AI by Hand ✍️

AI by Hand ✍️

Compaction

Summarizing history to make room

Prof. Tom Yeh's avatar
Prof. Tom Yeh
Aug 19, 2026
∙ Paid

Library › Context Problems

  1. The Context Window

  2. Window Occupancy

  3. Growth per Turn

  4. Remaining Space

  5. Turn Budget

  6. Retrieval Footprint

  7. Retrieval per Turn

  8. Search Rounds

  9. Window Sizing

  10. Truncation

  11. Compaction

  12. The Compaction Bill

  13. Compacted Retrieval

  14. The Compaction Threshold

  15. Verbatim Tail

  16. Stateless API

  17. Retrieval Cost

  18. Context Tax

  19. Quadratic Cost

  20. Cost Forecast

Truncation throws the oldest turns away. Compaction keeps them, but small: the model reads the history and writes a summary that stands in for it. The system prompt is not part of the history and does not get compacted; it is pinned at the front and survives intact. So the window after a compaction is the same system prompt, followed by a summary a fraction of the size, and a great deal of space that was not there before.

Paid members: the worksheet and its printable PDF are below ↓

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2026 Tom Yeh · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture