field notes ⋅ The Assessment
Reading a new person
If the person who taught you what people are like was someone who admired you while you were useful and turned on you when you were not, then dating carries a particular problem: the early evidence looks identical either way. Everyone is warm at the beginning. This is a manual for the tests that actually separate them — built from the pattern rather than from a checklist.
on this page
the pattern to test for
Being treated well is not the signal. Almost anyone treats you well while you are an asset — while your success reflects on them, while you are pleasant to be near, while you are not asking for anything. That period contains almost no information.
The signal is conditionality. Admiration that holds while you are doing well and curdles when you struggle. Pride in your achievements that lasts exactly until you outgrow the person expressing it. If that shape is familiar from your own family, you are already equipped — you have seen the whole arc at close range, which most people have not.
Warmth is easy. What matters is what happens to it when you stop being convenient.
So the tests below are not about detecting cruelty early. Cruelty is not usually visible early. They are about deliberately creating the small conditions under which conditional warmth reveals itself, while the cost of finding out is still low.
The Pedestal Is the Red Flag, Not the Good Part
This is the hardest one to accept, because it arrives feeling like the best thing that has ever happened.
Intense early certainty — you are different, you are the only one who has ever understood me, I have never felt this before, this is unusually fast and that must mean something — is not strong feeling. It is a valuation made before he knows you.
And a judgement formed on no evidence can be revised on no evidence. Someone who decided you were extraordinary within three weeks can decide you are worthless within three months, and nothing will have changed except his need. The height of the pedestal is a fairly reliable predictor of the drop, because they are the same mechanism running in two directions.
Real regard is slower and more boring. It accumulates. It refers to specific things you actually did rather than to what you represent, and it does not require you to be exceptional in order to continue.
six tests you can actually run
None of these require confrontation, cleverness, or catching anyone out. Most require only that you stop smoothing things over and watch what happens in the gap.
I · Watch what happens when you struggle
Not when you shine. Be ill, be behind on work, be unable to go out, need something small. Does the warmth hold? Or does it cool, get impatient, become advice that is really criticism? This is the single most informative thing available, and it arrives on its own eventually — you do not have to manufacture it.
II · Watch what happens when you win
A raise, a finished project, praise from someone whose opinion carries weight. Safe people are straightforwardly pleased. Watch instead for the immediate deflation: anyone could have done that, the abrupt change of subject, the joke that is not one, the good news somehow becoming a conversation about him.
III · Watch what happens when you struggle
Not when you shine. Be ill, be behind on work, be unable to go out, need something small. Does the warmth hold? Or does it cool, get impatient, become advice that is really criticism? This is the single most informative thing available, and it arrives on its own eventually — you do not have to manufacture it.
II · Watch what happens when you win
A raise, a finished project, praise from someone whose opinion carries weight. Safe people are straightforwardly pleased. Watch instead for the immediate deflation: anyone could have done that, the abrupt change of subject, the joke that is not one, the good news somehow becoming a conversation about him.
what costs you nothing
Tests four and five are worth separating out, because they are free. They require no risk, no disclosure, and no waiting — they arrive unprompted in ordinary conversation, and they are about how he treats other people, which means you learn it before anything is at stake for you.
How someone speaks about people who are not in the room is how they will speak about you when you are not in the room.
Mockery of an ex's mental health. Contempt for a sibling who is too sensitive. A friend who is in therapy for no reason. That is not gossip — it is a demonstration of the filing system your own disclosures will enter.
Believe it the first time.
what you are Looking for, not Just avoiding
A list of warnings produces vigilance, and vigilance is exhausting and inaccurate. It helps to know the shape of the thing you actually want, so you can recognise it when it is in front of you rather than only scanning for threats.
signals Worth noticing
- Admits fault without a performance of self-hatred attached
- Accepts a no without renegotiating it
- Is pleased by your successes without needing to match them
- Gets details slightly wrong and remembers the thing that mattered
- Is the same person with strangers as with you
- Tolerates you being unavailable without punishing it
- Curious about you rather than certain about you
signals Worth slowing for
- Intense certainty about you very early
- Every ex was crazy, unstable, or obsessed
- Apologies that never change the behaviour
- Warmth that fluctuates for reasons you cannot trace
- Repeats a small detail back with unusual precision
- Presses for depth you have not offered
- Charming to the useful, cold to the rest
Note the fourth entry on the left. Safe people usually get the specifics slightly wrong and remember what actually mattered. Unusual precision about small disclosed details — repeated back weeks later, in a conversation it has no business being in — is not evidence of good listening. It is evidence the detail was stored somewhere retrievable.
when the apology comes
At some point a person you are testing will get something wrong and say sorry. This is the moment most worth reading carefully, because a good apology and a performed one feel almost identical in the room — the difference is in what is actually inside it.
