Use Shneiderman's overview-first protocol for graph exploration — overview, zoom-and-filter, then details-on-demand

Why This Is a Rule

Ben Shneiderman's Visual Information Seeking Mantra — "Overview first, zoom and filter, then details-on-demand" — was developed for data visualization but applies directly to knowledge graph exploration. The three phases exploit how human visual processing works: the overview engages pre-attentive processing (detecting clusters, gaps, and density patterns without reading), zoom-and-filter engages focused attention on regions of interest, and details-on-demand retrieves specific content only when you've identified what's worth reading.

Most people skip straight to details — opening a specific note and reading it. This is like reading a map by examining individual street names. You miss the topology: where the dense areas are, where the structural holes sit, which clusters bridge to each other. The overview phase reveals structural properties that no amount of individual note-reading would surface.

The protocol also prevents premature commitment. Reading a specific note's content creates anchoring effects that bias your subsequent exploration. Seeing the shape first, before reading labels, lets your visual system detect patterns that your reading mind would filter through existing expectations.

When This Fires

• When opening your graph visualization tool for any exploratory session
• During periodic graph reviews when assessing structural health
• When looking for patterns, gaps, or unexpected clusters
• Before diving into any specific region of your graph — always start with overview

Common Failure Mode

Jumping to details immediately: opening a graph view and clicking on the first interesting-looking node to read its content. You've now anchored on that node's perspective and will explore outward from it rather than from the topology. The overview phase must happen first and without reading labels — pure shape, density, and connectivity.

The Protocol

(1) Overview: Open your full graph visualization. Don't read labels. Look at shape: Where are the dense clusters? Where are the sparse regions? Where are the isolated nodes? Where are the bridges between clusters? Spend 60 seconds just seeing topology. (2) Zoom and filter: Select a region that interests you — maybe a dense cluster, maybe a surprising bridge, maybe an isolated node. Zoom in. Now read labels but not full content. What themes emerge? (3) Details-on-demand: Only now open individual notes to read content. You've arrived here through structural navigation, not random access — the notes you're reading are structurally interesting, not just topically familiar.