08
What the threshold tells you
You have the baseline. You have seen them leave it, and you know which way. This chapter answers the cluster’s last question, and it is the question this whole field answers wrongly: how far does that signal allow you to conclude?
A single threshold crossing gives exactly three things
Not two, not four. Three. And counting correctly matters more than people think, because the fourth thing — the one everybody quietly adds — is what ruins everything.
1 · Sign
Inhibition or arousal. Which way along the axis the other person went. Chapter 07 is the whole of the technique for answering this.
2 · Intensity
How far from their own baseline. Not measured on any absolute scale — only as the distance from that person’s baseline.
3 · Timing
Which topic, which question, which word it is tied to. This is the most valuable of the three, and most readers skip it to rush off elsewhere.
What it does not give
Content. Why they are uncomfortable. What they are hiding. Whether what they just said was true or false. A threshold crossing contains no information about any of those, and no observational technique can extract it, because it is not in there.
The Othello error does not vanish once you have a signal
This is where the threshold principle is most easily misread, so let us be blunt.
A reader coming out of chapters 06 and 07 easily draws a very attractive wrong conclusion: “Ah, so the problem is only that nobody in the lab is pushed over the threshold. If people are genuinely pushed over the threshold in the field, body reading will catch lies.”
No. Look at what actually happens when a heavy question is put to two different people.
The person involved
Hears the question. Knows they are involved. Knows the consequences of being found out. Over the threshold.
The wrongly suspected person
Hears the question. Understands they are under suspicion. Knows the consequences if nobody believes them. Over the threshold.
The same question. The same moment. The same sign, usually the same intensity. And they look alike — not nearly alike, alike.
This is the Othello error, raised in chapter 03, and what has to be added here is this: the threshold principle does not fix it. The threshold principle makes it worse. Because now you have a clear signal, measurable and repeatable — and a clear signal produces a far stronger feeling of certainty than a vague hunch. Certainty rises while the amount of information does not. That is the definition of a good trap.
Timing is the asset
Of the three, timing is the one worth money. The sign tells you which way they went — useful but narrow. Intensity ranks things — useful but easily misread, as the next section will say. Timing is different in kind: knowing exactly which topic pushed the other person over the threshold is knowing where to ask next.
And that is the whole job of a compass. It does not tell you what lies to the north. It only tells you which way north is, and going there to look is your own work. The title of this series is not a pretty metaphor — it is an exact technical description of the tool’s range.
Three sources of noise to rule out before assigning timing to a topic
Timing is worth something only when it really belongs to the topic. Three things push people over the threshold with no bearing at all on the conversation, and they must be ruled out first — not weighed afterwards.
| Source | What it looks like | How to rule it out |
|---|---|---|
| The body cold, coffee, lack of sleep, hunger, a full bladder, pain | The deviation drifts with the length of the session rather than jumping with the topic; it appears equally across every topic | Ask before the session; offer water; take a five-minute break and see whether the deviation returns to baseline |
| The context the room, noise, a third person, the hour, deadlines | The deviation is tied to an event in the room — a door opening, a phone vibrating, someone walking past | Record every event in the room with its timestamp; check against them before assigning anything to the topic |
| ⛔ The observer | You tense first — voice hardens, body leans in, questions come faster — and then they tense after you | Self-check at the start and every twenty minutes. If you yourself have crossed the threshold, the session is not scored |
The third source deserves a longer word, because it is the biggest source of error in the whole trade and the only one you control completely. You cause exactly what you measure. Someone who asks questions in an already tense voice will get tense bodies, and will record them as data about the person opposite. No observational technique detects this error from the inside — it is blocked only by the discipline of self-checking, and chapter 30 gives a whole chapter to it.
What to do after seeing a threshold crossing
Almost everyone’s instinct on seeing a person leave baseline is to press on it there and then. That instinct is wrong, and it is wrong for a technical reason rather than a courteous one.
- Record the topic — no comment, no follow-up question
One line: the moment, the topic, the sign, an estimated intensity. That is the whole of what has to be done on the spot. Do not say “you seem uncomfortable”, because that sentence turns an observation into an event in the room, and from that second onwards you are measuring the reaction to your own remark.
- De-escalate — change the topic, give it time, let them return to baseline
Move to a neutral topic and stay there long enough. If the sign is inhibition, stay longer: a freeze returns to baseline more slowly than arousal, sometimes much more slowly. The sign of a return to baseline is that the channels you recorded at the start come back — illustrative gestures reappear, sentences lengthen, the voice regains its rise and fall.
