Local disk sweep: design / data-visualization readings

Sweep date: 2026-09-26

Scope (read-only): workspace; C:\Users\mjmon\Documents; Downloads; OneDrive; Zotero; Research Streams Literature / lit folders.

Missing roots (do not exist on this machine): Desktop, Books, Library, Kindle content paths, Calibre Library, Dropbox, iCloudDrive.

Method: recursive filename globs for target authors/titles; Zotero SQLite title search (0 matching items); home-directory EPUB/MOBI scan for target names (0 hits); Read-tool / extracted-text coverage of every candidate PDF/HTML/MD below.

Legal note: No pirate mirrors used; no copyrighted books downloaded. Only files already on disk.

---

Summary verdict

No complete commercial books from the Tufte / Cairo / Knaflic / Lupi–Posavec / McCandless / Few / Munzner / Bertin / Wilkinson / Cleveland target list were found as full PDF/EPUB on disk.

What was found: (1) the target Bostock–Ogievetsky–Heer 2011 D3 TVCG paper (complete); (2) related Heer/Bostock toolkit papers; (3) Tufte precursor / adjacent open or journal materials (1976–77 VDQI sketches, Citation Classic note, clinical-summary papers, 2020 open-source Truth Matters essay); (4) public D3 documentation / tutorial HTML extracts.

Duplicates: several papers exist both under Literature\ and under this design_readings\ folder.

---

Files found (candidates)

1. Bostock, Ogievetsky & Heer (2011) — D3: Data-Driven Documents

Design / system principles actually stated in the file:

• - Prefer representational transparency: bind data to the native DOM rather than hiding a toolkit-specific scenegraph.
• - Balance expressiveness, efficiency (notational effort), and accessibility (learnability); refine further into compatibility, debugging, and performance.
• - Tools should sit in an ecosystem of standards (HTML, CSS, SVG, JS); reuse prior knowledge and offload aesthetics to CSS.
• - Selections + operators: identify elements with predicates, then mutate them; values may be constants or functions of data.
• - Data join produces enter / update / exit so element lifecycle matches changing data; key functions preserve object constancy across transitions.
• - Prefer immediate evaluation of operators (easier debugging) over delayed property evaluation.
• - Prefer specifying transformations of scenes (what changes) over regenerating whole intermediate representations.
• - Animated transitions should interpolate attributes over time; delay/duration can be data-driven; use appropriate interpolators (including polar for arcs).
• - Optional helper modules encode Tufte’s quoted maxim: “Don’t get it original, get it right” — reuse proven layouts/scales rather than inventing for novelty.
• - Direct DOM access enables developer-tool inspection and console iteration; sticky data supports post-hoc restyling.
• - Performance claim grounded in benchmarks: D3 at least ~2× faster than Protovis on tested tasks; focus updates on changing attributes only.
• - Close cognitive mapping between toolkit primitives and desired graphics reduces Norman’s gulf of execution (cited regarding low-level libs; D3 aims for convenient abstractions without encapsulation costs).

---

2. Bostock & Heer (2009) — Protovis: A Graphical Toolkit for Visualization

Principles actually stated:

• - Evaluate tools on expressiveness (“Can I build it?”), efficiency (“How long?”), and accessibility (“Do I know how?”).
• - Chart typologies are quick but highly restrictive; unsupported types force compromise or tool-switching.
• - Low-level drawing is flexible but tedious / error-prone for complex, live, interactive viz.
• - Prefer a representation close to how designers think: hierarchy of graphical marks with properties as functions of data.
• - Inheritance / cascading of mark properties (CSS analogy) for concise specs.
• - Tool choice biases resulting designs (Maslow “hammer/nail” cited).
• - Cite Wilkinson and Tufte on limits of consumer chart typologies and questionable defaults.
• - Gap remains between high-level viz systems (foreign abstractions) and low-level graphics (too much work).

---

3. Heer & Bostock (2010) — Declarative Language Design for Interactive Visualization

Principles actually stated (from extracted text / sibling D3 framing that cites this work):

• - Declarative DSLs let users focus on domain specification while implementers optimize execution.
• - Decoupling specification from execution details supports interactive visualization language design (paper’s topic per title and D3 citations).

(Full principle list deferred relative to the 2011 D3 paper; treat as supporting toolkit literature.)

