Radar Perene / Archive / science

Voluntary transparency has a cost: what the house opened and what it kept closed

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

Science

"Be transparent" is the cheapest advice in the world — for whoever gives it. For whoever practices it, every opened page is a commitment that never expires: the published error stays published, the old version sits next to the new one, the reader ten years from now will be able to check what the house was saying today. Real transparency is not a declared virtue; it is a debt signed in a public registry. And debts are taken on with criteria.

Not everything the house produces becomes a public page with a DOI — and the boundary between what opens and what stays internal is not embarrassed secrecy, it is a criterion: what becomes public is what can be sustained, maintained and audited indefinitely; what stays internal is what has not yet crossed that standard, or what constitutes the craft that finances the openness itself.

The cost the word "transparency" hides

Opening a study with a DOI in a repository like Zenodo carries three costs the rhetoric of openness tends to omit. The first is irrevocability: the deposit preserves versions forever — whoever publishes a finding that later falls cannot delete it, only revise it in the open, as the house's own series has already had to do. The second is maintenance: public pages age, and aging in public demands continuous review — every opened material is a permanent commitment of vigilance. The third is competitive exposure: what is opened, anyone can use, including those who will never cite the source.

Whoever says "we open everything" either has not measured these costs or has nothing whose outside use would cost anything. Both hypotheses say something — neither of them flattering.

The criterion, stated without the manual

The house's ruler fits in a double question: can this material bear being eternal, and can the house bear keeping it eternal? The published indices answer yes to both — stable concept, documented question, revisions possible in the open — and that is why they sit in the repository with DOIs, even without open formulas. The operational detail of how each such decision is made is an internal document; the outcome of each one is public and checkable, and that is the correct asymmetry: the reader does not need to see the meeting, the reader needs to be able to audit the consequence.

What stays closed falls into two honest categories. The first is material that has not matured — draft studies that will go up when they cross the standard, not before; publishing too early transfers to the reader a risk that belongs to the bench. The second is the craft: the treatment of inputs, the accumulated construction decisions, what economically sustains the operation that pays for the open part. A house that opened its entire craft would not be more transparent — it would be more brief.

The contrast the reader can check today

The declaration above would be mere rhetoric without public counterproof. It exists and has an address: the house's published indices are in the repository with active DOIs and preserved versions — open in concept, closed in formula, with a single deliberate exception of public weights and a data package verifiable by hash. That set is the criterion applied, visibly: each item in the archive sits at exactly the degree of openness the standard described here prescribes for it, and the rest of the archive has that same standard as its declared target.

Voluntary transparency, in the end, is not measured by the volume of what is opened — it is measured by the coherence between what is declared and what can be checked. The declared part is in this article. The checkable part is in the repository.

The criterion also explains what the reader does not find in the archive: promises of future openness with a marked date. Announcing tomorrow's openness today is spending transparency's credit before paying its cost — and the house prefers the reverse order, in which each material appears open once it has crossed the standard, with no pre-announcement. The absence of promises is, here, part of the policy, not a lack of it.

Frequently asked questions

Doesn't keeping something closed contradict the openness discourse?

It would if the house promised total openness. The promise is different: auditable coherence — what is open can be checked by anyone, and what is closed has a declared category (maturation or craft), not a denied existence.

How does this "closed" differ from the black box the trail criticizes?

The black box hides even what it measures and whether it errs. The house's standard opens concept, versions, revisions and data — and closes the treatment that constitutes the craft. The black box's reader can audit nothing; the house's reader can audit what was claimed.

What would make a closed material open?

Crossing the standard: a stabilized concept, documentation able to withstand indefinite scrutiny, and the house's capacity to keep the revision commitment. It is the same path the already-public materials traveled.

Why publish in an external repository instead of the house's own site?

Because the cost is the argument: in the repository, the house cannot erase its own past. Transparency that can be silently undone is worth less than transparency placed beyond the reach of whoever declared it.

---

Continue the trail: Publishing one's own hit rate, even when it is not good →

House reading: today's reading is in the Diário; the archived episodes, in the Atlas.

Applying this openness criterion to a reader's own archive or material is bench work — the house talks about it outside the article.