← ClaudeAtlas

tufte-data-mapslisted

Use when designing or auditing a map that carries data and something is off — a plan view spending both axes on geography the question does not need, poster-thin density that collapses on close inspection, ornament filling empty regions, one rigid scale forcing a choice between detail and overview, aggregation to political areas distorting the signal, or map small multiples restyled per panel.
jpoindexter/tufte-skills · ★ 0 · AI & Automation · score 70
Install: claude install-skill jpoindexter/tufte-skills
# Tufte: Data Maps as the Standard of Information Design ## Overview In *Envisioning Information* Tufte does not treat the map as one chart type among many; he treats high-quality maps as the **gold standard of all information design** — bountiful detail, several layers of close reading combined with an overview, rigorous survey data. A good map is the most demanding escape from "flatland": it increases (1) the number of dimensions shown on a plane surface and (2) the data density per unit area, and it rewards both a distant macro glance and close micro inspection. The chronic failure is the **"duck"** — pretend dimensions, ornament, and posterized thinness substituted for data. **Scope note.** This skill covers Tufte's map argument in *Envisioning Information*, chiefly ch. 1–2 (Escaping Flatland, Micro/Macro Readings). *EI* does carry its own choropleth critique: pp. 40–41 fault the shaded "blot map" on three grounds — non-uniform area sizes, color filling often-nearly-empty land so large unpopulated regions get the greatest visual emphasis, and historical boundary changes breaking statistical comparison — and present the fix, mesh maps built on Japan's division into 379,000 equal-sized cells (developed in `tufte-micro-macro-readings`; cite EI pp. 40–41 for it). What remains *VDQI* territory: the fuller "area is not data" patch-map treatment and classification schemes. Modern conventions (Jenks, Moran's I, ColorBrewer) are out of scope for both sources. --- ## §1. Maps