Radar Perene / Archive / science

From self-assessment to open data: closing the loop between what the house audits and what it publishes

◦ Index methodology v2.2 (working papers with DOI). See the methodology.

Science

Seen up close, the pieces of this trail look like separate subjects: a regulatory vacuum here, a stopping rule there, a published hit rate, a citation check, a data package with hashes. Seen from afar, they draw a single figure — and this closing piece exists so the reader sees it whole before moving on. The figure is a loop, and it always turns in the same direction: from inside out.

The pattern running through the house's entire archive is a four-stroke cycle: audit internally, under criteria defined in advance; decide what opens, under a declared cost; publish what crossed, with DOIs and versions no one can rewrite; and return verifiable data, so the reader checks instead of trusting. Each stroke feeds the next, and the last feeds the first — because everything that became public becomes material for the next audit.

The cycle, told as one study's journey

Follow a hypothetical house study through the circuit, and the trail's pieces fit together on their own. It is born at the bench and first faces the internal sieve — the retest under variants, the rule that closes it if the central statistic does not come back, the periodic assessment that re-examines it even after approval. If it survives, it reaches the second station: the opening decision, which weighs whether the material can bear being eternal and to what degree — concept always; formula, only when openness strengthens; data, when the documentation supports the deposit.

What crosses is published in the repository with a DOI, preserved versions, revisions possible only in the open. And what was published returns verifiable data — spreadsheets with integrity verification, confrontable findings, a hit rate that enters the next account. The loop closes where it began: the public material is audited again, and whatever ages badly will be revised in public, as the house's series has already done more than once. There is no final station; there are turns.

Why the order of the strokes matters

The cycle works because each stroke only receives what the previous one approved. A house that published first and audited later would be asking readers to finance its drafts with their own trust. A house that audited forever and never opened would be asking for trust without offering a verification surface. And a house that opened without returning the data would be offering conclusions auditable in appearance only. The four strokes, in the right order, are what allow the sentence that sums up this whole trail: nothing the house claims in public demands trust — it demands, at most, the work of checking.

This design was not born of virtue; it was born of the vacuum described at the trail's opening. Where there is no supervisor, either the operation builds its own accountability circuit, or it operates with none. The cycle is the house's answer — and, as the previous piece recorded, it is an answer other providers can adopt.

Back to the beginning — of both tracks

This article closes the second great trail of the science archive: the one that told the house's studies — the published indices, the findings that survived and the ones that fell, the governance around them. It was written on the foundations of the first trail, which explained the concepts everything here was built with — from how a research project is born to the reading rulers. Whoever arrived at this point through the second trail has the natural way back: starting the first from its beginning, with what a working paper is — the format that carries all of the house's public output — or revisiting how a Brazilian open dataset becomes research. And whoever wants to retrace this trail from its first question finds the starting station at beating the trivial benchmark, where the house's ruler first gets applied to its own indices.

The complete archive of studies, with DOIs and versions, remains gathered on the research page — which is, in the end, the whole cycle seen at once: what the house audited, decided to open, published and returned as data. The rest is the next turn.

Frequently asked questions

Is the cycle a formal methodology or an after-the-fact description?

The rules composing it — retest, stopping rule, self-assessment cadence, opening criterion — exist in writing and prior to the episodes they apply to; this trail presented them one by one. What this closing piece adds is only the overview: the rules were always strokes of the same circuit.

What can the reader actually verify of the cycle?

The outputs of each turn: the working papers with DOIs and preserved versions, the revisions admitted in the texts themselves, the data package with integrity verification, the periodic accountability of the editorial production. The internal circuit is not visible; its public consequences are what it exists to produce.

Does the cycle guarantee the archive is right?

No — it guarantees the archive is correctable in the open, which is the most an honest system promises. The difference between the two promises is the subject of this entire trail.

Where to continue reading?

Along whichever track is missing: the concepts, starting with what a working paper is; or the studies, from beating the trivial benchmark. The two cross constantly — by design.

---

The cycle restarts where the archive began: What a working paper is — and why the house publishes its own →

House reading: today's reading is in the Diário; the archived episodes, in the Atlas; the complete studies, on the research page.

What the cycle does not deliver in an article — the turn applied to a reader's specific question — remains where it always was: in conversation.