27
Disciplined Scoring
A scoring system is not there to produce a number. It is there to block — to make drawing conclusions from weak data systematically hard, instead of leaving it to how careful you happen to be at the moment you are most tired.
Weights by fakeability
The weights in the table below are assigned on two criteria: the signal's fakeability and its level of empirical support. Not how dramatic it is, not how visible it is, and above all not how certain it makes you feel.
The accumulation rule
Action thresholds
Why rank is more trustworthy than an absolute score
An absolute number depends on a great many things you do not control: the length of the meeting, the lighting, whether you can see their legs, how alert you are that day. Two different meetings give you two scales that cannot be compared.
But rank is stable. Within one meeting, the fact that topic A scores twelve while topic B scores four is usable information — whether or not the absolute scale is right. Use this table to rank topics; do not use it to measure how much someone is lying.
The observation sheet template
This template prints onto one side of A4 or one notebook page. Writing by hand while you talk is normal in most working contexts; where it is not, fill it in straight afterwards, while the memory is fresh.
THE SUBJECT: ________________ DATE: __________ SETTING: ______________ MY ROLE: ___________________ (stranger / acquaintance / colleague / close) MY OWN POSTURE at the start: _______________________ (check the contagion trap before reading anything) ─── BASELINE (first 5–10 minutes, mandatory) ────────────────────────── Speech rate: ______ Blinks per minute: ______ Eye contact %: ______ Default hands: ____________________ Default legs: __________________ Baseline movement: ________________ Default pacifying: ______________ Distance they chose: ________ cm ─── LOG ───────────────────────────────────────────────────── T TOPIC Ô FIDG DESYNCHRONY OTHER SIGNALS SCORE ──── ─────────────── ── ──── ─────────── ───────────── ──── ____ _______________ __ ____ ___________ _____________ ____ ____ _______________ __ ____ ___________ _____________ ____ ____ _______________ __ ____ ___________ _____________ ____ ____ _______________ __ ____ ___________ _____________ ____ ____ _______________ __ ____ ___________ _____________ ____ ─── HEAT MAP ──────────────────────────────────────────────── Topics +++ : ______________________________________________ Topics ++ : ______________________________________________ Topics + : ______________________________________________ ─── CELL TRAJECTORY ─────────────────────────────────────────────────── ____ → ____ → ____ → ____ → ____ Turned bad at: _____________________________________ ─── CONCLUSION ──────────────────────────────────────────────────── Leading hypothesis: ______________________________________ Alternative NOT yet ruled out: ________________________ Needs independent verification by: ______________________________
The last two lines are the two that matter most
This sheet deliberately forces you to write the alternative hypothesis not yet ruled out right beside the leading one. If that box is empty, you have not finished — because there is almost never a case in which no alternative hypothesis is left. And the last line forces you to name a specific action, instead of letting the meeting end in a feeling.
Three common scoring errors
Scoring backwards
You already had the conclusion, then went looking for signals to add points to it. How to recognise it: the alternative hypothesis always scores low, in every meeting, with everyone.
The imaginary baseline
You know this person, so you “know what they are normally like”. But you know what they are normally like with you, in a different context. The baseline has to be established again for every context.
Class drift
A class C signal gradually comes to be treated as class A because it “always holds true in my experience”. This is how class X is reborn. Chapter 30 has a way to measure it.
Chapter exercises
One. Copy the sheet template into your notebook and use it in three meetings. Do not try to fill it completely — the aim is to find out which boxes you habitually leave blank.
Two. Take a conclusion you once drew about someone. Score it again with the weight table and the three multipliers. Does it still stand?
Three. For one week, every time you are about to score, ask about the context factor first. Count how many times it comes out zero.
Chapter summary
- A scoring system is not there to produce a number. It is there to block — to make drawing conclusions from weak data systematically hard, instead of relying on how careful you happen to be at the moment you are most tired.
- Weights are assigned on fakeability and level of empirical support. Not on how dramatic it is, not on how visible it is, and above all not on how certain it makes you feel.
- Six classes, from S, five points — pupil, sweat, blushing, cold damp hands, breathing, dry mouth — down to X, zero points. Class X is not “few points”, it is zero, absolutely: gaze aversion, gaze direction, microexpressions as a marker of deception, consistency as evidence.
- Two factors that multiply to zero: no baseline established, and a plausible alternative explanation still not ruled out. These are the two errors experienced people make more often than beginners, because familiarity creates the illusion that the baseline is already known.
- The cluster factor: ≤ 2 signals × 0 · 3 signals × 1 · ≥ 4 pointing the same way × 1.5.
- Four bands: 0–3 noise, ignore it, make no note · 4–8 record the topic, come back to it later by an indirect route · 9–15 dig with the funnel and unanticipated questions, DO NOT ACCUSE YET · 16+ deploy the evidence strategically, prepare to verify.
- NEVER conclude “lying” from this table, at any score. Even at the highest level, its output is an action, not a conclusion.
- Rank is more trustworthy than an absolute score. The absolute scale depends on the length of the meeting, the lighting, whether you can see their legs, how alert you are that day. But topic A scoring twelve while topic B scores four is still usable information. Rank the topics; do not measure how much someone is lying.
- The baseline is taken in the first 5–10 minutes, without exception. And the first line of the sheet is your own posture.
- The last two lines of the sheet are the two that matter most: the alternative hypothesis not yet ruled out and the method of independent verification. An empty alternative box means you have not finished — there is almost never a case in which no alternative hypothesis is left.
- Three scoring errors: scoring backwards, recognised when the alternative hypothesis always scores low with everyone · the imaginary baseline, because you only know what they are normally like with you, in a different context · class drift, when a class C signal comes to be treated as class A. Class drift is how class X is reborn.