# Tufte Meetings And Documents > Use when a slide deck is scheduled for a decision meeting, design review, PRD review, or briefing; when planning a meeting's time budget, agenda, or pre-read; when pre-reads go out by email and nobody reads them; when a talk to a large audience needs a handout; when you are the low-power party who keeps getting interrupted; or when choosing between presenting a thing and circulating a document about it. - Skill: `jpoindexter/tufte-meetings-and-documents` (Agent Skill) - Install (CLI): `npx skillmds@latest add jpoindexter/tufte-meetings-and-documents` - Raw SKILL.md: https://api.skillmd.com/api/skills/jpoindexter/tufte-meetings-and-documents/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Product & Planning - Author: jpoindexter (https://skillmd.com/u/jpoindexter) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/jpoindexter/tufte-meetings-and-documents --- # Smarter Presentations and Shorter Meetings ## Overview Chapter 8 of *Seeing With Fresh Eyes* (printed pp. 151–161) is the operational companion to Tufte's PowerPoint critique: it specifies the **meeting protocol** that replaces the deck. The protocol in one line: begin every nonfiction meeting with a real document and a silent study hall, then discuss — and finish early. The chapter supplies the time budgets, the presenter's playbook, the audience's obligations, a scaling method for large lectures, and a low-power variant (the Medical Concerns List) for when the meeting belongs to someone else. The theoretical spine is resolution: **spatial adjacency versus temporal stacking** (p. 154). A document puts the evidence adjacent in space, inside the reader's eyespan, under the reader's control; a deck stacks thin fragments in time, under the presenter's control. Which one you choose decides who navigates, who compares, and how long the meeting lasts. > Division of labor: `tufte-cognitive-style` owns the format critique — the ten deficiencies of slideware, the density tables, the Columbia case, and the *Beautiful Evidence* 4-pager spec. This skill owns the SWFE meeting protocol: how to actually run the room. The two Columbia investigation boards' condemnation of engineering-by-PowerPoint (recalled at SWFE p. 154, at Tufte's prompting) is covered there; the A3 handout instances here (pp. 150, 155) are SWFE's own worked examples, not restatements of the 4-pager. --- ## §1. The Protocol — Document, Study Hall, Discussion The core prescription (p. 151), stated as rules: 1. **Write a document, not a deck.** A briefing paper, narrative document, or technical report — 2 to 6 pages, written in **sentences**, with appropriate images and data displays. Paper or electronic. 2. **Do not send it out in advance.** People won't read it. Hand it over as people arrive or sign in: "Read this, then we'll talk about it." 3. **Run study hall.** Silent reading, taken seriously: **20% to 50% of total meeting time**. Meetings with several topics may have several silent reading periods. 4. **Then talk** — about the document, not through it (§3). 5. **Finish 10% to 20% early** (§5). Why reading beats listening: audience members read **2 to 3 times faster** than the presenter can talk, and every reader chooses their own pace and their own points of close attention. In a slide presentation, viewers control neither pace nor sequence — Tufte's image is sitting in the dark waiting for the diamonds in the swamp. Why documents beat decks at the source: decks are easier to *prepare* — documents require coherence, thinking, sentences — and that convenience is exactly the problem. > "Optimizing presenter convenience is selfish, lazy, and worst of all, replaces thinking." — Tufte, *Seeing With Fresh Eyes*, p. 151 The two industry witnesses (p. 151): > "People who know what they're talking about don't need PowerPoint." — Steve Jobs, quoted in Walter Isaacson, *Steve Jobs* (2015), 337; reproduced SWFE p. 151 **Bezos on Amazon (Charlie Rose interview, 2012, quoted p. 151):** staff meetings begin with about 30 minutes of silent reading — study hall — because a presentation gives the audience very little information, only bullet points; easy for the presenter, difficult for the audience. Every Amazon meeting is structured around a 6-page narrative memo instead. **The practical guarantee (p. 151):** meetings become smarter and more efficient, the audience more active, and meetings run 10–20% shorter. None ever wished them longer. --- ## §2. Why It Works — Spatial Adjacency vs. Temporal Stacking (p. 154) Documents and the web show information **adjacent in space** — on high-resolution paper or scrolling screens, within the viewer's common eyespan. Spatial adjacency activates the highest human skills: scanning spatially distributed information and drilling into micro-spaces for the diamonds. Every viewer proceeds at their own pace and their own sense of what matters. **Resolution governs the trade-off.