Radar Perene / Archive / science

Research on demand: when research does not depend on a department

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

Science

For most of the twentieth century, a serious research question had one path: finding a university department — or a bank's research desk — willing to adopt it. Data was expensive, literature sat locked in physical libraries, and credibility lived on letterhead. Whoever had a good question and no institution kept the question. That arrangement grew old, and the piece that retired it was not ideological: it was infrastructure.

Research on demand is investigation conducted with scientific method — declared hypothesis, traceable data, documented testing — starting from a question brought by whoever commissions it, rather than by a department's agenda. The rigor is that of academic research; what changes place is the origin of the question.

The term has no established tradition in Portuguese — this house uses it because it needed to name what it already practices. And the practice only became possible because three institutional monopolies fell almost at once.

The three monopolies that fell

The first was the monopoly on data. A central bank with thousands of open series, a statistics institute with searchable archives, treasury and securities regulators publishing raw datasets: the input that once justified an entire department is now public, free and documented. The raw material stopped being the privilege.

The second was the monopoly on record. Scientific credibility has always depended on public record — who claimed what, when, on what basis. Open repositories such as Zenodo, operated by CERN, hand any researcher what only the institutional seal used to grant: dated deposit, permanent identifier, preserved versions. An independent study with a DOI is citable and auditable exactly like a university working paper — what changes is the author's address. The house described that circuit in what is a preprint and where ours live.

The third was the monopoly on judgment. The critical reading a department performed in an internal seminar now happens in public: an open study is examined by whoever wishes, as many times as they wish. Scrutiny did not become gentler for the independent researcher — it became more visible.

What "on demand" adds to the arrangement

Independent research describes who does it; on demand describes where the question comes from. The distinction matters because the traditional academic agenda answers to incentives of its own — what is publishable, what is theoretically fashionable, what earns citations. Legitimate questions are orphaned for not fitting those incentives: a specific pattern in one sector, a series too short to interest a journal, a methodological doubt from someone running a real business.

Research on demand is ready-to-wear inverted: instead of the reader choosing among finished studies cut for the average, the study is cut over the question that exists. What does not change in the tailored cut is the same thing serious tailoring never compromises — the fabric. Method is method: hypothesis before test, data with provenance, results documented even when they displease. A commissioned investigation that can only end in a pleasing result is not research; it is stationery.

The proof that the arrangement works

This article's argument is not hypothetical — this house is its own test case. Radar Perene operates outside any department and maintains verifiable public output: in numbers, six working papers with active DOIs on Zenodo and an open data package — series, dated events and prior registration of hypotheses — whose integrity is verifiable by cryptographic hash, deposited under the identifier 10.5281/zenodo.21399426. The complete list lives on the research page.

Two traits of that output say more than the volume. First, one study in the series carries a published conclusion that is negative — the tested hypothesis did not hold, and the text says so under a permanent identifier. Second, new versions openly revise findings of older ones, with the history preserved in the repository. Those are the two behaviors that separate research from promotional material, and neither requires an institutional badge — they require only the willingness to be audited.

What a question needs to have

Not every question becomes research, and part of the on-demand craft is saying so early. A researchable question has three properties: it is specific enough to be tested ("does this pattern exist in this series?" rather than "is the market doing well?"); it has reachable data with declarable provenance; and it admits "no" as an answer — the question whose only acceptable answer is known in advance belongs to another genre. The full road between a loose question and a testable hypothesis is the subject of the article that opens this trail.

Frequently asked questions

Is research on demand the same as consulting?

There is overlap, but the center is different. Classic consulting delivers experienced opinion; research on demand delivers a documented test — hypothesis, data, method and a result another researcher can audit. The difference shows in the product: an opinion report cannot be replicated; a study can.

Doesn't a commissioned result tend to please whoever pays?

The risk exists and has a name in the literature — sponsorship bias. The antidotes are structural: hypothesis registered before the test, data with open provenance, and the declared commitment to document whatever result arrives. Whoever commissions serious research accepts "no" as a possible answer; that is what distinguishes a commission from an errand.

Is independent research worth less than university research?

A study's value lives in the method it exhibits, not in the author's address. Work with traceable data and a public deposit is auditable whatever its origin — and the audit, not the badge, is the best thing science has.

What kind of question fits this format?

Specific questions with reachable data: verifying a pattern in a series, retesting a relation folklore takes for granted, measuring an effect in a segment the literature has not covered. Questions about predicting the future stay out — by method, not by modesty.

This trail closes where all research begins: on the road between a loose question and a hypothesis one can test — how a research project is born →

House readings: the public daily record · the precedents, in the Atlas.

The house keeps the category in operation: specific questions find an interlocutor in conversation, outside the public article.