There is an unusually precise old framework for this. In Jewish tradition, real repair — teshuvah, literally return — is not an apology and not a feeling. It has a fixed shape, and the shape works as a detector: genuine repair contains all of these, and the performance contains almost none.
I · acknowledgement
Names the specific thing, plainly. Not if you were hurt, not mistakes were made — the actual act, said without fog.
II · facing the Harm
Understands what it did to you, not the abstract fact of being wrong. Looks at the real injury directly rather than around it.
III · ceasing
The behaviour stops. Present tense. An apology offered while the conduct continues is management, not repair.
IV · restitution
Repair aimed at you, in whatever form fits — not good deeds performed elsewhere to settle the feeling privately.
V · the difference Choice
The one that matters. Proven only when the same situation returns and they choose differently. Demonstrated over time, not promised in the moment.
Tears are the cheapest part. Distress at being confronted is not change.
Score the apology against the five. Only crying and a plea to move on is zero of five — and the absence is the answer. Watch, too, which way it points: if it keeps returning to how bad they feel, the person apologising is still at the centre of the story. Real repair moves you to the centre. The performance never does, because its whole purpose is to relocate the discomfort back onto you.
This is the version that will test you most, because it looks like exactly what you are hoping for. Someone repeats the behaviour, apologises with real tears, and adds that they are working on it in therapy.
It is still zero of five — and in one way it is harder to resist than a bare apology, for three reasons worth naming:
It comes with proof of effort attached, which buys more chances than the conduct earned. It flips the burden — now staying wary means doubting someone who is trying, so you become the unsupportive one. And "I'm working on it" never expires: it can absorb the same harm indefinitely, because there is always more work to point to.
Therapy is a tool for someone who has already decided to change. It is not the change, and it is not evidence of it.
A person can attend for years and choose the same thing every time — some even use it to get better at describing the harm while continuing to do it, which reads as depth and functions as cover. The test does not move: not what they understand, not what they are working on. What they did the next time it counted.
Empathy is owed to repair, not to its performance.
When the tears come and the behaviour has already repeated, you will often be expected — by them, by friends, by the whole script about forgiveness — to soften. But where no repair has happened, there is nothing there that empathy is the right answer to. Tears are a request. You are allowed to decline it.
Being pressured to feel it does real damage, and it is worth knowing the shape: it makes you responsible for managing their pain; it builds a bind where staying wary is cruel and softening puts you back in range; and it teaches you to distrust your own alarm — the exact faculty that keeps you safe. It works by exploiting the empathy you kept, turning the intact part of you into the lever that pulls you back.
Hold all three at once: this person may be suffering · their suffering is not yours to fix · understanding why someone became harmful never obligated you to stay in reach of the harm. Compassion and access are different things. You can grant the first, even sincerely, and still refuse the second for good.
None of this means demanding a perfect five-part apology from someone over something small. Ordinary people apologise clumsily and mean it. The test is not eloquence — it is direction. A real apology, however awkward, points at you and at their own future conduct. A counterfeit one points at their feelings and asks you to tend them. Once you have seen the difference, it is difficult to unsee.
time beast vigilance
You cannot out-detect a skilled performance in the first month, and attempting it will exhaust you while producing false positives on perfectly ordinary people who happen to be nervous.
But nobody sustains a persona for a year. Maintenance is expensive and consistency is not, so the gap opens on its own if you give it time. The whole strategy is therefore not sharper detection but slower investment — keeping the stakes low for long enough that the pattern surfaces before you are entangled in it.
Evidence first. Then trust. Then depth.
Not because people deserve suspicion — because slowness is the only filter that has ever worked, and it admits a different kind of person than speed does. People who move fast get in first. The ones who would have been safe are often still standing outside, waiting to be let in gradually.
you will not catch them all
Some people are genuinely skilled at this, and being deceived by a practised deceiver is not a failure of judgement. It is what practised means. Any framework promising you a perfect filter is selling vigilance, and vigilance costs more than it returns.
What you can actually control is not whether you are ever wrong, but how much is at stake when you find out. How entangled. How financially exposed. How far from your own friends. How many months, rather than years.
The goal is not to be uncatchable. It is to stay recoverable — and that is a matter of pacing, not perception.
related manuals
continue
The tactics named here are covered in depth elsewhere in the codex — love bombing, intermittent reinforcement, and the trauma bond in the manipulation manual, and the disclosure sequencing in the security briefing.