** Higher display resolution permits more spatial adjacency; lower resolution forces more temporal stacking. Tufte's limiting case is the phone answering tree — the dreary asymptote of binary take-it-or-leave choices stacked in time ("choose 1 to leave a message, 2 for appointments…"), where slightly higher resolution would show all choices at once and callers would choose by seeing. In the last 25 years, screen resolution rose from ~5% of paper's to more than 80%, finally making displays worthy of the human eye-brain system — which removes the last technical excuse for stacking. | Adjacent in space (document, web page) | Stacked in time (deck, phone tree) | |---|---| | Reader controls order, pace, re-reads | Presenter — an unreliable narrator — controls all three | | All comparisons available at will | Only the presenter's chosen sequence of comparisons | | Lives on after the meeting; can be re-consulted, cited, audited | Doesn't live beyond the moment (or survives as a 4-per-page printout, worse than either form) | | Imposes no structural model on the content | Imposes a hierarchical model — NASA and military contractors run 6 bullet levels; the F35 program was documented entirely in slides | The institutional aside (p. 154): high-ranking officers keep roomfuls of hundreds of soldiers and sailors working full-time producing decks — a standing measure of what temporal stacking costs an organization. (The format-level analysis of why bullets fail: `tufte-cognitive-style` §§2–4.) --- ## §3. The Presenter's Playbook **You have not lost control — you wrote the document.** Study hall is a gift to presenters (p. 151): everyone showed up, everyone is reading your work, and nobody is looking at you. If someone peeks at email during study hall, gently glare — the purpose of gathering is total concentration on the content at hand. Your job is intellectual leadership, which is why you're the one presenting. **Questions answer themselves downstream.** A document naturally handles questions by answering them further down the page — the reader keeps reading instead of interrupting (p. 151). **After study hall, do not repeat what they just read** (p. 153). The audience already knows a lot; recitation insults the reading. Instead, dig deeper into specific parts by address: "In the third paragraph of the budget statement, we don't have a consensus." The audience turns to the paragraph, you put it on the screen, and — the chapter's nice escalation — you may hand out *another* document on the budgetary specifics. **Rehearse** (p. 153). Rehearsal improves performance; a diplomatic colleague or a rehearsal video ruthlessly reveals incoherence, nervous habits, and filler placeholders (*like, um, as I mentioned earlier*). The audience appears to take notes but is actually keeping score — Tufte imagines them betting "3 *likes* in the next minute, for a latte." Identifying problems eliminates them; don't let rehearsal nerves deflect you. **Content and credibility (p. 152).** The audience seeks exactly two things: the substantive content, and reasons to believe the presenter. | Lever | Practice | |---|---| | The trouble test | A sure sign of trouble is inability to write one paragraph stating: what the problem is; why it is relevant and why anyone should care; what you are going to do to solve it | | Content quality ladder | Documents model information better than decks; data paragraphs beat sentences; sentences beat bullet lists. If your numbers are boring, get better numbers | | Expert quotations | Quote named experts with credentials, on-topic — they teach correct technical vocabulary, say things presenters can't, and demonstrate analytical thinking. Never faux-inspirational cheerleading | | Documentation as QC | The fundamental integrity mechanism: authors named, sponsors revealed, interests and agendas unveiled, measurements verified; provide data downlinks with a clean codebook. Fear that others may inspect your data encourages getting it right in the first place | | Reputation | Credibility = a continuing record of honest communication *and* getting it right — helped by not having lied to your colleagues last week | | Respect the audience | The common audience-research error is underestimating the audience. Don't spin, pander, or dumb down. Get it right, be honest, make everyone smarter — all three | --- ## §4. Named Time Budgets — Show Up Early, Finish Early (p. 153) | Budget | Value | Mechanism | |---|---|---| | Study hall | 20–50% of total meeting time | Reading is the meeting's information transfer; talking is its analysis | | Amazon variant | ~30 minutes of silent reading on a 6-page memo | Fixed opening block, every staff meeting | | Arrive | Early, before your own presentation | Start arrivals on study hall immediately — document at each seat or screen, plus a gracious slide: "read this now," "don't touch your cell phone here." Early arrival also catches logistics failures (double-booked rooms). Higher-ups especially: bosses who flounce in late with an entourage are disrespecting their colleagues | | Finish | 10–20% early | The audience is thrilled; word spreads through the building ("they finished early!"); people leave saying "great meeting" | | Multi-topic meetings | One reading period per topic | Several short study halls, not one long one | Tufte closes the finish-early