---

4. Heer, Card & Landay (2005) — prefuse: A Toolkit for Interactive Information Visualization

Principles actually stated (high-level from toolkit framing shared with later Heer work):

• - Provide composable operators for data management, visual encoding, and interaction (InfoVis Reference Model lineage as discussed in D3 related work citing this paper).

---

5. Heer & Shneiderman (2012) — Interactive Dynamics for Visual Analysis (CACM)

Principles / taxonomy items actually stated:

• - Map data attributes to position, size, shape, color to leverage perception.
• - Analysis is iterative: create, explore, refine views; tools must keep pace with human thought.
• - Confusing UI, hidden ops, or slow response curtail deliberation and introduce errors.
• - Taxonomy — Data & view specification: visualize, filter, sort, derive.
• - Taxonomy — View manipulation: select, navigate, coordinate, organize.
• - Taxonomy — Process & provenance: record, annotate, share, guide.
• - Formal grammars (Wilkinson Grammar of Graphics, ggplot2, Protovis) as flexible alternatives to fixed chart typologies.
• - Classic mantra cited: “Overview first, zoom and filter, then details-on-demand.”
• - Tools should preserve analysis history (undo/redo), support annotation, and enable collaboration/sharing.

---

6. Tufte (1976/77) — Improving Conventional Graphics for Statistical Data (“VDQI First Sketches”)

Design ideas actually stated in this file:

• - Opening warning: people can be lied to easily with charts if they treat “graphical” as automatically true (Rockefeller anecdote).
• - TOC themes as design agenda: forgetting the lying graphic; word-worth of one picture; less is more (communication, exploration, reconstruction); blot maps; area and effect; integration of written text and graphics; small multiples; friendly graphics; parallel plots; consciously 3-D graphics; aesthetics; case studies.
• - A graph can replace a long verbal news report (franked-mail example: ~700 words vs. time-series plot).
• - Critique of “lousy” component / stacked bar presentations (Schmid handbook example).
• - Scatterplot redesign notes: drop unnecessary “.0” trailing digits; cleaner scales.
• - Blot maps called out as problematic (“AND WHAT IS WRONG WITH BLOT MAPS”).
• - When encoding area (e.g., bomb yield circles), area ∝ magnitude must be explicit; captions should not violate figure principles.
• - Critiques of a Science admissions plot: missing size legend; caption violates figure principle; fitted line doesn’t fit; lettering quality.
• - Integrate text and graphics so they flow together; avoid divorcing numbered figures from prose; drop redundant pointer words like “below.”
• - Small multiples as a recurring structure (plant freezing distributions; NYT multi-panel economic charts).
• - Parallel time-series can work, but if the claim is a relationship (e.g., inflation vs. unemployment), a scatterplot (or differences) is more revealing than parallel strips.
• - Prefer a short sentence stating the main thrust of a graph to lead readers into detail (Business Week example praised).

---

7. Tufte (1992) — Citation Classic note on The Visual Display of Quantitative Information

Principles / claims actually stated:

• - Book analyzes excellence (esp. scientific work, cartography) vs. failure (esp. news graphics) since Playfair.
• - Practical theory: maximize data density and the multivariate / comparative quality of displays.
• - Book should be self-exemplifying: design of the artifact embodies its intellectual principles.
• - Integrate graphics into the text, sometimes mid-sentence — eliminate usual text/image separation.
• - Precise control of words/images, maximizing resolution, and pushing print technology are part of information-design scholarship.
• - Mentions subsequent Envisioning Information and forthcoming Visual Explanations (historical note only).

---

8. Powsner & Tufte (1997) — Summarizing Clinical Psychiatric Data

Principles actually stated:

• - Put the important facts on a single page while retaining diverse representation methods (numeric, narrative, diagrammatic, temporal, relational).
• - Combine small graphs, timelines, genograms, and recent notes so viewers can assess intervention–outcome relationships.
• - Use small multiples of clinical measures/doses over nonlinear time (last year large; prior years compressed context).
• - Align findings above interventions in columns to emphasize connection.
• - Rescale measures so vertical position ≈ clinical significance (normal / elevated / critical), making implicit clinical thresholds visually explicit.
• - Most recent value printed and plotted as the rightmost point.
• - High-resolution display that invites comparison across multiple coherent chronologies (generations → lifetime → clinic → recent weeks), not one jumbled chronology.
• - Cite Tufte VDQI / Envisioning / Visual Explanations page ranges as design basis for the small-multiple matrix.

