Capture and synthesis are different jobs.

When they get mixed together, the system becomes confused. Either capture becomes too heavy and you stop doing it, or synthesis becomes too shallow and everything gets called knowledge before it deserves the name.

Capture protects material from being lost.

Synthesis makes material useful.

You need both. You should not pretend they are the same.

Capture should be fast

Capture is the inbox function. Its job is to collect potentially useful material with enough context that future-you can understand why it mattered.

A good capture item answers:

  • What is this?
  • Where did it come from?
  • Why might it matter?
  • What topic or draft might it relate to?

That is enough.

Capture should not require a full essay, perfect tags, or a permanent folder location. If the threshold is too high, useful material never enters the system. If the threshold is too low, the inbox floods with garbage. The right standard is simple: capture enough to evaluate later.

Synthesis should be slower

Synthesis is where judgment enters.

It asks what the material means, what pattern it supports, what it contradicts, and what should happen next. It may update a topic page. It may create an evidence pack. It may merge with an existing theme. It may become a draft. It may be discarded.

Synthesis is not filing. It is interpretation.

This is why AI can help but should not fully own the job. An assistant can cluster notes, identify repeated themes, compare sources, summarize arguments, and propose draft angles. But the operator still has to decide what is true enough, useful enough, and important enough to promote.

The handoff matters

The most important part of the workflow is the handoff between capture and synthesis.

If raw material flows directly into permanent pages, the knowledge base gets noisy. If raw material stays forever in the inbox, nothing compounds. The system needs promotion logic.

A captured item should move forward when it meets one of these tests:

  • it appears in multiple sources;
  • it strengthens an existing topic;
  • it provides a concrete example;
  • it exposes a tension worth exploring;
  • it supports a draft already in progress;
  • it creates a new open question worth tracking.

Otherwise it can remain raw or be ignored.

The practical split

This is where this post differs from the broader “notes are not knowledge” argument. The issue here is not whether notes count as knowledge. The issue is operational ownership: capture owns intake, synthesis owns judgment, and promotion is the explicit handoff between them.

Use separate spaces for separate jobs.

The inbox is for capture.

Topic pages are for synthesized knowledge.

Evidence packs are for source-backed writing claims.

Draft queues are for publishable candidates.

Published posts are processed outputs that can become sources again.

When these spaces blur, the system loses leverage. You start drafting from messy notes, researching from old drafts, publishing from weak claims, and calling it a workflow.

The fix is not more complexity. It is cleaner boundaries.

Capture everything worth evaluating. Synthesize only what deserves to compound.

Source note

Draft informed by the 2026-05-05 Publishing & Knowledge Systems evidence pack and related vault notes on Publishing Pipelines, AI-Native Publishing Systems, Readwise Digest System, Profile Generation Pipelines, and the compiled knowledge layer.