passage with the one prediction he refuses: people leaving a great meeting still won't ask "when's the next meeting?" — not in your dreams. **Study hall discipline transfers whole:** phones off (his course slide: all cell phones OFF during study hall and class — use the lobby instead, p. 159); the reading period is sized to the document, not padded; reading is attendance. --- ## §5. The Audience's Obligations — Responsibilities of Spectatorship (p. 154) The protocol binds both sides. When you attend a presentation: - **Stay on the content** — that's why you, the presenter, and the observed objects are all there. - **Open mind, not empty head** (the Austin Kleon drawing of Tufte's maxim, p. 154). Be able to change your mind. - **Loot the presentation for useful material**; make the best of it. Don't let contrary elements spoil your seeing and learning. - **Disagreement is not diabolism.** Someone disputing the third paragraph of your budget statement is not Satanic; their motives are no better or worse than your own. If you require perfect agreement with presenters, Tufte suggests staying home and staring at your immutable self in the mirror. - Listen, see, think, learn — and treat presenters as you would like to be treated. The two epigraphs over the next section (p. 156) belong to every meeting's post-mortem: > "It's not what you say, it's what they hear." — Red Auerbach, quoted in *Seeing With Fresh Eyes*, p. 156 > "The biggest problem in communication is the illusion that it has taken place." — attributed to George Bernard Shaw, *Seeing With Fresh Eyes*, p. 156 --- ## §6. Scaling Up — Handouts for Large Lectures and Courses **The A3 handout (p. 155).** For large audiences, distribute a paper handout — A3 / US tabloid, printed both sides. Attendees look it over before the talk and follow along during it; the paper is a take-away that documents and memorializes the talk; and if your time runs short, **the audience still receives your information**. It also carries relevant material *not* presented in the talk at all. Tufte's own instance: the two-sided sheet prepared for his machine-learning keynote to 900 people — a dense quilt of sparklines, statistical maxims, Feynman and Tukey quotations, and data-forensics pointers (the quilt-of-sources form; see Chapter 9's visual-index concept, p. 162). **Q&A scales badly (p. 160).** Audience questions work in small groups (under ~25 people); in large audiences questions become off-point mini-speeches. Tufte's ranking of what audience members want to hear: first themselves, second the presenter, and last their fellow audience members. Remedies: work answers to recurring questions into the lecture and the handout — and note that study-hall reading already answers many questions before they're asked. **The one-day course as a study-hall system (pp. 159–160).** Tufte's "Presenting Data and Information" course (27 years, 930 course days, 320,000 students) hands every student four books and opens with a 30–40 minute study hall against specific page assignments — read carefully these pages, scan those. Study hall recurs through the day, including timed 8-minute reads before a topic. The instructive details: | Practice | Point | |---|---| | Look at the books, not slides — 4 books, ~700 pages, 2,000 illustrations, "the highest-resolution portable interface in the world in your hands" | Students conditioned by slideware stare at the projection screen even when holding something better | | "Books are meant to be used. Dog-ear the pages, mark them up, put notes in the spacious margins." | The handout is a working surface, not a keepsake (see `tufte-annotation-as-thinking`) | | Reading periods deliberately vary the rhythm — video, live commentary, silent reading, plain talking | A 4–5 hour session needs mode changes; study hall is one of the instruments | | Sound checks, speech exercises, venue specs (presenter at floor level with the audience, two large 16:9 screens) | Logistics are part of the presentation, rehearsed like content | | Course material changes ~15% per year; teaching exposes incoherencies in the book-in-progress | A good way to learn about something is to teach it | --- ## §7. The Low-Power Variant — the Medical Concerns List (pp. 156–158) The chapter's second protocol inverts the first: what to do when the meeting is *not yours* and the other party controls the clock. The instance is the doctor's appointment (section co-written with Dr. David S. Smith, Yale Internal Medicine), and the motivating statistic is brutal: empirical studies ~50 years ago found doctors first interrupt the patient after **18 seconds** of talking; after years of medical schools teaching "listen to the patient," the recent replication found **22 seconds**. The fix is a patient-authored document — a written list of current concerns, symptoms, and questions, **recent and relevant to this appointment** — deployed with study-hall mechanics: | Element | Rule | Why | |---|---|---| | Authoring | Prepare at home, in serenity and privacy; review with someone who knows you well; rehearse | Information composed free of social pressure beats answers