Open work past the guide
The field guide holds her patterns against my data. Once I was inside four years of my own watch, other questions surfaced that her guide never set out to answer. This is the complete state of play on those: everything I raised past her guide, sorted by how settled it is and what stage it sits at. It includes the findings that graduated to their own pages, the observations I can only describe, and the open questions still on the list. Showing the work in progress is the point.
This is the full list behind beyond the guide, and the twin of the guide, statement by statement: that ledger holds her patterns against my data; this one holds the questions my data raised past hers.
How to read the labels
Two things about each item: what kind of work it is, and what stage it's at. A described shape and a tested prediction are different kinds, held to different bars; a characterised description is a finished result, not a lesser not-yet-tested one. The label on each item is those two together.
- characterized
- A shape I could characterize; no pass/fail bar by design. A finished result, not a lesser one.
- scoped, a description
- A characterization I have designed but not yet run.
- first tested
- A prediction held to a pre-registered bar; a verdict on the linked finding.
- scoped, a test
- A test designed and pre-registered, not yet run.
- inconclusive
- A test that ran but could not resolve either way.
- open idea
- Noted and worth doing, not yet designed or locked.
- held back
- Deliberately held, usually awaiting a proper pre-registration.
Things I could describe
Shapes I characterized without a pass/fail bar. A characterized item is a finished result, never a not-yet-tested one.
-
After the crash: a recovery signature the feeling doesn't carry
characterizedAfter a crash my body reads different for about two weeks even as the feeling comes back in a couple of days: a real recovery signature the feeling itself doesn't carry.
-
When the strain isn't physical: what the watch catches, and what it misses
characterizedSorting my days by what the demand was: the watch catches my emotional load (in stress and battery, never the heartbeat) and stays blind to cognitive load. A refinement of her mental-activity line, not a new claim.
-
The changing kind of crash
characterizedThe kind of crash seems to have changed: the early ones ran loud (heart rate up), the recent ones quiet (heart rate down). An eyeballed pattern, described not tested, and tangled with a medication change.
-
Does emotional load help trigger a crash?
held backA first descriptive look hints emotional load rises in the days before a crash, but it is suggestive only and sits in the medication-shifted years. Held back for a proper pre-registered test, not claimed here.
-
Overnight stress runs higher on crash nights
characterizedOvernight Garmin stress runs clearly higher on my crash nights: a solid descriptive gap I never turned into a formal test. A freestanding read; her guide's night-stress line makes a related claim.
Things I held to a test
Predictions held to a pre-registered bar, plus the ones still designed or awaiting a test. A stage is not a verdict.
-
Recovery debt: does a crash stack the odds of the next?
first testedThe feared snowball, each crash stacking the odds of the next, didn't show up when I tested it. No measurable debt.
-
The danger window: is exertion riskier right after a crash?
inconclusiveIs exertion riskier in the felt-recovered-but-still-unsettled window after a crash? I pre-registered and ran it; the data couldn't resolve it either way.
-
Stress while I am sitting still
inconclusiveMy own refinement of her stuck-stress idea, with a filter that throws out the minutes I was moving. The clean sample fell one crash short of the floor I set in advance, so it landed inconclusive: not a yes, not a no.
-
The character-flip, tested properly
open ideaThe changing-kind-of-crash pattern above, tested properly: early-versus-late crashes read channel by channel, guarding the medication change. Designed as an idea, not yet pre-registered or run.
-
Is 'best in the middle' really an activity map?
open ideaAnother of my extensions: does the 'best in the middle' stress curve survive holding activity constant, or is it partly a map of how much I did? Noted as an idea, not yet run.
-
Too much in the days before a crash
first testedHeavy exertion in the roughly four days before a crash: present but weak, below the bar I set, and awaiting a cleaner single-pool recompute. The lived-experience cousin of the scorecard's exertion lead-up, kept as its own thread.
-
Body-battery floor before a crash
held backDoes end-of-day body-battery dropping below my personal floor predict a crash the next day? Designed, but blocked: it needs per-minute body-battery this watch doesn't export cleanly.
-
Late-afternoon body-battery drain before a crash
held backDoes a steeper late-afternoon body-battery drain than my morning envelope predict a next-day crash? Same blocker as the floor version: it needs per-minute body-battery.
-
Stress that won't come down, any time of day
held backStress that stays high at rest at any hour, not just after exertion, and worse in the evening: my broader-than-the-guide extension, blocked on a stress-with-motion-by-time-of-day signal I haven't built yet.
This list is provisional and it moves. An idea becomes a scoped design, a scoped test runs and lands (or can't resolve), a parked lead waits on a pre-registration or on data this watch doesn't give up easily. That's the honest state of it, kept in the open the same way the closed-hypothesis ledger keeps what's already settled. Where a thread has grown into a finding, it links out to it.