Radar Perene / Archive / science

Why we never write \"you should\": the P7 filter explained

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

Science

There are sentences the reader will never find on any page of this house. "You should." "We recommend." "The best option is." Not for lack of occasion — occasions arrive daily — but because there stands, between writing and publication, a rule that blocks those constructions without exception. Internally it carries a protocol name: the P7 filter. This article explains why it exists, what it forbids, and what that prohibition says about how the house understands its own craft.

The P7 filter is Radar Perene's editorial rule vetoing prescriptive language on every public surface: no text of the house tells the reader what to do. What the data show is described; the decision about what to do with the description belongs, entirely, to the one reading.

The distinction the filter protects is old in science and rare in markets: the border between describing and prescribing. "Historically, episodes similar to this one ended in very different ways" is a description — verifiable, datable, falsifiable. "Seize this moment" is a prescription — unverifiable by nature, because it embeds a bet on the future and a guess about the life of whoever reads it.

The first reason: the future does not cooperate

The house organizes probabilities; it does not manufacture certainty. That sentence, which closes every published piece, is not decoration — it is the epistemological premise of everything that leaves here. An honest historical archive shows that similar episodes ended differently: the same configuration that preceded recoveries preceded declines. Whoever knows that archive knows that any "do X" is a simplification the data itself overrules.

Prescription demands two certainties no one has: about what the future will do and about what suits a specific reader. Description demands only one thing the house does have: the record of what happened.

The second reason: the house does not know the reader

Every serious recommendation is tailor-made — it depends on horizon, obligations, tolerance for loss, circumstances that vary from person to person. A public text speaks to thousands of readers at once; a published prescription treats them all as one. That is why financial regulation, in Brazil and elsewhere, rigorously separates descriptive analysis from individual recommendation: they are different crafts, with different responsibilities. The house chose the first craft and writes inside it.

The image that best describes the choice: the house works as a cartographer, not a guide. The cartographer delivers the most faithful map it can — reliefs, cliffs, roads already traveled — and the traveler chooses the trail, being the only one who knows where they want to arrive and how much weight they carry.

The third reason: incentives stay clean

There is a less obvious and perhaps more valuable effect. Whoever prescribes comes to be measured by the outcome of their own prescriptions — and gains, in the same instant, an incentive to defend them, dramatize them and hide whatever contradicts them. Whoever describes is measured by another ruler: fidelity to the record. A description's error is corrected by the next description; a prescription's error becomes a liability.

The archive holds an episode that shows the rule at work. In May 2026, the house's statistical engine flagged an extreme reading in one of the calmest series in the database — and the memo that recorded it declined the headline in the same paragraph, classifying the move as series noise, not an economic event. No "attention, time to act"; a record, a weighing, and the decision not to dramatize, all published and dated. In numbers: the house's two indices are released on a 0-to-100 scale, published daily — numbers that describe a state, never an order.

What the filter forbids, in practice

The visible layer is a list of constructions vetoed in any position of the text — title, body, caption, FAQ answer: the action imperatives, the "we recommend," the "the best option is," the urgency superlatives that sales language uses as punctuation. The less visible layer is a second review pass that hunts the disguised prescription: the rhetorical question that pushes ("wouldn't now be the time to..?"), the insinuating conditional, the praise of an outcome that works as a nudge. Like every editorial rule of the house, the filter is a condition of publication, not a style suggestion — a text that fails does not ship.

The attentive reader will notice this very article quotes the forbidden sentences. Quoting is not using: the mention in quotation marks describes the object of the rule, and describing is exactly what the rule permits.

Frequently asked questions

Doesn't the P7 filter make the texts evasive?

The house's bet is the opposite: removing prescription forces the text to become more precise, not less. Without the shortcut of "do X," what remains is the hard work — saying what happened, how often, with how much variation. Evasion usually lives precisely in vague advice.

Isn't describing a historical extreme already an implicit recommendation?

It is a real risk, and the second review pass exists because of it. The operational difference: an honest description includes what contradicts the tempting reading — the episodes where the extreme persisted, the times the ruler misled. The implicit prescription is dismantled by publishing the counterexample alongside.

Where does the name "P7" come from?

From the internal numbering of the house's editorial protocols. The name matters less than the function; it stuck because it is short and became working vocabulary.

Does the rule also apply to paid reports?

It applies to every editorial surface of the house, public or subscriber-only. The border between describing and prescribing does not move according to whether the reader pays.

The filter decides how the house writes; the next question is how the house decides what to trust — how the Atlas tests whether a precedent is a stylized fact or a coincidence →

House readings: today's record, in the Daily · the memory of episodes, in the Atlas.

For those researching their own editorial process, the experience accumulated with this rule is a subject the house discusses in conversation.

Characters: Method