus publish-or-perish creates an exploit kit: HARKing, p-hacking, forking paths, and file-drawer deserts where negative results go to die The press-office-to-headline pipeline amplifies “maybe” into “OMG” on Monday; the retraction on Friday never ships the patch notes Peer review is a heroic volunteer fire brigade asked to inspect skyscrapers with a garden hose, no time, and occasional conflicts — then blamed when the wiring they never saw catches fire Reproducibility fails because the repo is gone, the data is under NDA, and the environment requires Python 2, Java 6, and a blood moon

§We Have the Parts to Rebuild This

The supply chain for truth can be fixed:

Preprints decouple time-to-knowledge from venue latency Registered Reports flip the incentive: publish the question and method before you know the answer, then report what you found — boring or not Open peer review and overlay journals separate evaluation from distribution Funder reform means buying what funders actually want: reusable datasets, robust software, and independent replications — with credit counted in hiring and tenure Data/code availability, containers, and CI turn “works on my machine” into “passes in prod” Post-publication review (PubPeer-style), red-team grants, and adversarial collaborations can pressure-test claims before they ossify into dogma Corrections and retractions must propagate with the same velocity and UX as the original claim — no more silent edits in the footnotes of history

§A New Layer: Signed, Portable Review

Nostr is a useful place to prototype this. The concept: signed, timestamped attestations linking claims to artifacts.

Think Proof-of-Review — a short, legible review posted as a Nostr event that includes:

Content hashes of the paper, dataset, and code Conflict disclosures A verdict bounded to specific claims, not general vibes

A replication note becomes a reusable, portable, rankable, subscribable signal. Follow lists become trust graphs. Overlay journals emerge as curated feeds with transparent criteria. Embargo theater dies because the network routes around it.

We’re basically running npm install science — this time with a lockfile.

§Forbidden Questions #1 (Draft): How Publishing Broke, Who Profits, and How to Patch It

Three sections:

Metrics theater (IF, h-index) as governance failure Replication deserts and the economics of invisible nulls Embargo games and why corrections don’t travel

Three concrete proposals:

Make Registered Reports + code/data/CI the default for publicly funded research Fund and formally credit replications the same way we credit original work Deploy signed post-publication review and replication attestations with content-addressed artifacts

§Call to Action

Send three case studies — one metrics fiasco, one replication that flipped a field, one embargo/press-release whiplash that misled the public. Links over takes.

Join the 30-day Nostr Review Club — each week, pick a high-signal preprint, run a light red-team, and post a signed review plus reproducibility notes. A minimal template and a “corrections bot” (threads updates to prior viral posts) will be provided.

Developers — help sketch a simple NIP for review/replication attestations: content hashes, conflict disclosures, claim-bounded verdicts.

Curators — propose overlay feeds.

Funders — pledge microgrants for replications and negative results.

If we can ship CI for facts, we can ship trust without demanding faith. Hold my DOI — let’s push to main, but only after the tests pass.