Field notes · July 2026
What the notebook noticed
Plover began as one runner, one folder of .fit files, and a notebook full of
plots. The goal was modest: look at my own Coros exports and see what a season
had actually done to my running.
Weekly mileage was the easy chart, and the least interesting — every app already draws it. The plot that stopped me was smaller. At the same easy pace, over about six weeks, my stride length was quietly shrinking, and the time each foot spent on the ground was creeping up. Nothing in my splits looked wrong. No single run stood out. But the mechanics were drifting, run after run, in the direction fatigue points.
Illustrative example — not your data. The band is your usual range at this pace.
That is the finding conventional tools miss. A readiness score would have shown green. A charts platform would have buried it on page four. The signal was real, it was early, and it was mechanical — exactly the kind of thing you want a quiet observer to point at before it becomes a story you tell your physio.
Two rules came out of that notebook and never left. First, the headline is the plain-English change — “your steps are getting shorter on easy runs” — not the metric name; the number is the evidence, not the point. Second, a finding only earns attention once it clears a real threshold: matched-pace ground-contact drift starts at 3% between the first and last thirds of a run, stride fade at a 2% drop across a set of reps. That is what keeps the feed short and the quiet weeks quiet.
Everything else in Plover is downstream of that afternoon — the detectors, the evidence grades, the honest gaps. The methods paper has the full math. This is just where it came from.