agentleFS
jpoindexterjpoindexter/tufte-skills/skills/tufte-disinformation-design/SKILL.md

tufte-disinformation-design

Skill1 starsChanged 4 months agoMITView on GitHub

Use when auditing a chart, demo, interface, or vendor presentation for concealment tactics — masking, dazzle, decorative diversion, buried disclosures, presenter-dependent displays — when designing pictorial instructions that must show both the audience view and the hidden mechanism (flaps, cutaways, operator viewpoints), when verifying inherited or copied diagrams, or when structuring an honest technical reveal.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not compress the explanation of a hidden mechanism to match the brevity of its visible effect.

    line 46

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Disinformation Design: The Magician's Methods, Inverted

Source: Visual Explanations (VE), Edward R. Tufte, Graphics Press, 1997 — Ch. 3, "Explaining Magic: Pictorial Instructions and Disinformation Design," pp. 55–71, with supporting cases from pp. 35 and 74. All page numbers are the book's printed pages.

Overview

Stage magic is the most refined practice of information suppression ever developed: a magician designs an experience whose entire purpose is to prevent the audience from reconstructing what happened. Tufte mines that craft twice over. Read forward, the conjurer's methods are a working catalog of how presenters conceal — useful to anyone who must audit a display made by someone with something to gain. Read backward, the books that teach magic solve the hardest problem in pictorial instruction: showing a hidden, fast, deliberately disguised process to a reader who must perform it — useful to anyone who must explain a mechanism the eye cannot see.

This skill is the operational double catalog: detection tactics for consumers of displays, revelation tactics for designers of instructions.

Sibling coverage map (do not re-derive here):

Already covered inWhat lives there
tufte-narrative-and-sequence §4–§5Larger-motions-mask-smaller / visual masking; still-land vs. video-land; the 80% text/picture split; the full Automaton Chess Player case; the five-question integrity test; PGP and the six presentation rules
tufte-causal-reasoning-in-graphicsCYA notices / displays needing an oral escort; selective databases; the Challenger chart failures
tufte-evidence-corruptionPresentation-level corruption in Beautiful Evidence — cherry-picking, overreaching, punning evidence

What follows is the residue those skills do not carry.


§1. The Two Viewpoints — the Geometry of Revelation (pp. 55–58, 61)

A performed illusion has a front and a back. Knowledge of the revealed front view (what appears to be done) yields no reliable knowledge of the concealed back view (what is actually done) — and the audience's assumption that it does is the magic (p. 55). Every depiction of a process silently chooses among three options: show the produced effect, show how to produce it, or show both — and by means of what angles (p. 61). Making that choice consciously, before drawing, is the first act of honest instruction design.

DeviceWhat it doesVE case
Hinged flapAudience view on top, workings beneath, in the same footprintTufte's redrawing of a trick documented since 1581 (p. 57)
Layered depictionStacks revealed and concealed states as annotated strataCards-in-Envelope, ten layers (p. 56)
Cutaway / impossible viewpointShows what no camera position could captureMitral-valve surgery diagram (p. 57)
Rotation to operator viewFlips the scene 180° when the reader must perform, not watchStrike-second-deal drawn from the card-handler's side (p. 58)
Double-functioning outlineOne drawn line serves two depicted layers at onceSolid edge reading as both top card and the card beneath (p. 58)

The viewpoint question: who is this reader — spectator, student, or operator? A spectator needs the effect; a student needs both views in one figure; an operator needs the scene rotated into their own hands. Drawing the wrong viewpoint for the audience is a category error no rendering quality can fix.

§2. Verbs Are Expensive — Budgeting Description Density (p. 58, fn. 6)

The strike second deal is over in a fraction of a second, yet Hugard and Braué need 2.6 pages of text and illustrations to teach it; the record explanation of a single sleight — a move completed in under two seconds — runs 29 pages with 14 photographs and 14 drawings. The general law: explanation length scales with the speed and hiddenness of the action, not with its duration. Fast, concealed, or disguised steps deserve the most depiction, precisely because they are the steps the eye never caught.

