Seven Seconds: The Science of Writing That Gets Read, Trusted, and Remembered

18 min read

Seven Seconds: The Science of Writing That Gets Read, Trusted, and Remembered

A recruiter spends 7.4 seconds on your resume. A LinkedIn user decides in the first two lines whether to tap “see more.” A Substack subscriber opens your email and gives you one paragraph to earn the next. This path teaches the cognitive science under copywriting — not formulas to follow, but the reasons they work — so you can write for any platform, any audience, any argument.

Who this is for

You are a knowledge worker who knows things worth sharing. You have deep expertise in a domain — maybe several. You can explain complex ideas in conversation. But when you sit down to write, the result is either too dense for anyone to finish or too thin to be worth publishing.

This path is for you if:

  • You want to translate deep knowledge into articles, posts, and thought leadership — not just notes for yourself
  • You have tried writing for LinkedIn, Substack, or a blog and felt that something was off but could not name what
  • You want to understand why writing rules exist, not just memorise them
  • You are building a body of work and need each piece to earn the reader’s next minute

What this path is NOT

This is not a list of copywriting hacks. It is the cognitive science and structural principles under all good non-fiction writing, explained so you can apply them to any format on any platform. If you understand the why, the how becomes obvious.

Part 1 — The fluency principle

Every copywriting rule you have ever heard — “write short sentences,” “use simple words,” “break up your paragraphs” — is a surface symptom of a single underlying mechanism. Understand the mechanism and you will never need to memorise the rules.

The mechanism is cognitive fluency.

Reber and Schwarz demonstrated in 1999 that statements presented in easier-to-read formats are judged more likely to be true.1 Not more persuasive — more true. The reader’s brain uses processing ease as a shortcut for credibility. If it is easy to read, it feels honest.

This connects directly to Kahneman’s dual-process theory.2 Simple, fluent writing engages System 1 — fast, automatic, trusting. Complex writing activates System 2 — slow, deliberate, sceptical. When your reader hits a sentence they have to re-read, they shift from trusting to scrutinising. You lose them not because they disagree, but because you made them work.

graph LR
    A[Simple<br/>fluent writing] --> B[System 1<br/>fast, trusting]
    B --> C[Feels true<br/>keeps reading]
    D[Complex<br/>dense writing] --> E[System 2<br/>slow, sceptical]
    E --> F[Feels hard<br/>stops reading]
    style C fill:#5cb85c,color:#fff
    style F fill:#e74c3c,color:#fff

Meanwhile, Jakob Nielsen’s eye-tracking research (2006, n=232 users) showed that people scan web pages in an F-pattern: two horizontal sweeps at the top, then a vertical scan down the left.3 They do not read your article. They scan for reasons to keep reading or stop.

And George Miller’s working memory research (1956, refined by Cowan in 2001) tells us why long paragraphs fail: human working memory holds 3–5 chunks at once.4 A paragraph with six distinct ideas exceeds capacity. The reader forgets the first idea before reaching the last.

Every “rule” of copywriting — short sentences, active voice, white space, clear structure — is a strategy for maintaining cognitive-fluency. The reader’s brain is constantly asking one question: is this worth the effort? Your job is to make the answer yes, every sentence.

The principle under all the rules

Writing well is not about dumbing down. It is about reducing the cognitive cost of understanding so the reader can spend their mental energy on your ideas, not on parsing your sentences. Fluency serves depth, not simplicity.

Part 2 — The curse of knowledge

If cognitive fluency explains why bad writing fails, the curse-of-knowledge explains why smart people write badly.

Camerer, Loewenstein, and Weber coined the term in 1989.5 The finding: once you know something, you cannot accurately simulate what it is like not to know it. Better-informed agents systematically overestimate how much less-informed agents understand.