---

9. Powsner & Tufte (1994) — Graphical Summary of Patient Status (Lancet)

---

10. Tufte (2020) — The Truth Matters: On the Relationship Between Evidence and Conclusions (open source)

Principles actually stated (evidence / analytical design):

• - Take anonymous statistical lives as seriously as named individual lives.
• - Standards of statistical reasoning about truth are universal.
• - Prefer rescue/prevention investments that save many invisible lives over only named rescues when costs differ.
• - Ask How do I know that? / How could anyone possibly know that?; Feynman’s law: you are the easiest person to fool.
• - Analytical thinking: reason about causality and mechanism, make comparisons, assess measurement credibility, validate, enforce integrity.
• - Observe data collection at the moment of measurement, not only at publication (Cuthbert Daniel).
• - Retain primary data and clear records; published figures must be traceable to unmanipulated primary data.
• - Track origins and life histories of measurements (who looks, consequences, who profits).
• - Database at hand may not contain the answers; enemies of truth include no theory, no relevant data, confirmation bias, conflicts, cheats.
• - False alarms can be the most prevalent (and lucrative) measurement error in healthcare.
• - Models need package-insert-style warnings about assumptions and adverse effects (mock “black box” for stepwise regression).
• - “You can observe a lot by just watching” — but you must know what to watch and keep watching (Daniel / Yogi Berra).

---

11. Public D3 documentation extract — What is D3?

Principles actually stated:

• - D3 is a low-level toolbox, not a charting library with a “chart” abstraction; compose primitives.
• - No default presentation — only the code you write (or copy).
• - Embrace web standards (SVG/Canvas/DOM) for CSS, debugging, imperative immediacy.
• - For bespoke, high-audience graphics (e.g., newsrooms), D3’s expressiveness pays; Amanda Cox paraphrased: use D3 if writing ~100 lines for a bar chart seems normal.
• - Overkill for private dashboards / one-off analysis — prefer higher-level tools (Observable Plot) when time-constrained.
• - Data join (enter/update/exit) exists so you control exactly what happens when data change; shines for dynamic viz.

---

12. Bostock bar-chart tutorial HTML parts

---

Non-matches / noise excluded

• - C:\Users\mjmon\OneDrive\...\tzdata\zoneinfo\Africa\Cairo — timezone file, not Alberto Cairo.
• - Zotero library: 0 items matching target author/title strings in zotero.sqlite.

---

Target titles NOT found on disk (as complete books or standalone essays)

Edward Tufte (full Graphics Press books)

• - The Visual Display of Quantitative Information (book)
• - Envisioning Information
• - Visual Explanations
• - Beautiful Evidence
• - The Cognitive Style of PowerPoint

Alberto Cairo

• - The Functional Art
• - The Truthful Art
• - How Charts Lie
• - The Art of Insight

Cole Nussbaumer Knaflic

• - Storytelling with Data
• - Storytelling with Data: Let’s Practice

Giorgia Lupi / Stefanie Posavec

• - Dear Data
• - Data Humanism (essay/book materials)

David McCandless

• - Information is Beautiful
• - Knowledge is Beautiful

Amanda Cox / NYT Upshot

• - No standalone Upshot essays, talk PDFs, or Cox-authored HTML collections found under searched roots (only a paraphrased Cox quote inside the public D3 docs extract above).

Related classics

• - Stephen Few — Show Me the Numbers; Now You See It
• - Tamara Munzner — Visualization Analysis and Design
• - Jacques Bertin — Semiology of Graphics
• - Leland Wilkinson — The Grammar of Graphics (book; cited by name in Heer papers, but book file absent)
• - William S. Cleveland — The Elements of Graphing Data; Visualizing Data

---

Practical takeaway for the LEC poster goody bag

On this machine, the legally available full target text for design craft is essentially the 2011 D3 paper plus Tufte-adjacent open/journal materials and Heer interaction/toolkit papers. The major named books in the brief are not present as files; physical copies (if owned) are not mirrored digitally in the searched folders.