My/Dotfiles/Process/Justifying Dependencies via Spaced Repetition/Idea
Prune what I can't explain: spaced repetition as a retention gate
The idea
- Give every dotfiles-managed dependency a flashcard: why it's installed, what job it does, which My/Principle or My/Pref it serves. Review the deck on a spaced-repetition schedule, the same mechanism already running Logseq/Flashcard/Review/Favorite decks for tools like mise, chezmoi, tmux, and yazi.
- If I can't answer a card — I've forgotten why something is there, or what principle it's supposed to serve — that entry gets flagged, not pruned right away. If I still can't explain it after a check-in period, it gets pruned from the dotfiles.
- I kinda like this because it makes me responsible for knowing what's in my dotfiles. Anything I don't know about eventually gets pruned, just by the passage of time and my own forgetting.
Why it fits here
Open questions
- What counts as "explained" — remembering the answer on my own, or is recognizing it on the card enough?
- Per-tool or per-principle: if I forget one small library dependency of a tool I do use and understand, would that wrongly get pruned?
- Schedule and grace period — how many missed reviews before an entry is flagged, and how long flagged before it's actually pruned?
- Scope for a first pass — Homebrew entries only, tying into My/Dotfiles/Process/Adding a New Global Dependency/Idea's Try-before-commit window, or every dotfiles-managed tool including mise?
- Who does the pruning — a weekly check I run myself, or a skill/agent that suggests removals from failed reviews for me to confirm?
- See 2026-07-17 Fri.