Steven Pinker called it “the single best explanation I know of why good people write bad prose.”6 The expert writes for someone who already shares their vocabulary, their assumptions, their frame of reference. The reader — who does not share those things — bounces.

graph TD
    A[Expert's mind<br/>dense, connected,<br/>full of context] --> B[Writes as if<br/>reader has<br/>same context]
    B --> C[Reader is lost<br/>blames themselves<br/>or stops reading]
    D[Expert recognises<br/>the curse] --> E[Bridges the gap<br/>with analogies,<br/>structure, rhythm]
    E --> F[Reader follows<br/>feels smart<br/>keeps reading]
    style C fill:#e74c3c,color:#fff
    style F fill:#5cb85c,color:#fff

The antidote is not “write simply.” It is test your assumptions about the reader. Specifically:

  • Name what you are assuming. If your sentence works only for someone who already knows the concept, you are writing for yourself.
  • Lead with the familiar. Attach new ideas to things the reader already understands. Analogy is the expert’s bridge.
  • One idea per paragraph. Cowan’s 3–5 chunk limit means that a paragraph doing two things does neither.
  • Read it aloud. Paul Graham’s test: “Is this the way I’d say this if I were talking to a friend?”7 If not, rewrite.

The Flesch-Kincaid readability research puts a number on it: general-audience writing should target Grade 8 (age 13–14). The New York Times writes at Grade 8–9. This is not condescension — it is respect for the reader’s time and attention.8

The paradox

The deeper your expertise, the harder it is to write about it clearly. Fluency is not a natural byproduct of knowledge — it is a separate skill that must be practised. The curse of knowledge is the reason that the most knowledgeable person in the room is often the worst communicator in the room.

Part 3 — Architecture: how a piece holds together

You know why writing works (fluency) and why it fails (the curse). Now: how to build a piece that carries the reader from start to finish.

Three frameworks have survived decades of practice because they map to how the brain processes sequential information:

graph LR
    A[AIDA<br/>Attention<br/>Interest<br/>Desire<br/>Action] --> D[Best for:<br/>persuasion<br/>calls to action]
    B[PAS<br/>Problem<br/>Agitation<br/>Solution] --> E[Best for:<br/>pain points<br/>thought leadership]
    C[Inverted Pyramid<br/>Conclusion first<br/>then supporting<br/>detail] --> F[Best for:<br/>informational<br/>SEO, news]
    style B fill:#4a9ede,color:#fff

AIDA — Attention, Interest, Desire, Action — was created by E. St. Elmo Lewis in 1898, drawing from William James’s psychology.9 It is a persuasion sequence: grab attention, build interest, create desire, prompt action. Still used in sales copy, landing pages, and ads.

PAS — Problem, Agitation, Solution — emerged from direct-response advertising. Dan Kennedy called it “the most reliable formula for sales ever invented.”10 For thought leadership, PAS is powerful: name the problem your reader faces, agitate it (show the consequences of not solving it), then offer your insight as the resolution. The article you are planning on AI for learning is a natural PAS structure: the problem (AI tools are widely available), the agitation (but they often damage learning rather than help it), the solution (design principles that make AI actually teach).

The inverted pyramid — conclusion first, then supporting detail — originated in Civil War-era journalism when telegraph costs demanded essential facts first.11 It works for web writing because of Nielsen’s scanning research: readers may leave at any point, so front-load the value.

The reason these frameworks work is the serial position effect (Ebbinghaus): people remember the first and last items in a sequence best; the middle fades.12 Your opening and closing sentences carry disproportionate weight. Build your architecture so that the strongest ideas land in those positions.

Go deeper: applying these to thought leadership

For articles that translate deep expertise into public argument, PAS is your default. You are not selling a product — you are reframing a problem. The agitation phase is where your expertise lives: you see consequences the reader doesn’t. The solution is your original insight, not a product pitch.

Part 4 — The hook and the hold

Architecture gets the reader through the piece. But first you have to earn their attention — and then keep it.

David Ogilvy quantified the stakes: “On average, five times as many people read the headline as read the body copy. When you have written your headline, you have spent eighty cents out of your dollar.”13

The reason headlines matter disproportionately is George Loewenstein’s information gap theory (1994): curiosity arises when attention is focused on a gap in one’s knowledge.14 The curiosity-gap functions like hunger — the person is motivated to close it. A good headline creates a gap that can only be closed by reading.

graph LR
    A[Headline<br/>creates a gap] --> B[Reader feels<br/>the gap]
    B --> C[Opens the piece<br/>to close it]
    C --> D[First paragraph<br/>earns the second]
    D --> E[Rhythm and<br/>structure<br/>sustain attention]
    style A fill:#4a9ede,color:#fff

Headline mechanics (from CoSchedule’s corpus analysis): the sweet spot is 11–14 words, ~55 characters. Headlines with 10–15% emotional words outperform. Specificity beats vagueness — “How AI Tutoring Doubled Learning Gains in a Harvard Study” outperforms “AI Is Changing Education.”15

Once inside the piece, rhythm holds the reader. Gary Provost’s demonstration from 100 Ways to Improve Your Writing (1985) is the best illustration ever written: five identical-length sentences feel monotonous; varying length creates music that pulls the reader forward.16

And transitions — the “bucket brigade” technique from direct-response copywriting — reset attention between sections. Short phrases (“Here’s what that means:”, “But there’s a problem:”, “The evidence says otherwise:”) act as micro-hooks within the piece, pulling the reader past the points where they would otherwise stop.17

The economics of attention

Every sentence is a transaction: the reader spends attention and expects value. If a sentence gives less value than it costs in effort, the reader leaves. Headlines, openings, rhythm, and transitions are all strategies for keeping the exchange favourable — for the reader, not for you.

Part 5 — Platform physics

You have the principles. Now: why do they apply differently across platforms? Because each platform has different physics — algorithmic mechanics, reader context, and attention economics that shape what works.

Understanding the physics means you never need to ask “what’s the formula for LinkedIn?” You can derive it from first principles.

graph TD
    A[Platform physics] --> B[Algorithm<br/>what gets distributed?]
    A --> C[Context<br/>where is the reader?]
    A --> D[Attention window<br/>how long do you have?]
    B --> E[Write to be<br/>distributed]
    C --> F[Write for the<br/>reader's state]
    D --> G[Write within<br/>the window]
    style A fill:#4a9ede,color:#fff

LinkedIn — the dwell-time machine. The algorithm operates in stages: initial classification (first 60 minutes), then engagement testing with a small follower group. Dwell time — how long a reader stays on your post — is the primary ranking signal.18 “See more” clicks are explicit engagement. This is why LinkedIn rewards short paragraphs, hooks in the first two lines, and line breaks between sentences: they maximise dwell time and “see more” clicks. You are not writing for the reader — you are writing for the reader and the algorithm, and the algorithm measures time-on-post. Also: author responses within 30 minutes yield 64% more comments and 2.3x more views.18

Substack / Newsletter — the subscriber relationship. Your reader chose to receive this. They opened their inbox, saw your name, and clicked. The attention contract is entirely different: they have already signalled interest before reading a word. This is why long-form works on Substack and not on LinkedIn — the audience has self-selected for depth. You don’t need to earn attention in the first two seconds; you need to reward it over 5–10 minutes. The writing can be denser, more nuanced, more discursive. It should still be fluent (Part 1 applies everywhere), but the pressure to hook instantly is lower.

Blog / SEO — the search-intent match. A blog reader arrives from a Google search with a question. They are not browsing — they are hunting. This is why blogs use the inverted pyramid: answer the question immediately, then elaborate. Google’s own documentation prioritises search intent matching and structured content (headings, featured snippet formatting).19 Write the answer in the first paragraph, then use the rest to earn trust and depth. The reader stays because you answered them; they explore because you demonstrated expertise.

Twitter/X — the compression engine. 280 characters is not a limitation — it is a discipline. Each tweet must stand alone as a complete thought. Threads work because each tweet hooks the next (the bucket brigade at scale). Retweets and quote-tweets are the distribution mechanism, so each tweet must be independently shareable. This is why threads outperform single long tweets: they give the reader multiple entry points and multiple reasons to share.

The meta-principle

platform-native-content is not about following rules that change every month. It is about understanding the physics of each platform — what gets distributed, where the reader is, and how long you have — and writing to those constraints. The physics change slowly. The tactics change fast. Learn the physics.

