What counts as a crash.

Every finding on this site answers a question of the form "what happens around a crash." For the answers to mean anything, "a crash" needs one fixed definition: chosen and locked before any test ran, so it couldn't be quietly bent to fit a result. Here it is.


A crash is a run of bad days, not a single one

The felt-state score runs 1 to 10, but in four years it never once rose above 6. Against a lived ceiling of 6, a 3 is not a mild day. See what a 3 actually means, in lived terms. A crash is a run of two or more consecutive days scoring 3 or below, with separate runs within three days of each other merged into one episode. By that rule, four years hold 29 crashes.

crash line: 3 or below one crash episode a dip
Illustrative: a teaching diagram, not real dated scores (those stay private). Taller bars are better days; the warm, short bars are the bad ones.

Two tiers: crashes and dips

A later refinement keeps the crash definition unchanged and adds a second tier for the bad days the first rule misses:

Why a score, and not a "PEM" tag?

There's no clean clinical tag in this record that marks a post-exertional crash: nothing applied the same way, every day, for four years. What there is, is the score: one number, typed by feel, that blends the physical, the cognitive, and the emotional into a single read of the day. It isn't arbitrary: each level maps to real, describable differences in what the body could do, set down in what a 3 actually means. And it has the one property a crash definition can't do without: it's logged every single day, an unbroken line across the whole record. So the crash is built on the score: the complete, reasoned signal we actually have. The scattered symptom notes that do exist are presence-only, kept for describing crashes, not defining them.

Full definition, locked and dated, in the research repo: crash_v2 definition.md ↗

Open full page ↗