Radar Perene / Archive / science

Levels of research: from curiosity to working paper, from preprint to peer review

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

Science

The vocabulary of research reads like a catalogue of different objects: working paper, preprint, refereed article, technical note. Read from the outside, it suggests distinct genres, each on its own shelf. Read from the inside, it describes something else — a single study climbing a staircase, gaining exposure and validation at each step. Whoever understands the staircase reads any scientific text knowing exactly how much weight to give it.

The levels of research are stages of exposure and validation of the same work: from private notes to internal exploration, from the public unrefereed study (working paper or preprint) to the article approved by referees — and, beyond it, to replication and citation, which keep validating the text after publication. No map of this funnel circulates in Portuguese; this article is the house's attempt to draw one.

The invisible step: curiosity and the bench

Every study begins as material no one will see: notes, readings, exploratory tests, dead ends. This step carries no prestige and serves an enormous function — it is where most ideas die cheaply. A mistake made here costs hours; the same mistake discovered three steps up costs reputation.

The next step is still internal, but already has form: the declared hypothesis, the test run under protocol, the result written for the house's own archive. We call it the bench. What separates it from curiosity is discipline — as described in How research is born, this is where free exploration parts ways with the declared test.

The first public step: working paper and preprint

When the study holds up, it can choose to be born in public before any external validation. That is the step of the working paper — the vocabulary of economics — and of the preprint, the equivalent term in the natural sciences. The author bears the risk alone: the text has a date, an author and an identifier, but no referee has filtered it.

It is a step, not a shortcut. The public unrefereed study gains three things the drawer manuscript lacks: precedence (the date fixes who claimed it first), criticism (any reader can contest it) and traceability (a DOI makes it citable forever). What it does not gain is the stamp — and an honest text at this step says so on its cover.

The filter step: peer review

The refereed article is the same text after crossing the sieve of anonymous referees and an editor — the process described in What is peer review. What changes is not the truth of the study but the nature of the guarantee: other specialists searched for flaws and did not bring the argument down. The filter is slow, fallible, and remains the most demanding one science has built.

The step that never ends

The staircase does not end at publication. A refereed article no one can reproduce is worth less than a working paper replicated by three independent groups. Citation, replication and post-publication criticism are the continuous validation — the step where time is the referee.

| Level | Who has seen it | What it guarantees | |---|---|---| | Curiosity / notes | Only the author | Nothing — and it need not | | Bench (internal study) | The author, under protocol | The discipline of the declared test | | Working paper / preprint | Any reader | Date, authorship, precedence | | Refereed article | Referees + editor | An active search for flaws, survived | | Replication and citation | The community, over time | Survival beyond the author's hands |

Where this house operates

Radar Perene's public output lives today at the working paper step — by choice, not by accident. In numbers: six studies in the RPWP series are public on Zenodo, each with its own DOI; the first of them, on the Perene Risk Index, under identifier 10.5281/zenodo.21325743. Each says on its cover what it is: research in progress, exposed to criticism before any stamp. The internal steps — the bench, the protocols, the tests that did not survive — sustain those six texts and remain where they belong: out of the shop window.

Which step a question deserves is a case-by-case decision, and less obvious than it looks. Not every question needs to climb the whole staircase; some yield everything they have at the bench, others only make sense if they face referees. There is no fixed ruler — there is judgment, and it is learned by erring.

Frequently asked questions

Can a study skip steps?

It can, and some do — manuscripts go straight to journals without circulating first. In economics, though, the route with a stop at the working paper is the classic flow: the text matures in public before facing the formal filter.

Is a working paper worth less than a refereed article?

It guarantees less, which is different. The working paper guarantees date and authorship; the refereed article adds the specialists' sieve. The informed reader does not ask "is it worth anything?" but "what guarantee does this step offer?".

Should every study reach the top of the staircase?

No. Some research exists to inform a decision, feed a product or document a method — and fulfils that entirely at the public unrefereed step. The top costs years; not every question deserves them.

What decides when a text is ready to climb a step?

At the internal steps, the house protocol; at the public ones, other people's criticism. The honest answer is that the text never decides alone — the confrontation with readers decides, and each step widens their circle.

---

Continue the trail: Robustness testing: why a result that only works one way is not a result →

House reading: the most daily step of all — the dated record — is in the Diário.

Deciding what level of rigor a question deserves is a conversation, not a fixed ruler — and it is the kind of conversation the house has, when sought out.