Part 6 — From expertise to thought leadership

You have the science of reading (Part 1), the expert’s trap (Part 2), the architecture (Part 3), the hooks (Part 4), and the platform physics (Part 5). The final piece: how to turn deep knowledge into thought-leadership — writing that changes how people think.

The Edelman–LinkedIn Thought Leadership Impact Study (2024) found that 75% of decision-makers said thought leadership prompted them to research products they hadn’t previously considered, and 70% of C-suite executives said it led them to reconsider a current vendor.20 The 2025 report adds: 95% of potential clients are not actively seeking services at any given time — thought leadership reaches this hidden majority.

The distinction matters: content marketing demonstrates solutions to known problems. Thought leadership reframes the problem itself. Content marketing reflects existing knowledge. Thought leadership offers something the reader did not know they were missing.21

Eugene Schwartz’s Five Levels of Awareness (1966) clarifies the job.22 Your audience is mostly at levels 1–3: Unaware (no problem recognition), Problem Aware (knows the issue, not the solution), or Solution Aware (knows solutions exist, hasn’t chosen). Thought leadership moves people from Unaware to Problem Aware — it names what they have not yet seen.

graph LR
    A[1. Unaware] --> B[2. Problem<br/>Aware]
    B --> C[3. Solution<br/>Aware]
    C --> D[4. Product<br/>Aware]
    D --> E[5. Most<br/>Aware]
    style A fill:#fde8e8,stroke:#e74c3c
    style B fill:#4a9ede,color:#fff
    style C fill:#4a9ede,color:#fff

For your planned article on AI for learning, the structure maps directly:

  • Move 1-to-2: Name the problem most readers haven’t seen — that AI tools marketed as learning aids can damage learning (the Bastani et al. evidence from your Two-Sigma Machine path)
  • Move 2-to-3: Offer the framework — the five mechanisms that produce learning, and why answer machines violate them
  • The thought leadership contribution: the original insight that is yours — the harness view, the scaffold-fade principle, the distinction between cognitive offloading and cognitive leverage. This is what no one else in the room is saying.

The translation from learning path to article is not summarisation. It is re-architecture: pick one insight, build a PAS structure around it, write it at the audience’s awareness level, deliver it in the platform’s physics.

The thought leader's discipline

A learning path is a map of a territory. A thought leadership article is a single trail through it, chosen for a specific reader in a specific context. The expertise is the same. The architecture is different. You are not simplifying — you are selecting.

What you now understand

Concepts you have gained

  • cognitive-fluency — the mechanism under all copywriting rules: easy to read = perceived as true
  • curse-of-knowledge — why experts write badly and how to bridge the gap
  • copywriting-frameworks — AIDA, PAS, and inverted pyramid as cognitive architectures, not formulas
  • curiosity-gap — Loewenstein’s information gap theory and why headlines carry 80% of your dollar
  • platform-native-content — how algorithm mechanics, reader context, and attention windows shape writing on each platform
  • thought-leadership — how to move readers from Unaware to Problem Aware using Schwartz’s five levels

Check your understanding

Where to go next

Build the content engine

You know how to write. Now build the system for producing and distributing consistently. content-strategy covers cadence, formats, repurposing, and sustainability.

Best for: anyone ready to publish regularly.

Own LinkedIn

Part 5 gave you the physics. linkedin-ecosystem goes deeper: the algorithm in detail, content formats that perform, engagement mechanics, and how to build traction from zero.

Best for: anyone whose primary audience is on LinkedIn.

Build the portfolio

Articles are the material. A portfolio is the architecture. Read digital-portfolio to understand how to curate your published work into a visible body of evidence.

Best for: anyone turning writing into career capital.

Write the AI-for-learning article

You have the copywriting principles and the domain knowledge (from two-sigma-machine and ai-self-learning). The article is a PAS structure: the problem (AI as answer machine), the agitation (the evidence it harms learning), the solution (the harness view). Write it.

Best for: you, right now.