improvised under it — and embarrassing items make it onto paper that wouldn't survive the room | | Delivery | Hand it over at the **exact moment of need** — the start of the appointment. Do not pre-email it | Doctors are busy; a pre-read won't be read (the same rule as §1's "don't send in advance"). See `tufte-instructions-at-point-of-need` | | The study-hall cue | Hand over the list, then look pointedly at your own copy: "Here it is, I have 5 issues" | The reader reads several times faster than you can talk; the cue starts the reading | | Agenda function | The list sets the agenda and the allocation of time; check items off as the visit proceeds | Ensures lower-level issues aren't abandoned when time runs short | | The doorknob save | A list read early prevents "by the way, one more thing — I have chest pain" as the visit concludes | That closing aside is usually the patient's *main* worry, and often serious | | Distribution | Multiple copies — doctor, nurses, students, techs; refreshed if transferred; essential in the ER | Everyone who acts on you reads the same document | | Content discipline | Report precise observations, dates, frequencies; photos/video of transient symptoms; speculations labeled as speculations | Observation beats interpretation — see Lucky, below | **The Lucky case (p. 158).** A Golden Retriever's Medical Concerns List — dated episodes, heart rates by year, 12 pages of attached test results — taken to a top cardiology department. The list noted that only her *hind* legs went wobbly; her owners "had no idea this meant anything, just described what we saw." That observed-not-interpreted detail ruled out the vagus-nerve hypothesis (which affects all four legs). The lesson generalizes: record what you saw, not what you concluded, because you cannot know which details are diagnostic. **Transfer beyond medicine:** any asymmetric meeting where you hold the information and the other side holds the clock — skip-level 1:1s, vendor escalations, legal consultations, incident reviews, user-research sessions. Bring the list, hand it over at the start, cue the reading, march the items. --- ## §8. The Fire Chief Case — the Protocol Demonstrated (p. 150) Chapter 7 closes with the protocol's cleanest worked example (an annotated Turgot *Plan de Paris* standing in for the city). At Town Council meetings the Fire Chief reports monthly on fires, alarms, false alarms, inspections, emergencies, and expenditures — hundreds of data points, in a meeting with 10 other agenda items. Instead of a slideshow, the Chief hands out **one A3 / 11×17 sheet: an annotated map of monthly activities on the front, budget line-item expenditures on the back** — to council members, news reporters, and citizens alike. The Chief opens: "Look this monthly report over, and then we will talk about it." What happens next is the payoff: council members start asking *specific, located* questions — why all the Marina false alarms? what caused the Trois Marie fires? why so few Southend inspections, so many fires? what about these out-of-town travel expenses? The document made the audience active interrogators instead of passive viewers. And two structural advantages compound: - **Documents live on, unlike slides.** Each member can look at any part of the report at any time — including later in the meeting, while the Chief's rival the Police Chief marches through a tiresome slideshow, inviting the comparison and perhaps a request to stop it with the slides and bring a document like the fire map. - **One artifact serves every audience** — council, press, public — at their own depths (compare the two-reader formats in `tufte-annotation-as-thinking` §5). Also a judicial sibling (p. 153): in a trade-secrets trial, Federal District Judge William Alsup provided the jury with paper handouts like opera program notes — a "handy-dandy list" of the alleged trade secrets — and urged the two sides to submit a glossary of the top dozen people, so jurors could navigate a complex case. --- ## §9. Modern Transfer — Design Reviews, Standups, PRD Reviews, Async Culture | Ritual | Deck-era failure | Protocol version | |---|---|---| | Design review | Designer click-throughs a Figma deck; feedback is reaction to the last frame shown | 2–6 page document (flows, screenshots, the *why*, open questions) read silently in-room for the first 20–40% of the slot; then discussion by section | | PRD / spec review | Pre-read emailed 3 days early; nobody read it; author re-presents it as slides | Doc handed out at meeting start; timed study hall; discussion addressed to paragraphs, not vibes | | Decision meeting | Options deck steered by the presenting team's sequence | Written options memo with the trouble-test paragraph (§3) up top; readers compare options adjacently, at will | | Standup | Status recited serially — temporal stacking of thin fragments | Written status adjacently visible (board/doc); the meeting handles only exceptions and blockers | | Large all-hands | 45 slides, Q&A mini-speeches | Two-sided A3-equivalent doc; short talk digging into parts of it; recurring questions pre-answered in the doc (§6) | | Incident review | Slide chronology curated by the on-call team | Timeline document studied silently, then interrogated — the Fire Chief pattern (§8) | **The async caveat, stated honestly.