- Return to that topic with a different question
Genuinely different: a different angle of approach, a different tense, a different grammatical subject. If the first time you asked “why did you leave that job”, the second time might be “if your old company invited you back now, what would you think?”. Asking almost the same sentence again measures how well they remember the question, and nothing new.
- A second threshold crossing on the same topic — that is a repeated hot spot
And it is still only a hot spot. Two threshold crossings on the same topic, through two different doors, rule out most of chance — but they do not rule out the Othello error, they do not rule out an unrelated private matter, and they do not turn themselves into a conclusion. Conclusions come only from independent verification, and “independent” means from a source that is not the person sitting in front of you.
Why you must de-escalate before returning
Two reasons, and the second is the main one. One: for a person far from baseline, every later measurement loses its reference — you no longer know where the deviation is counted from. Two: pushed too far over the threshold, speech closes as well. The higher the pressure, the less information you obtain, even from someone entirely cooperative. Pressing a frozen person is a sure way to learn nothing more in the thirty minutes that remain.
The point that has to be conceded — intensity does not measure importance
Of the three things you may say, intensity is the most abused, because it looks like a scale of seriousness. It is not.
People cross the threshold very hard over something small but embarrassing — a personal debt, a rejection, a family detail they do not want to raise here. And people cross it very lightly over something large but well rehearsed — an event told twenty times already, with the words ready, no longer capable of surprising them.
The precise statement
Intensity ranks discomfort. It does not rank importance. The topic that moves a person furthest is not the most important topic of the conversation — it is the topic most uncomfortable for them, and those two coincide far less often than intuition suggests.
The practical consequence is simple: use intensity to order what you dig into, not to weight your conclusions. The topic heat map in chapter 19 is designed exactly for ordering, and it is valuable precisely because it concludes nothing.
Traps peculiar to the three-chapter cluster
The threshold principle opens a temptation the other chapters do not have, and that temptation has to be named rather than left for the reader to infer: if data exists only above the threshold, why not push people over it?
| Trap | What it looks like | Blocking condition |
|---|---|---|
| ⛔ Pushing in order to have something to read | Raising the pressure continuously because “nothing has shown up yet” | Pushed too far over the threshold, speech closes as well and information falls. Raising cognitive load is permitted only inside the question funnel, and only when the other person is free to leave |
| Reading a freeze as calm | “They were perfectly fine talking about it” | With no baseline you may not use the word “fine” |
| Assigning content to the sign | “Crossing the threshold at that question means there is something to hide” | The Othello error. Sign and intensity may never stand alone in a conclusion |
| ⛔ Your own threshold becoming their data | You tense first, they tense after, you record it | Self-check at the start of every session. If you have crossed the threshold yourself, the session is not scored |
| Intensity becoming importance | The thing that moves them furthest is taken to be the biggest thing | Intensity ranks discomfort; it does not rank the matter itself |
| A catalogue creating the illusion of a threshold crossing | Collecting enough odd signals to see a “threshold crossing” in anybody | A threshold crossing is a gradient against baseline inside a time window, not a total of countable signals |
The boundary
The whole three-chapter cluster comes down to one sentence, and that sentence is not an ethical note added for decoration. It is the condition under which the numbers above mean anything.
The condition for the whole cluster
Work with the threshold only of a person who is free to stand up and walk out.
The reason is both ethical and technical, and the two point the same way. For a person who is not free to leave, the threshold loses all meaning: they were pushed over it by the situation itself before you opened your mouth, so every deviation you measure afterwards is tied to no topic at all. The measurement is not merely ethically wrong — it does not run.
That is why this system does not apply to interrogating someone in custody, does not apply to a superior reading a subordinate who fears losing their job, and does not apply to an adult reading a child. In all three cases the baseline does not exist, the threshold was crossed beforehand, and what you take for reading a person is in fact re-reading the very power structure you stand at the top of.
Chapter exercises
One · Three columns. After your next long conversation, rewrite what you observed into exactly three columns: sign, intensity, timing. If a sentence does not fit into any of the three columns, it is inference — delete it.
Two · De-escalate and return. Pick a low-risk conversation and practise the four-step procedure exactly: record, de-escalate, return with a different question, see whether it repeats. The hardest part is not any of the four steps — it is restraining yourself from pressing at step one.
Three · Count backwards. Recall the last three times you “read” somebody. For each one, ask: did I have their baseline? did I rule out the three sources of noise? did I verify independently? The number left after those three questions is usually zero, and it is worth facing squarely once.