Sources

Further reading

Resources

  • Zinsser, W. (2006). On Writing Well — the best single book on non-fiction writing. Four principles: clarity, simplicity, brevity, humanity. Read it first.
  • Pinker, S. (2014). The Sense of Style — the cognitive scientist’s guide to writing. Dense but worth it for the chapter on the curse of knowledge alone.
  • Kleon, A. (2014). Show Your Work! — the manifesto for making your thinking visible. Short, practical, complements this path.
  • Handley, A. (2014). Everybody Writes — practical copywriting guidance for content marketers. More tactical than this path; good as a reference.
  • Graham, P. (2015). Write Like You Talk — one essay, five minutes, will change how you draft.
  • Provost, G. (1985). 100 Ways to Improve Your Writing — the source of the sentence-rhythm demonstration. Short and direct.
  • Edelman & LinkedIn. 2025 B2B Thought Leadership Impact Study — the data on why thought leadership works at the executive level.

Footnotes

  1. Reber, R. & Schwarz, N. (1999). Effects of Perceptual Fluency on Judgments of Truth. Consciousness and Cognition, 8(3), 338–342. ↩

  2. Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux. ↩

  3. Nielsen, J. (2006). F-Shaped Pattern for Reading Web Content. Nielsen Norman Group. n=232 users. ↩

  4. Miller, G.A. (1956). The Magical Number Seven, Plus or Minus Two. Psychological Review, 63(2), 81–97. Revised by Cowan, N. (2001). The Magical Number 4. Behavioral and Brain Sciences, 24(1), 87–114. ↩

  5. Camerer, C., Loewenstein, G. & Weber, M. (1989). The Curse of Knowledge in Economic Settings. Journal of Political Economy, 97(5), 1232–1254. ↩

  6. Pinker, S. (2014). The Sense of Style: The Thinking Person’s Guide to Writing in the 21st Century. Viking. “The single best explanation I know of why good people write bad prose.” ↩

  7. Graham, P. (2015). Write Like You Talk. ↩

  8. Flesch, R. & Kincaid, J.P. (1975). Flesch–Kincaid Readability Formulas. General audience target: Grade 8. NYT writes at Grade 8–9. Overview: Readable.com. ↩

  9. Lewis, E. St. Elmo (1898). AIDA model. First appeared anonymously in Printers’ Ink, 9 February 1898. ↩

  10. Kennedy, D. (2006). The Ultimate Sales Letter. Adams Media. “The most reliable formula for sales ever invented.” ↩

  11. Poynter Institute. Birth of the Inverted Pyramid. ↩

  12. Ebbinghaus, H. (1885). Uber das Gedachtnis. Serial position effect validated by Glanzer & Cunitz (1966). ↩

  13. Ogilvy, D. (1983). Ogilvy on Advertising. Crown. ↩

  14. Loewenstein, G. (1994). The Psychology of Curiosity. Psychological Bulletin, 116(1), 75–98. ↩

  15. CoSchedule. What Separates the Best Headlines. Corpus analysis: 11–14 words, ~55 characters optimal. ↩

  16. Provost, G. (1985). 100 Ways to Improve Your Writing. Mentor/Penguin. ↩

  17. Bucket brigade technique. Direct-response tradition (Sugarman, Halbert). Adapted for web by Brian Dean (Backlinko). ↩

  18. Hootsuite (2025). How the LinkedIn Algorithm Works. AuthoredUp. LinkedIn Algorithm Data-Backed Facts. Author responses within 30 minutes: 64% more comments, 2.3x more views. ↩ ↩2

  19. Google. Search Quality Evaluator Guidelines. Structured content, search intent matching, E-E-A-T. ↩

  20. Edelman & LinkedIn (2024). B2B Thought Leadership Impact Study. 75% of decision-makers researched new products; 70% of C-suite reconsidered vendors. 2025 report: 95% of clients not actively seeking. ↩

  21. MarketingProfs. Content Marketing vs. Thought Leadership. ↩

  22. Schwartz, E. (1966). Breakthrough Advertising. Boardroom Books. Five levels: Unaware, Problem Aware, Solution Aware, Product Aware, Most Aware. ↩

Graph View

Backlinks