  • Do not compress the explanation of a hidden mechanism to match the brevity of its visible effect.
  • Arrows can conduct tempo, not just direction: the coin-exchange sequence (p. 60) depicts the action at two frames per beat, with heavy arrows conducting the rhythm of images — the diagram controls reading rhythm the way a metronome controls practice.
  • A pair of hands moving through a move traces a curve through flat space over time — Tufte notes the resemblance to a Phillips-curve path (p. 60): motion depicted on paper is a time-series, and inherits time-series design obligations (explicit ordering, legible tempo).

§3. Effect-Marketing vs. Method-Instruction (pp. 59–60)

Magic catalogs advertise tricks by depicting only the effect — the Flying Glass of Water ad shows what the buyer's audience will see, never the gimmick. Viewpoint selection is itself a sales strategy: when a depiction shows only outcomes, you are being sold, not taught. The modern equivalents are the vendor architecture diagram with no failure paths, the demo that never shows the config, the before/after with no mechanism between.

Two further tactics from the same pages:

  • Text as image: the word GONE lettered in vibrating stripes (p. 59) makes the typography itself perform the vanishing — Tufte admires it as idea, word, and drawing adding up to one coherent, vivid whole. The caution is contextual, not absolute: stripes read as dazzle and deft speed beside a magic trick and as something else entirely on a prisoner's uniform (Rand, cited p. 59) — attention-getting devices take their meaning from their setting.
  • Reader control as the antidote: on paper the reader sets pace, sequence, direction, and focus; read fast, the trick performs — read slowly, it dissects (p. 60). Any medium that removes reader control (autoplaying video, timed slides, unpausable demos) restores the magician's advantage.

§4. The Strategies of Disinformation — a Detection Catalog (pp. 64–65)

Fitzkee's analysis of conjuring reduces to two master strategies: disguise (a thing appears to be what it is not) and attention control (looking is steered away from the method). Every concealment tactic in a display is one of the two. The consumer's catalog:

Conjuring moveDisplay equivalentDetection question
Mask a small motion under a large oneLayout animation, redesign, or transition landing at the same moment as a data changeWhat changed in the numbers while everything else was moving?
Dazzle — vibrating pattern where an edge mattersNoisy texture at the exact point of discrimination (the repetitive stair carpet masking each step edge, p. 65; a railroad-station stair striped parallel to the tread edge — 1,400+ falls in six weeks, p. 65, fn. 18)Does decoration sit precisely where the reader must discriminate?
Divert scrutiny with framingHeavy frame and blocky capitals flattening a damaging disclosure (the Surgeon General's warning on a cigarette billboard, p. 65)Is the mandated or unflattering content styled to repel reading?
Drop the clue under a distractionThe material fact released at the exact moment something louder claims attention — the Agatha Christie method: just as the clue is dropped, a distracting incident occurs (p. 64)What else was demanding attention at the moment the key fact appeared?
Patter and gestureDisplay illegible without the presenter's talk track (see tufte-causal-reasoning-in-graphics on oral escorts)Does the display survive being read alone, cold, on paper?
Self-working trickA format that deceives with no active presenter: truncated default axes, unlabeled baselines, auto-scaled comparisonsWould this display mislead even if its maker had no intent?

"Where scrutiny is damaging, scrutiny is diverted." — Dariel Fitzkee, Magic by Misdirection, quoted in Visual Explanations, p. 65

The incentive audit. Concealment correlates with motive. Before trusting a display, ask who benefits if it is skimmed rather than studied — a question about the presenter, not the pixels. Tufte places conjuring in a family with political fallacies (Bentham), strategic intelligence (Dulles), lie detection (Ekman), and the psychology of deception (Hyman) (p. 64, notes): deception methods generalize across domains, and magic is simply the domain where they are practiced openly enough to study.

The perceptual substrate. These tactics work because of visual masking — a target stimulus made less visible by a spatiotemporally overlapping mask (Breitmeyer, cited p. 64). The tactics are not rhetorical tricks; they exploit measured limits of the visual system, which is why good-faith viewers cannot simply "look harder."

