A knowledge base should support fluid resolution-switching

Type: note · Status: seedling

Good thinking is not staying at one level of abstraction — it is constantly moving between levels. Start broad to see the landscape, narrow in when something is interesting, zoom back out to check bearings, dive deep into the load-bearing detail, then abstract up to see the pattern. A knowledge base that supports good thinking must support this motion fluidly.

The commonplace KB already has several mechanisms that serve resolution-switching, but they aren't usually described under that framing:

Titles vs bodies are a resolution pair. Claim titles give the zoomed-out view — the principle, the assertion. The note body gives the zoomed-in view — the mechanism, the evidence, the specifics. Scanning a list of titles is surveying the landscape; opening a note is examining the territory. Since title as claim enables traversal as reasoning, following links between claim titles reads as a chain of reasoning at the abstract level — without requiring the reader to descend into any specific note.

Indexes and notes operate at different resolutions. An area index like kb-design.md is the broad view — it shows what topics exist and how they relate. Following a link from the index zooms in. Returning to the index zooms out to check bearings. This is the two kinds of navigation distinction: local link-following is narrow and contextual; search and index browsing are broad and orienting.

Link semantics encode zoom direction. "Since [X]" zooms into a foundation — following it takes you deeper, toward the grounds of the current argument. "This extends [Y]" zooms out toward a generalization. "Contradicts [Z]" shifts laterally to a competing view at the same level. The relationship words in link strength aren't just categorization — they tell the reader which direction in abstraction space they're moving.

Progressive disclosure is a resolution gradient. The context loading strategy layers information from always-loaded (CLAUDE.md — broadest, least specific) through on-demand descriptions (medium) to full note bodies (narrowest, most specific). An agent traversing this hierarchy is adjusting resolution.

The evaluative criterion

This framing suggests a quality criterion for knowledge bases that complements retrieval accuracy: resolution-switching fluidity. A good KB lets you move between abstraction levels with low friction. A bad one traps you — either stuck in abstractions (indexes that link to indexes, never reaching specifics) or stuck in details (dense notes with no outward links to broader context).

Concrete symptoms of poor resolution-switching: - Notes with no outbound links — you can zoom in but can't zoom back out - Indexes with bare links (no context phrases) — the broad view has no resolution; everything looks the same - Topic-titled notes — titles don't carry the abstract-level argument, so you must open every note to learn anything - Missing relationship articulation — links exist but don't tell you which direction you're moving

Connection to discovery

The discovery note describes three depths of abstraction in connection: shared feature, shared structure, and shared generative model. Resolution-switching is the navigation skill that makes discovery possible — you can only see the particular as an instance of the general if you can move between the two levels. A KB that traps you at one level suppresses discovery.

Open questions

  • Can resolution-switching fluidity be measured? Candidate signal: for a random note, how many clicks to reach an index (zoom out) and how many from an index to reach a specific mechanism (zoom in)?
  • Does the KB have resolution dead-ends — areas where you can zoom in but not out, or vice versa?
  • Is there a sweet spot for note granularity that maximises resolution-switching? Too fine-grained and zooming out requires too many hops; too coarse and zooming in means reading irrelevant material.

Relevant Notes:

Source: - Adapted from a social media post on "The Art of Good Thinking: Moving Between Levels" — the core insight about resolution-switching applied to KB design

Topics: