Snowball / Flywheel

A hypothesis that accumulated success or decline is making each next turn easier in the same direction.

A year ago you could barely finish an article on machine learning; now you read papers on the train for fun. Or the reverse: you skipped two French lessons in March and somehow have not opened the app since, each missed week making the next one feel more hopeless. Both stories share one suspected structure, running in opposite directions.

Snowball / Flywheel is the hypothesis that accumulated success or decline is making each next turn easier in the same direction: skill makes reading easier, easier reading builds skill; missed practice erodes confidence, eroded confidence blocks practice. To use it as a diagnostic rather than a flattering story, make it predict. If compounding is real, your effort per unit of progress should be visibly falling over time (or rising, in decline), and interrupting the loop briefly should show momentum, not instant reset. State the strongest rival: maybe an outside factor, a new job using these skills daily, is driving everything and there is no self-feeding at all. Find the separating observation, such as tracking comprehension across a month with no job exposure. If confirmed, protect the loop's early turns and its stored gains; if in decline, the priority is any small win that restarts accumulation. Do not declare a flywheel from two good weeks; noise compounds nothing.