§5. Copied Errors — Provenance Heuristics (pp. 66–67, 71; p. 35)

The full Automaton Chess Player case lives in tufte-narrative-and-sequence §4. What this skill adds is the working heuristics the case (and its neighbors) yield:

  1. Craft is orthogonal to truth. The most beautifully executed diagram of the Automaton's interior was a wrong guess; execution quality is evidence of effort, not of accuracy. Never let rendering polish stand in for verification.
  2. Copying substitutes for observation, then compounds. Reginald Scot's 1584 Discoverie of Witchcraft founded a 300-year course of literary theft in conjuring texts, some copied mostly word-for-word from Scot's pages (p. 67, fn. 24); a 1622 map depicting California as an island was reproduced in 182 variants, the last copyist publishing in 1745; Dürer's two-horned rhinoceros was taken as real for some 200 years (p. 71); a six-fingered hand in a card-magic illustration went unnoticed for years until a fellow illustrator finally spotted it — and answered with a deliberately six-fingered homage (p. 71, fn. 38). Each copy is one more authority the next copyist will cite.
  3. Degradation is a copying error too. Textbook redrawings of Snow's cholera map drop the labels on the brewery and workhouse, turning the essential compared-with-what cases into mysterious cholera-free zones; some facsimiles omit the map entirely (p. 35). A faithful-looking copy that sheds labels, scales, or contrary cases is a false diagram with an honest ancestor.
  4. Detecting a clever fraud takes a practitioner of the craft. Brilliant non-specialists are the easiest marks — Franklin played the Turk and lost; Tufte's sources (Gardner, Hyman, Randi; p. 66, fn. 22) insist that fraud review requires someone fluent in how the deception is done. Staff the audit accordingly: a designer to catch design tricks, a statistician to catch statistical ones.
  5. The provenance question for any inherited diagram: did its maker observe the mechanism, or copy another drawing? If the chain ends at a drawing rather than at the thing itself, treat the diagram as a hypothesis.

§6. The Honest-Presentation Counterpart (pp. 68–70)

Hoffmann's 1876 maxims for magicians — never announce the trick in advance, never repeat it — are context-suppression rules; inverted, they become teaching (announce, then repeat). The PGP structure and the six rules for technical presentations are catalogued in tufte-narrative-and-sequence §5. The residue here:

  • The audience frame decides everything. Conjuring treats spectators as marks to be kept ignorant; teaching treats them as minds to be informed. Every downstream choice — announce or surprise, repeat or vary, paper or patter — follows from which frame the presenter is actually in (p. 68). A presentation optimized for astonishment is using the wrong frame for evidence.
  • Attention choreography can serve revelation. Nelms' stagecraft (p. 69) channels attention with honest mechanics: point without a counter-gesture that splits the pointing two ways; change hands when passing a prop across your body; show a small prop fully, against a contrasting background; step off with the foot nearest where you are going. The same choreography that hides a palm can walk an audience through a proof — attention control is a tool, and only its purpose is moral.
  • Paper is the anti-patter. Handouts create a testimonial record — the audience can later ask "didn't you say this?" (p. 69). Accountability is a feature of high-resolution paper and a structural impossibility of evanescent talk.
  • Rehearse like a conjurer. Magicians sometimes spend years perfecting a few minutes of material (p. 69); Mosteller's presentation discipline (pp. 69–70) imports that rehearsal ethic into teaching. Mastery of detail is what makes calm, non-manipulative delivery possible.
  • The bridge to interfaces: stage-magic principles have been explicitly carried into human-interface design (Tognazzini, cited p. 69, fn. 30) — the inverted toolkit applies to software surfaces, not only talks and charts.

§7. The Two Integrity Tests (p. 70)

The five-question display test (truth revealed? representation accurate? data documented? spurious readings avoided? comparisons and context shown?) is catalogued in tufte-narrative-and-sequence §4. Tufte adds a second, consequentialist test that belongs here:

Judge a display by the decisions it escorted. The cholera map is vindicated because the epidemic analysis ended an outbreak; the Challenger charts are condemned because the launch proceeded; the corrected staircase stops causing falls; a good route map gets the traveler there. When outcomes are observable, they audit the display more honestly than aesthetics ever will.

