How we document a new technique

The reference library holds thirty-four techniques. We are often asked how they got in, and above all what we turn away. Here is the method, with no mystery to it: it is fairly boring, and deliberately so.

A technique does not get in because it is annoying. It gets in because somebody other than us can spot it, on posts we did not choose. That is the only criterion that stops a reference library turning into a list of opinions.

The four steps

01NoticingSomeone spots a pattern that keeps recurring. At this stage it is nothing: an impression, often wrong. We note the URL of the post and move on.

02CollectingWhen five unrelated posts carry the same pattern, we open a file. We gather about fifty, across different accounts, different sectors and, where possible, two languages. It is the longest step: a few weeks.

03Defining it in the negativeThe step that eliminates the most candidates. We write down what the technique is not, and hunt for counter-examples. If we cannot describe a post that resembles it but does not qualify, the definition is too broad. It is dropped.

04The blind testTwo people annotate thirty posts without conferring, ten of which do not carry the technique. If they disagree on more than three, the entry goes back to step three.

What we turn away, and why

Roughly two candidates in three never make it. The reasons are almost always the same.

Matters of taste, first. "Posts that are too long" or "emoji abuse" are not techniques, they are preferences. A technique describes a device and its effect, not what we happen to like.

Assumed intentions, next. You cannot document "the post that wants to impress": intent is not observable, only means are. Every entry we publish describes means.

And patterns that are too rare. A technique met once in five hundred posts may well be real, but it clutters more than it helps. We keep it in reserve.

"If a second person cannot spot it without being told, the entry is not ready."

THE ENTRY CRITERION

What an entry contains

Always the same things, in the same order: a one-sentence definition, the signals that let you spot it, a real example rephrased so as not to expose its author, and the question to ask when you meet it.

That last line is the one we care about most. An entry that only names produces cynicism; an entry that hands you the question produces a reading. It is the difference between "this is a manipulation technique" and "coached toward what, with what results?".

WHAT THE EXTENSION DOES

Every technique flagged in a post links through to its full entry, with its definition and its question. The whole catalogue is readable without an account.

An entry is never finished

Techniques wear out, as we have written elsewhere. A published entry is therefore reread periodically: its examples age, its signals shift, and some will end up describing a practice nobody follows any more.

When that happens we do not delete it; we date its decline. Knowing that a technique worked until 2025 and then stopped is part of what we are trying to document. It is arguably the most interesting part.