** Tufte's protocol is *synchronous silent reading*, and his rule is explicit: don't send the document in advance, because people won't read it (p. 151). An "async docs culture" that emails pre-reads is running the part of the protocol Tufte rejected while skipping the part that works — the protected, observable reading block. Async documents succeed only where reading is genuinely budgeted and verifiable (comment deadlines held, decisions blocked on documented review); otherwise schedule the study hall. The meeting *is* the reading time. --- ## §10. Named Failure Modes **M1 — Pre-Read Theater.** Document sent days ahead; nobody read it; the meeting proceeds as if they had — or worse, the author re-presents the unread doc as slides. *Fix:* hand out at the start; run study hall (§1). **M2 — Deck Rehydration.** A good document exists, and someone converts it into bullets "for the meeting." The conversion strips the sentences that carried the reasoning. *Fix:* project the document itself when a shared view is needed (§3). **M3 — Study Hall Skipped Under Pressure.** "We're short on time, let's just discuss." The information transfer is deleted to protect the talking. *Fix:* the reading is 2–3× faster than any spoken summary — cut discussion, never reading. **M4 — Repeat-the-Read.** The presenter narrates what the room just read. *Fix:* go deeper by address ("third paragraph of the budget statement"), or go home early (§3). **M5 — Padding to the Slot.** The meeting expands to fill its calendar block. *Fix:* finishing 10–20% early is a named deliverable, not an accident (§4). **M6 — Bullet Memo.** A "6-page memo" that is six pages of bullet fragments — the deck's cognitive style smuggled into the document container. *Fix:* sentences with agents and verbs; the trouble-test paragraph up top (§3; format analysis in `tufte-cognitive-style` §3). **M7 — Doorknob Concern.** The real issue surfaces as the meeting ends, unexamined. *Fix:* a concerns list handed over and read at the start guarantees the main worry is on the table in minute one (§7). **M8 — Q&A Mini-Speeches.** Open floor in a large audience produces speeches, not questions. *Fix:* pre-answer recurring questions in the handout; reserve open Q&A for groups under ~25 (§6). **M9 — Underestimating the Audience.** Spin, pandering, dumbing down — the common audience-research error. *Fix:* endless respect; get it right, be honest, make everyone smarter (§3). **M10 — Boss Flounce.** Senior people arriving late with an entourage, resetting the room and taxing everyone's reading. *Fix:* the higher your rank, the earlier you show up (§4). **M11 — Presenter-Convenience Optimization.** Choosing the artifact that is easiest to produce over the one that transfers most. The deck's whole appeal — and the tell that thinking is being replaced (§1). --- ## §11. Application Checklist Before your next decision meeting, review, or briefing: - [ ] Is there a 2–6 page document in sentences, with real images and data displays — not a deck, not a bullet memo? - [ ] Does it open with the trouble-test paragraph: the problem, why anyone should care, what you propose to do? - [ ] Are authors named, sponsors and interests revealed, sources documented, data reachable? - [ ] Is the document distributed at the meeting's start — not emailed in advance? - [ ] Is study hall on the agenda, sized to 20–50% of the meeting (one period per topic)? - [ ] Phones off during study hall; reading treated as the meeting, not a delay before it? - [ ] Is your talk planned as deeper-by-address, not a recitation of what was read? - [ ] Have you rehearsed in front of a diplomatic critic or a camera? - [ ] Will you arrive early and finish 10–20% early — on purpose? - [ ] For large audiences: two-sided A3-class handout carrying the talk plus what won't fit in it; recurring questions pre-answered? - [ ] If it's someone else's meeting: is your concerns list written at home, copied for everyone, handed over at minute one with the reading cued — observations, not interpretations? - [ ] Afterward, apply Auerbach's test: not what you said — what did they hear? --- Source: Edward Tufte, *Seeing With Fresh Eyes: Meaning, Space, Data, Truth* (Graphics Press, 2020), Chapter 8, "Smarter Presentations and Shorter Meetings," printed pp. 151–161, with the Fire Chief case from Chapter 7's close (p. 150, annotated map adapted from Turgot & Bretez, *Plan de Paris*, 1739) and the visual-index/quilt concept from Chapter 9 (p. 162). Medical Concerns List section co-authored in the book by Dr. David S. Smith (Yale Internal Medicine) and ET. Quotation credits per the book's back matter (p. 171): Steve Jobs in Walter Isaacson, *Steve Jobs* (2015), 337; Jeff Bezos, Charlie Rose interview, 16 November 2012. The deck-format critique, density data, and Columbia case are covered in `tufte-cognitive-style`. All passages paraphrased; quoted lines limited to single attributed sentences with page citations.