Notice one ordinary flaw, deliberately
The counterintuitive intervention. Naming something mundanely imperfect about them while you feel they are perfect keeps the whole person in view and softens the eventual inversion.
It is not that your judgment of someone changes. It is that the filter changes, and both versions feel equally like the truth while you are inside them.

Chapter 2 describes this as predictable rather than random, which is what makes it interruptible.
Their flaws stop registering. Not minimised deliberately — filtered out before you see them. They become exceptional, nearly perfect, the answer. The relief is real: the intensity has somewhere to go and the emptiness has someone in it. Investment follows: constant contact, analysing every message, thinking about them when apart.
No one can hold the role. Eventually they are busy, or set a boundary, or say something that contradicts the image, or simply fail to understand something. Nothing has gone wrong — they have been human in a way the idealization did not allow for.
The filter inverts. The same person, unchanged, becomes selfish, cruel, someone who never cared. This reading feels exactly as true as the first one did, which is the part that makes it so hard to argue with from inside.
The most useful thing to understand about this cycle is what is not moving.
They changed — they were kind and now they are not
They stayed the same. The filter selecting what you notice inverted.
You finally saw them clearly
Both phases are selective. Neither view included the whole person.
The idealization was the mistake
The idealization was the setup. The crash was built into it from the start.
This is about them
It is a regulation strategy. Someone was assigned a job no person can do.
Earlier is dramatically easier. By devaluation there is little available except delay.
The counterintuitive intervention. Naming something mundanely imperfect about them while you feel they are perfect keeps the whole person in view and softens the eventual inversion.
Contact frequency, plans, other friendships. The steeper the investment, the harder the reality check lands.
“They were busy” rather than “they are pulling away.” This is the single highest-value moment in the entire cycle.
How many times has this specific signal meant what you are about to conclude? The honest answer is usually rarely.
Send nothing, end nothing, decide nothing. The view will move again. Nothing decided in this phase survives contact with the next one.
Chapter 26's material. The relationship is often recoverable if the devaluation phase did not produce something unrecoverable.
From outside, the swing between adoration and contempt reads as calculated, and it is one of the main reasons people with a BPD diagnosis are described as manipulative. Chapter 2 is clear that this is a misreading of the mechanism.
The filtering is not deliberate and is not experienced as filtering. During idealization the flaws are genuinely not visible; during devaluation the warmth is genuinely not retrievable. Nobody is switching between two known versions of a person for effect — each version arrives as simple perception, complete and convincing.
That matters practically for both sides. If you are living this, the intervention is not trying harder to see them accurately in the moment, because in-the-moment accuracy is exactly what is unavailable. It is noticing which phase you are in and refusing to act irreversibly from inside it.
And if you are the person on the receiving end, understanding the mechanism does not oblige you to absorb its effects — but it does change what you are responding to, which usually lowers the temperature considerably.
Both versions of the person feel like accurate perception. Neither is chosen.