Every brief on this site carries a review badge, and this page is the contract behind it. The badge is derived at build time from facts stored on the brief's own record — its review tier, its clock, and any real review that has happened — and is never typed by hand. A badge that could be edited independently of the facts would be worth nothing.
The four states
Not independently reviewed. The honest default, and what almost every brief says today. Each brief was researched, written, and checked against its sources by its author in one pass — a real process, and not peer review. The badge says so rather than letting a reader assume otherwise.
Reviewed by <name>, <date>. A person read the brief against its sources after publication and stands behind it. This state appears only when a review has actually occurred; the Institute does not invent reviewers, dates, or approval.
Review due since <month year>. The brief's review interval has elapsed. This state is shown, not hidden: an overdue review is a fact about the Institute's capacity, and a badge that concealed it would make every other badge on the site worth less. Expect to see it — a corpus this size on a single-principal institute will run clocks out, and the policy was adopted knowing that.
Under revision — <reason>. An evidence trigger fired and the brief is being reworked. The reason is always stated, because "under revision" without one reads as evasion.
The tiers
One interval for every subject would be as uninformative as none: a brief on fault-tolerant quantum computing ages in months, a brief on the Manhattan Project does not. Each brief is assigned one of three tiers, stored on its record:
| Tier | Review interval | Currently assigned |
|---|---|---|
| Fast | 6 months | 61 briefs |
| Standard | 18 months | 250 briefs |
| Slow | 4 years | 74 briefs |
Adopted 2026-09-14. Every figure in this table is computed at build time from the brief records and the taxonomy's one declaration of the intervals; none of it is typed here.
Fast is for subjects whose evidence base turns over within a year — active clinical trials, benchmark-driven fields, live standards processes, programmes reporting on a launch or funding cadence. A fast-tier brief must carry a named owner: a six-month clock nobody owns is a clock nobody reads. Standard is the honest default — established science with an active but not volatile frontier. Slow is for historical and methodological subjects, and for theory whose evidence moves by reinterpretation or decadal experiment rather than by result: the whole of Categories X and XI, and the named megastructure and exotic-propulsion concepts.
The five triggers
A trigger overrides the clock: the brief goes into review immediately, whatever its tier, when any of these occurs. A cited source is retracted — this corpus has already published a correction of exactly that kind. A cited link dies or moves and the verification ledger records it. A decisive experiment the brief itself named reports a result — section 6 of every brief exists to make this trigger fire. A flag would change — the strongest trigger, because a claim moving between speculative and frontier is this site's whole subject matter. A reader challenges a claim and the challenge is not obviously wrong.
The clocks
Clocks run from this policy's adoption or from a brief's last review, whichever is later; briefs published after adoption run from their publication. No historical per-brief dates were invented to make the clocks look older — a fabricated date under an honest badge would be the exact failure this system exists to prevent.
What this policy is not
It is not peer review, and it does not claim to be. It is a published commitment about freshness, with the overdue state as the enforcement: when the Institute falls behind, the site says so on the page where it matters, in the brief's own header. Corrections and flag challenges are welcome on every brief and feed the trigger above.