Four · Test it in reverse. Find a time you reached a correct conclusion about someone thanks to “reading the body”. Then ask: if that person had been innocent, how differently would they have reacted at that exact moment? If the answer is “no differently”, then what made you right that day was not body reading.
Reading matrix — the threshold principle cluster
The last time, in full. The highlighted row is the one belonging to this chapter — and it is the only row in the table that is about you rather than about them.
| Observed mode | May be said | May not be said | What to do next |
|---|---|---|---|
| Still at baseline | “Nothing to read yet” | Anything about state | Keep recording baseline; change topic along the funnel |
| Left baseline towards arousal | Sign, intensity, the topic that caused it | Why; true or false | Note the topic, de-escalate, return with a different question |
| Left baseline towards inhibition | Sign, intensity, the topic that caused it | “They are calm”; why | As above, but de-escalate more slowly |
| No baseline to compare against | “Cannot be read” | Everything | Take the baseline first — three minutes of neutral topics |
| ⭐ You yourself have crossed the threshold | Nothing about them | Everything about them | De-escalate yourself first, then retake their baseline. The session is not scored |
Intervention
The de-escalate-and-return rule — the intervention of chapter 08
On seeing a threshold crossing you may not press on it there and then.
De-escalate. Wait for the return to baseline. Come back with a different question. Only a second threshold crossing may be recorded as a hot spot — and a hot spot is still not a conclusion.
| Goal | Action | Mechanism |
|---|---|---|
| Keeping the measurement valid | Write one line, stay silent, change the topic | Commenting on their state turns the observation into an event in the room; from then on you are only measuring the reaction to yourself |
| Confirming a hot spot | Return to the same topic from a different angle, a different tense, a different subject | Two different doors rule out most of chance. Asking the same sentence again only measures how well they remember the question |
| Ordering what to dig into | Use intensity to rank the topics, not to weight conclusions | Intensity ranks discomfort, not importance |
| ⚠️ Do not do | Raising the pressure further once strong threshold signs have appeared | Information peaks at moderate pressure and then falls. Pushed too far over the threshold, speech closes |
| ⛔ Never | Using this cluster on a person who is not free to stand up and walk out | The baseline does not exist, the threshold has already been crossed by the situation itself, and the measurement does not run |
Chapter summary
- A single threshold crossing gives exactly three things: sign (inhibition or arousal) · intensity (how far from baseline) · timing (which topic, which question it is tied to). It does not give content.
- The Othello error does not vanish once you have a signal — it gets worse. The wrongly suspected person and the person involved both cross the threshold at the same question, and they look alike. A clear signal raises certainty while the amount of information stays where it was.
- Timing is the most valuable of the three. Knowing exactly which topic pushed the other person over the threshold is knowing where to ask next. This is the whole job of a compass: it points the direction, not the destination.
- Three sources of noise to rule out before assigning timing to a topic: the body (cold, coffee, lack of sleep, a full bladder — drifting with session length, not jumping with the topic) · the context (the room, a third person, events in the room) · and the observer (you tense first, they tense after, you record it as data about them).
- Four steps after seeing a threshold crossing: record the topic without comment → de-escalate so that they return to baseline → return with a genuinely different question → if it repeats, record a hot spot. A hot spot is still not a conclusion; conclusions come only from independent verification.
- Why you must de-escalate: for a person far from baseline every later measurement loses its reference, and pushed too far over the threshold speech closes as well. Information peaks at moderate pressure and then falls — even with someone entirely cooperative.
- The point that has to be conceded: intensity is not proportional to importance. People deviate very strongly over something small but embarrassing, and very lightly over something large but well rehearsed. Intensity ranks discomfort, not importance.
- Six traps peculiar to this cluster, each with its own blocking condition: pushing in order to have something to read · reading a freeze as calm · assigning content to the sign · your own threshold becoming their data · intensity becoming importance · a catalogue creating the illusion of a threshold crossing.
- The condition for the whole cluster: work only with the threshold of a person who is free to stand up and walk out. For someone without that freedom the baseline does not exist and the threshold has already been crossed by the situation itself — the measurement is not merely ethically wrong, it does not run.
- The de-escalate-and-return rule: on seeing a threshold crossing, do not press on it there and then. De-escalate, wait for the return to baseline, come back with a different question. Only a second crossing may be recorded as a hot spot.
The closing sentence of the three-chapter cluster
The body does not tell you what they are hiding. It says only, once and in one place, that here they can no longer hold still. The rest is the work of the next question.
The nine zones of the body begin in the next chapter. Every signal entry in them carries a reminder line at the top, and now you know what that line means: read only once the threshold has been crossed.