Lore No. 2 · verdict undecided

Lore // Entry two

The Gap

Everyone tells you what they want. Fewer show you. The distance between the two is the most honest signal in any room, and ARCS is built around it twice: once as a way to read other people, and once as a promise never to read you too hard.

I · The clock

Two recruiters called an hour apart on the same Wednesday (the full account is With Life, and Without It). I left the first call ten minutes early, because some part of me could not get out of the room fast enough. I stayed on the second ten minutes late, because I did not want to go.

I had not decided anything when I left early or stayed late. My body had. The clock knew before I did which call had someone home.

II · The economist's version

The idea has an older name. In 1938 the economist Paul Samuelson proposed that you stop asking people what they prefer and watch what they choose: preference, revealed by behavior. Everyone else calls it watch what they do, not what they say.

III · Two rules that look like one

ARCS keeps two rules about the gap, and on the surface they point in opposite directions.

The two rules

For othersWeight the revealed. Every recoil war room carries a pair of fields, Stated, watch and Revealed, weight, and the canvas stamps every card: Revealed preference > stated preference. For youNever overrule the revealed. NOTOMATION, primitive one: the tools may scaffold and sequence your thinking, never compress, replace, or overrule revealed preference.

Sources: the recoil war-room renderer (18 war rooms carry the lens) · the ARCS constitution, primitive one (NOTOMATION)

A counterpart who says I need to loop in the team is stated. One who stays past the half hour and asks follow-up questions is revealed. And when the compass reads GREEN on a job offer and you say it feels wrong, the system records the gate as undefined, not GREEN. Your lived experience outranks the inference.

With others, ARCS reads the gap and acts on it. With you, ARCS is forbidden to act on it. The reconciliation is in who holds the pen. Reading a counterpart's behavior informs your decision; you are still the one deciding. Reading your behavior and then deciding for you would make the machine the author of your preference, which is exactly the replacement ARCS was built not to be.

Behavior outranks words, and only the person whose behavior it is gets to say what it means.
IV · The open question, answered

The NOTOMATION spec leaves one question open, in writing: when the operator's stated preference contradicts their revealed behavior over time, should ARCS surface it, or stay silent until asked?

My own record had already answered it, without meaning to. I state that ARCS should close loops. My revealed behavior, counted on 3 October 2026, is 261 receipts filed and one real outcome closed. ARCS surfaced the gap. It did not decide what the gap means.

So that is the answer, and it is now the rule: surface the gap as an observation, never as a verdict. Show the two numbers side by side. Let the person whose numbers they are write the caption.

V · The crack, and the falsifier

The honest weakness: a revealed read is not an observation. It is an interpretation of one. Stayed past thirty minutes is a fact; therefore interested is a guess, and guesses carry the reader's hopes. A rule that says trust the revealed can quietly become trust my reading of you, which is just a stated preference wearing a lab coat.

The falsifier not yet run

ClaimIn the recoil record, the revealed read predicts what actually happened better than the stated read did. TestFor every recoil file with both reads and a post-call close, mark which read the outcome matched. Tally. KillIf the stated read matches as often or more, the lesson drops from law to rule of thumb, and the stamp comes off the canvas. Status18 files carry the lens. Usable closes not yet counted. Told, not cited.
VI · Coda

You do not need to ask anyone what they want. Watch when they stand up to leave.

Then ask anyway, because they are the only one allowed to say what it meant.