tufte-viz

name	tufte-viz
description	Ideate and critique data visualizations using Edward Tufte's principles. Use when designing charts, improving dashboards, checking graphical integrity, reducing chartjunk, or planning small multiples and high-density displays.
version	1
user-invocable	true

/tufte-viz - Tufte Data Visualization Review

Apply Edward Tufte's principles to design clear, honest, high-density data visualizations. For general layout and UI critique, use /visual-critique instead.

Quick Start

/tufte-viz                                    → Guided questions, then describe or attach the viz
/tufte-viz [attach image]                     → Critique a chart, dashboard, or report screenshot
/tufte-viz [paste spec or describe dashboard] → Review viz decisions in a spec or flow

Output: Review saved to outputs/critique-notes/[feature]-tufte-viz-[date].md Time: 10–20 minutes depending on scope Dependencies: None. Works from images, specs, or descriptions. Figma MCP helps if connected.

---

Context Routing (Internal)

Before ideating or critiquing, check:

Source	Files/Folders	What to extract
Tufte reference	context-library/foundations/tufte-principles.md	Lie factor, data-ink ratio, chartjunk, small multiples, Tufte test
Semantic color	context-library/foundations/semantic-color.md	Color encoding vs decoration in charts
WCAG	context-library/foundations/wcag.md	Contrast, alt text for charts, color independence
Design principles	context-library/design-system/principles.md	Team rules that may constrain chart treatment
Storybook	context-library/design-system/storybook.md	Existing chart components if documented
Brief or spec	outputs/briefs/ or described in chat	What insight the viz must support

Read context-library/foundations/tufte-principles.md for the full reference before applying the frameworks below.

---

Guided Questions

Ask these if the designer hasn't given context:

1. "What am I looking at? (Single chart, dashboard, report page, in-product analytics)"
2. "What's the one comparison or insight this needs to support?"
3. "Who's the audience? (Exec summary, analyst, end user in workflow)"
4. "Is this new design or a critique of something existing?"
5. "Any chart library or component constraints? (e.g., Recharts, design system chart tokens)"

---

Workflow: New Visualizations

1. Clarify the data story

• What comparisons matter?
• What's the key insight to communicate?
• Who's the audience?

2. Select approach (Tufte-aligned)

• High comparison need → Small multiples
• Dense data → Data tables, sparklines, condensed time-series
• Time-series → Line charts with minimal grid
• Part-to-whole → Prefer bar or table over pie chart

3. Design with data-ink in mind

• Start minimal; add only what's necessary
• Every element must earn its ink
• Default to grayscale; use color purposefully (check semantic-color.md)

4. Apply the Tufte test

Run all seven checks from tufte-principles.md Quick Reference before handoff.

---

Workflow: Critiquing Visualizations

Work through each area. Only include findings where there's something worth saying.

1. Graphical integrity

• Does visual effect match data effect? (Lie Factor ≈ 1.0; flag if > 1.05 or < 0.95)
• Are baselines and scales honest? (No truncated axes that exaggerate change)
• Any 3D distortion on 2D data?
• Is context provided? (Sources, units, time range, definitions)
• Are intervals consistent?

2. Chartjunk

• Decorative elements, heavy grids, gradients, shadows without purpose
• Moiré patterns, busy cross-hatching
• "Duck" graphics where design competes with data
• Clip art or icons that don't encode data

3. Data-ink ratio

• Apply the eraser test: what can be removed without losing information?
• Redundant labels, borders, boxes, excessive tick marks
• 3D when 2D suffices

4. Excellence, comparison, and density

• Can viewers compare data elements easily?
• Multiple levels of detail (overview + fine structure)?
• Could more data fit in the same space without hurting readability?
• Would small multiples serve comparison better than one busy chart?

5. Color and accessibility

• Color used for encoding, not decoration (cross-check semantic-color.md)
• Colorblind-safe palettes; gray default, color for emphasis
• Not relying on color alone for meaning (cross-check wcag.md)

6. Suggest improvements

For each issue: specific before/after recommendation tied to a Tufte principle.

---

Quick Checklist

• Lie Factor ≈ 1.0 (no visual distortion)
• Maximum data-ink ratio
• Zero chartjunk
• Clear labeling (units, sources, scales)
• Enables comparison
• Reveals multiple levels of detail
• Appropriate data density

---

Output Structure

# Tufte Viz Review: [Chart, Dashboard, or Feature Name]

**Date:** [date]
**Source:** [image / spec / Figma link / description]
**Mode:** [new design / critique]

---

## Summary

[2–3 sentences: overall viz health. Is the data honest, clear, and dense, or buried in decoration?]

---

## Data Story

[What comparison or insight this viz supports, and whether the design serves it.]

---

## Tufte Test Results

| # | Check | Pass / Flag | Notes |
|---|-------|-------------|-------|
| 1 | Data-ink | | |
| 2 | Integrity (Lie Factor) | | |
| 3 | Chartjunk | | |
| 4 | Excellence (multi-level) | | |
| 5 | Comparison | | |
| 6 | Density | | |
| 7 | Context | | |

---

## Findings

| # | Issue | Location | Principle | Recommendation |
|---|-------|----------|-----------|----------------|
| 1 | [what's wrong] | [where] | [e.g., chartjunk, integrity] | [specific fix] |

---

## What's Working Well

[Specific decisions to preserve.]

---

## Suggested Next Steps

[1–3 actions. e.g., "Remove heavy grid," "Try small multiples for segment comparison," "Run /color-review on chart palette."]

---

Quality Check Before Saving

• foundations/tufte-principles.md read and referenced
• Tufte test (all 7 checks) completed
• Integrity and lie factor considered for proportional encodings
• Chartjunk and data-ink pass done
• Color checked against semantic-color.md where color is used
• Accessibility flagged via wcag.md where relevant
• Recommendations are specific (before/after), not generic "simplify"
• Scope stayed on data viz (did not duplicate full /visual-critique)