A lazy day is just packet loss. So my step goal backs off like TCP.

Описание и применение

Приложение Slow Start использует принципы TCP‑контроля перегрузки для динамического расчёта ежедневных целей шагов, повышая их достижимость и удержание пользователей.

Задача

Users struggle to set realistic daily step goals and often abandon fitness apps when goals are unattainable.

Результат

On held‑out days from 33 real users, the dynamic goal met 34.6% of days versus 0% for a fixed 10,000 goal. After tuning, the system achieved 71.1% success at 74% of a typical day, matching a simple 30th‑percentile rule.

Рабочий процесс

• 1. Import user step history or start from 1,500 steps.
• 2. Compute a "window" goal using TCP‑style rules: start small, double after a win, add 1,500 once near level, halve after a miss.
• 3. Compute a "cap" using TabPFN v2’s 30th percentile prediction of tomorrow’s steps.
• 4. Set the daily goal as the minimum of window and cap.
• 5. Log steps; if goal met, double goal next day; if missed, halve goal next day.
• 6. Update Slumbrock’s state (XP, speed, power) based on cumulative steps.
• 7. Provide feedback via Gemma 3 4B for check‑ins and recovery tips.

Доказательства

I reached 10,000 on none of them. Not one day, in a stretch where I logged 43 workouts in Hevy.

Источник

Slow Start sets your step goal like TCP: start small (1,500, or your own level if you import your history), double after a win, climb steadily near your level, halve after a miss, never to zero.

Источник

On held‑out days from 33 real people, a fixed 10,000 steps was met on 34.6% of days. My first controller hit 86.6%, but only by asking for half of what people usually walk. Tuned on a validation window, Slow Start meets 71.1% of days at 74% of a usual day, which ties a plain "your own 30th percentile" rule (69.7%).

Источник

TabPFN v2 (Apache 2.0) and Gemma 3 4B run on a server I control. Nothing goes to a closed AI service.

Источник