Horizons

Define what success means now, later, and after the initial gain decays.

Cramming works. Next week's exam will go fine, your practice scores prove it. But this course feeds directly into next year's sequence and, eventually, your actual profession, and the method that wins next week is precisely the one that leaves nothing behind in six months. Your goal has one horizon; the result has three.

Horizons makes you define success separately across time: the immediate effect, the intermediate result, the long-term state, and what remains after the initial gain decays. The cue: a goal judged at a single moment when its value plays out over multiple timescales, or a method that trades later horizons for the nearest one. Run it in three steps. First, ask the move question: what does success mean now, later, and after the initial gain decays? Second, write a success line for each horizon: next-week performance, six-month retention on held-out questions, transfer to unfamiliar problems in the following course. Third, check your current method against all lines, cramming passes one and fails two, so blend in spaced retrieval. What changes: you stop unknowingly buying the near term with the far term. The spec's warning: derive horizons from how the result causally unfolds, not from a ritual one-week/one-year template.