experiments
Book recommender: library audit
Audit of 40 unread and in-progress books with status labels, reading patterns, and calibration data for the recommender.
🌱
🌿
🪴
🌳
Labeled 40 books (10 physical, 30 e-books) across status, type, purpose, and “why stuck”. Built an interactive HTML labeling interface as part of the book-recommender project.
Key patterns
- Ereader invisibility: multiple books I didn’t know were on my ereader. Discoverability is a real problem.
- “Should read” guilt: some topics feel like a chore even when I think I should read about them.
- Density without momentum: interesting topics but textbook feel kills the urge.
- Lost track: started books that just faded away.
- Old sci-fi bias: “masculine sci-fi” doesn’t appeal, even when the topic is relevant.
What works
- Pratchett: loved it, every time. Language + humor + philosophy + layers.
- Art of Co-design: started and enjoying it, high contender.
- Earth Emotions: remembered being beautiful but didn’t stick.
- Some books work as “on the side” reads (Complete Robot, Typography Beyond Borders, Microcopy).
Prototype
Interactive HTML labeling tool: book-recommender/library-audit.html. Potential future use as a library management UI on the garden.
Features:
- 40 books grouped by theme
- Status, type, and purpose buttons per book
- “How did you like it?” review section for finished books
- Progress bar and filter (all/labeled/unlabeled)
- Auto-saves to localStorage
- Export to JSON