Divergent predictions

Before looking at the result, lock in what each rival explanation predicts you will see.

Your 10K times have plateaued for two months, and you have theories: you're overtraining, your shoes are dead, or your easy runs are too fast. You're about to take a recovery week, and you can already feel how this will go. Whatever your next race time is, you'll find a way to read it as supporting whichever theory you woke up favoring.

Divergent predictions is the move of locking in, before you look at the result, what each rival explanation predicts you will see on the same observable. The cue is that a result is coming, a race, a report, a reply, and multiple explanations are still live. The operation: pick the observable (next 10K time after a recovery week), then write each rival's prediction next to it. Overtraining predicts a clear improvement after rest. Dead shoes predict no change. Pacing predicts no change from rest alone. Freeze the list where you can't quietly revise it. What changes is that the result becomes an arbiter instead of a Rorschach test, because you committed before knowing. Skip the freeze when only one explanation is in play, since there's nothing to discriminate between yet.