Edit wiki/learning/spaced-repetition.md — Work Reflections

---
visibility: public-edit
---

# spaced repetition

the forgetting curve and how i use mochi to actually retain what i learn.

## the forgetting curve

ebbinghaus discovered in 1885 that you forget most of what you learn within 24-48 hours, following an exponential decay — but each review flattens the curve, and by the fourth or fifth spaced review the memory is basically permanent.

this is not controversial in cognitive science — meta-analyses of 250+ studies confirm that spaced practice produces 10-30% better retention than massed practice (cramming).

## why cramming feels effective

the cruel trick of memory: rereading material creates a feeling of familiarity that gets mistaken for learning. you read your notes, everything looks familiar, you think you know it. then you close the notes and can't reproduce any of it.

this connects to [[the-testing-effect]] — retrieval practice (actually pulling information from memory) is what builds durable memory, not re-exposure.

## how i use mochi

mochi is a spaced repetition app — basically anki but with better design. the system works by scheduling cards at increasing intervals based on how well you recall them.

what i make cards for:
- key concepts from [[research-workflow]] reading
- api patterns and syntax i use occasionally but not daily
- mental models and frameworks (the kind of thing from [[modeling]])
- definitions and distinctions that are easy to confuse

what doesn't work well as cards:
- anything too complex for a single recall prompt
- things i use daily anyway (no card needed — natural spacing)
- opinions or judgments — cards are for facts and frameworks, not takes

## the deeper principle

spaced repetition is an instance of a broader principle: learning happens at the edge of forgetting. you need to almost-forget something before reviewing it — that's what makes the review effortful and therefore effective.

## connection to building things

the best spaced repetition system isn't mochi — it's using what you learn in real projects. see [[building-to-learn]]. if i learn a concept and then use it in code the next week and again in a different project the following month, that's natural spaced repetition with the added benefit of contextual practice.

mochi fills the gap for things i need to remember but don't naturally encounter at spaced intervals.

## practical notes

- i do mochi reviews in the morning — it's low-energy work that gets the brain moving
- card quality matters more than card quantity. a bad card wastes time forever. see [[operation-optimization]] — small improvements to card writing pay compound dividends
- the habit of making cards is harder to maintain than the habit of reviewing them. mochi's scheduling handles reviews automatically; the bottleneck is creating good cards from new learning