Professional floor, same page: state monetary series in inflation-adjusted units as routine practice, and avoid rainbow color codings whose hue ordering confounds the data ordering.


§8. Named Failure Modes

Failure modeDescriptionVE page
Effect-only depictionShowing outcomes while withholding mechanism — marketing posing as explanation59–60
Wrong-viewpoint instructionSpectator view drawn for a reader who must operate; no rotation, no backstage view55–58, 61
Compressed verbHidden fast action explained at the same length as visible slow ones58
Dazzle at the discrimination pointDecorative pattern or texture placed exactly where the reader must distinguish65 (+ fn. 18)
Scrutiny diversionFraming, casing, or styling that repels reading of damaging content65
Clue-under-distractionThe load-bearing fact released at the moment something louder claims attention64
Patter dependenceDisplay that collapses without its presenter's talk track64; cf. CYA notices
Self-working deceptionFormat defaults (truncated axes, auto-scaling) that mislead with no operator intent65
Polish-as-proofTrusting a diagram because it is well-executed (the Automaton trap)66–67
Copy-chain diagramInherited drawing whose provenance ends at another drawing, not the mechanism67, 71
Degraded copyRedrawing that sheds labels, scales, or contrary cases while keeping the look35, 67
Astonishment framePresentation structured to surprise rather than inform — conjurer's frame on evidence68

§9. Do / Don't

DoDon't
Decide viewpoint first: effect, method, or both — and for which readerDefault to the spectator view for an operator's task
Rotate the scene into the operator's hands when the reader must performForce the reader to mentally mirror every step
Spend pages on the fast, hidden step; let slow visible steps be briefBudget explanation by elapsed time instead of hiddenness
Use arrows to conduct reading tempo through a sequenceScatter undifferentiated arrows that point without pacing
Demand the mechanism when shown only effectsAccept outcome-only vendor diagrams as explanation
Put damaging or mandated content in the display's calmest, clearest registerStyle disclosures to repel the scrutiny they exist for
Give the key fact a clear moment and the visual weight its role deservesRelease it while attention is drawn elsewhere
Test every display by reading it alone, cold, on paperLet patter carry what the display should
Trace inherited diagrams to an observation of the thing itselfCite a chain of copies as if it were replication
Staff deception review with practitioners of the relevant craftAssume intelligence alone detects skilled fraud
Choreograph attention to walk viewers through the argumentUse attention control to walk them past it
Judge displays by the decisions they escortedJudge them by execution quality and applause

§10. Pre-Ship / Audit Checklist

Designing an instruction or reveal:

  1. Reader identified — spectator, student, or operator? Viewpoint(s) chosen to match?
  2. Both the visible effect and the hidden mechanism depicted (flap, cutaway, layers) where the reader needs both?
  3. Hardest-to-see step given the most depiction space?
  4. Sequence tempo conducted (numbered frames, weighted arrows), not implied?
  5. Display self-sufficient on paper, without your voice?

Auditing someone else's display:

  1. Who benefits if this is skimmed? What would they conceal?
  2. What moved while the layout moved?
  3. Is decoration sitting on a discrimination point?
  4. Is any disclosure styled to repel reading? Any key fact released under cover of a distraction?
  5. Would the format mislead even with an innocent presenter (self-working deception)?
  6. Where does the diagram's provenance chain end — at the mechanism, or at another drawing?
  7. What decision did (or will) this display escort, and did the outcome vindicate it?

Sources

  • Visual Explanations: Images and Quantities, Evidence and Narrative, Edward R. Tufte, Graphics Press, Cheshire, Connecticut, 1997 — Ch. 3, pp. 55–71 (magic, pictorial instruction, disinformation, presentations); p. 35 (degraded redrawings of Snow's map); p. 74 (large-motion-covers-small-motion named among the universal principles).
  • Cross-references: tufte-narrative-and-sequence (§4–§5: masking, still-land, Automaton, integrity test, PGP, six rules), tufte-causal-reasoning-in-graphics (oral escorts, selective data), tufte-evidence-corruption (presentation corruption in Beautiful Evidence), tufte-instructions-at-point-of-need (where instructions should live), tufte-multifunctioning-elements (double-functioning marks).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.