Palace/Workshop

Learning By Writing

learningwritingstrategyinvestigationrationalityhypotheses

Learning By Writing

A method for turning investigation into retained, opinionated understanding by organizing learning around writing rather than reading. Reading and discussion always serve the formation, articulation, critique, and revision of an explicit bottom-line view.

Core Loop

  1. Pick a topic — claims that are important if true and might be true.
  2. Read/discuss the most prominent pieces on each side (lightly).
  3. Write (or speak) a defense of your current, very premature hypothesis — bold and aggressive.
  4. List weaknesses and gaps in your own case.
  5. Select the sub-question most likely to flip the headline claim; investigate narrowly.
  6. Revise the claim or switch sides.
  7. Repeat the write-critique-revise cycle many times.
  8. Incorporate external feedback and continue.

The decisive practice is always having (and re-articulating) a hypothesis. This keeps every round of reading focused on changing or supporting a bottom-line view instead of diffuse information gathering.

Why the Method Works

Writing the hypothesis early surfaces the exact version of the claim you care about and reveals which sub-investigations have leverage. Without it, multi-dimensional topics cause endless low-value reading.

Premature hypotheses are expected to fail; the point is rapid iteration through failure. Actively arguing the best case for the opposing side often reveals more than trying to be neutral from the start.

Sub-question selection is the meta-skill: stay narrowly on what can actually alter the headline rather than pursuing exhaustive coverage of every angle.

Relation to Other Ideas

This operationalizes several of the heuristics that humans do not apply automatically (see humans-are-not-automatically-strategic): explicit goal definition, progress tracking via written claims, systematic conjecture testing, focus on high-leverage methods, and deliberate revision.

It contrasts with passive "read everything first" models and complements frameworks that emphasize sampling and match quality in learning.

Sources