Appearance
10.9 — Working Out Someone's Situation, From Evidence
A colleague has been arriving late, missing details he never used to miss, and declining every social invitation for two months. He is short in meetings and apologises afterwards.
The intuitive read: he has become unreliable, or he has checked out, or he does not like you any more.
The disciplined read asks a different question: what set of circumstances would produce exactly this pattern in an ordinary person?
Illness — his own or somebody's. A relationship ending. Money. A second job. A new baby. Sleep destroyed by any of the above. Depression. Being managed out and knowing it.
Most of that list produces precisely this pattern, and none of it is about you or about his character.
This page is the method: how to build an accurate picture of what is going on with somebody, from evidence, rather than from impression.
Why this is the useful skill
From 10.5, the middle layer — the situation — explains most puzzling behaviour and is the layer nobody reads, because it is not visible.
It is not visible, and it is inferable. Behaviour leaves traces, and a situation constrains behaviour in patterned ways. This is the part of reading people that genuinely improves with practice, because unlike face-reading it runs on evidence and can be checked.
The method
Six steps. It sounds elaborate written down and takes about two minutes once it is habitual.
1. Separate observation from inference
Write only what a camera would have recorded.
Arrived after 9:30 on eleven of the last twenty days. Not he's become slack.Declined four invitations. Not he's avoiding us.Two errors in the March figures. Not he's careless now.
This step alone changes the answer more often than the rest of the method combined, because the inference was doing all the work and it was not in the evidence (2.2).
2. Establish when it changed
The single most informative question, and the one most often skipped.
When did this start? Behaviour that has always been this way is disposition. Behaviour that changed on a date has a cause with a date.
Then: what else happened around that date? A reorganisation, a project ending, a colleague leaving, something you heard about their family. The change point is where the explanation usually is.
3. Generate at least four hypotheses
Force the number. Three is where it starts to break (5.8).
And include the boring ones, because they are the most common: tired, unwell, money, someone ill at home, a relationship in trouble, overloaded, depressed.
Include one about yourself — did I do something? — and hold it lightly, since it is usually not the explanation and occasionally is.
And include the possibility that nothing is wrong. People have flat months.
4. Ask what each hypothesis predicts
This is what makes it a method rather than a list.
For each candidate, name something that would be true if it were the right one, and something that would be false.
If he is overloaded: his output should be down across everything, not just the parts involving me. He should be declining new work. If he is avoiding me specifically: the pattern should show up with me and not with others. If someone at home is ill: the lateness should cluster around specific times, and he should be leaving promptly rather than staying late. If he is depressed: it should show in things he used to enjoy, not only at work.
Now you have checkable differences, and you did not have them ten seconds ago.
5. Look for the discriminating evidence
Then go and look. Not intrusively — most of it is already available.
Is the pattern specific to you or general? The cheapest and most useful check, and it resolves the "is it me" question immediately.
What is the timing pattern? Late in the mornings only, or throughout? Leaves promptly, or stays late?
What is he still doing normally? What has been dropped tells you where the load is.
And what have other people noticed? Asked as a genuine question rather than as gossip: "has Ravi seemed all right to you?"
6. Then ask
The step people skip because it feels intrusive, and it is usually welcome.
One sentence, no agenda, private, and give them the option to decline:
"You've seemed a bit stretched the last couple of months. Everything all right? No need to go into it if you'd rather not."
The escape clause at the end is what makes it safe to answer honestly. Without it, the question is a demand.
And then be quiet. Most people say more than they intended to if you do not fill the pause (4.9).
The traces that carry the most information
Where to look, in rough order of usefulness.
Change points. Anything that started or stopped on a date.
What has been dropped. People shed in a consistent order under load: first the optional and pleasurable, then the social, then the discretionary parts of work, then the core (6.7). Where somebody is in that sequence tells you how much load they are under, and it is visible from outside.
Time patterns. When they are available and when they are not. Mornings, evenings, particular days. A person managing something at home has a shape to their week.
What they still protect. Whatever survives when everything else is dropped is what matters most to them.
Money signals. Changes in spending, in what they decline, in how they talk about cost. Frequently the earliest visible sign of something serious and the most awkward to ask about.
Physical signals. Weight, sleep evident in the face, illness, drinking. These are not intrusive to notice and they are frequently the first evidence of anything.
And what they talk about, and what they have stopped talking about (10.4). Omission is a signal.
The ethical line
This is a method for understanding people, and the same method can become surveillance. The line is not fuzzy.
Fine: noticing what is publicly visible, thinking about it carefully, and asking the person.
Not fine: reading their messages, tracking their movements, checking up on them with others behind their back, or assembling a file.
The test is simple: would I be comfortable telling them exactly how I formed this picture? If not, that is not investigation, and 13.7 covers what it becomes when it is done to a partner.
And the purpose matters. Understanding somebody in order to help or to work with them is one thing. Understanding somebody in order to use it against them is another, and it is 1.7's cunning.
The most common finding
Worth stating because it recurs so consistently that it is nearly a rule.
When somebody's behaviour changes and you cannot explain it, the answer is usually one of: someone is ill, money, sleep, a relationship, or their own mental health.
Almost never: they have changed as a person, or it is about you.
Starting from that list, rather than from your own hypothesis, will make you right more often — and it will make you a considerably better colleague, friend and partner, because the person on the other end is dealing with something and everybody else has concluded they became unreliable.
What to do with this
Run the six steps once this week on somebody whose behaviour has puzzled you. Written down. It takes ten minutes and it will produce at least one hypothesis you had not considered.
Always ask when it changed. If you take one thing from this page, take that question.
Force four hypotheses, including two boring ones.
And then ask them, with the escape clause. "You've seemed [X] lately. Everything all right? No need to say if you'd rather not."
That sentence, used two or three times a year, will do more for the people around you than the entire rest of this Part — because most people going through something difficult are being read as having become unreliable, and nobody has asked.
Next: 10.10 turns the method around and shows how it is used against people — cold reading, and why a stranger can appear to know things about you that they could not possibly know.