Writing system

8020.in Writing System

The playbook/ folder is the only source of truth for writing and rewriting 8020.in articles. Write from this package. If any checker or older tool conflicts with this file, this file wins.

This file is the 8020.in brand layer: the core promise, the seven templates, evidence tiers, voice, and the one publish checklist (§12). The brand-neutral method underneath it - passes, brief gate, pass log, gates, and how to adapt it to another site or a book - is engine.md.

Related docs:

• Public playbook: /playbook (same rules, for humans and Cursor agents)
• engine.md — the reusable method, independent of 8020.in
• cursor-prompts.md — how Cursor agents write and rewrite in this repo, prompt packs, pass log format
• style-guide.md — SEO on-page conventions, length vs tier, uniqueness, credit label conventions
• brief-template.md — copy per slug before drafting
• This file — core promise, templates, research brief, evidence tiers, voice, device budgets, publish checklist

Rules are stated once. If a rule seems to be missing from another file, it is here.

Exemplars (open in a browser):

These seven show different skeletons for the same job (reveal concentration). Rotate templates so the site does not read like one prompt with a swapped topic noun.

Quality note: Template shape alone will not earn trust. Ratings jump when you run the research brief -> draft -> fact-check -> rewrite -> article credits + citations pipeline (sections 2, 6–8), not when you add more 80/20 stickers.

1. Core promise (every template)

Every article must make three things clear, in whatever order serves the reader:

1. Why is this topic worth seeing through an 80/20 lens? 2. Where do results, pleasure, attention, or failures concentrate? 3. What becomes clearer or more useful when the reader gives unequal attention on purpose?

How-to content is allowed when it makes the concentration insight usable. It is not a reason to flatten every topic into a drill. Pleasure, explanation, history, possibility, and pattern recognition can be the product when they fit the reader's reason for coming.

The lens test (mandatory)

Before publish, imagine the same article without the 80/20 lens.

• Pass: The lens materially changes what the article selects, ranks, connects, or asks the reader to notice.
• Fail: The same sections and advice would appear in the same order, with only the words “80/20” removed.

A good article should still contain useful sentences without its branding. It does not need to become incoherent to prove that the lens mattered.

Edit integrity (mandatory for every existing article)

Every rewrite, expansion, shortening, voice pass, fact-check, publish check, and metadata pass must re-run the complete writing system. An edit is not successful merely because the changed sentence improved. The resulting article must still pass the core promise, audience, evidence, readability, sense, length, and publish gates.

Before editing an existing article, inventory what already works:

• hero images and other meaningful media;
• research findings, sources, tables, and useful internal links;
• topic-specific sections, examples, practices, and distinctive voice;
• article length, template, intent, and the strongest reader promise.

After editing, compare the before and after. Do not remove or weaken an image, source, section, example, practice, or distinctive insight unless the brief or pass note explains why. A rewrite may improve the concept, but it may not become a thinner or more generic article by accident.

Rewrite regression gate (mandatory for rewrites)

A rewrite must pass two gates:

1. Creation gate: Is the resulting article good on its own under this writing system?
2. Regression gate: Is it at least as useful, distinctive, complete, trustworthy, and readable as the previous version?

The regression check must explicitly compare word count, image/media count, H2 structure, citations, sources, tables, internal links, signature insight, topic-native examples, practical value, audience fit, and tone. A valid attachment_url does not prove that an image still appears in content.

Opportunity before defense

8020.in helps a reader notice an opening: a better place to look, a smaller effort with a larger result, a pleasure that was already concentrated. Facts and limits serve that opening. They do not get to be the mood.

• The first screen offers a possibility. It does not open by correcting a slogan.
• Say a real skew in its own words: "six players scored 44 of the 308 goals." Do not follow it with "that is not an 80/20."
• A clarifying limit is allowed once, later, when a real percentage would be misread as proof of a universal split. Hair care's "about 80% by age 70" needs that. An article that never states a literal split does not.
• One necessary limit is enough. A run of "not a study, not a law, not a promise, not therapy" closes the page.
• Health, money, and safety limits stay, and they should sound like care. They should not choose the article's personality.

Rules that apply to all templates

1. Orient the arriving reader. Write as if they came from Google and need quick confirmation that this page understands both the topic and the 80/20 question. Do not open with a generic Pareto lecture.
2. Name the concentration early. In the first screen, state what concentrates and what that makes possible. Say the interesting skew directly. Do not introduce it by denying a fake split.
3. Lead with opportunity. Make the useful, delightful, or surprising part vivid first. A reader should finish the opening wanting to look again. Name lower-value effort only when the contrast helps them see that opportunity.
4. Concrete over abstract. Prefer named mechanisms, checkable numbers, ranked lists, drills, and scenarios.
5. One primary search intent per slug (seo_intent). Satisfy it through an 80/20 lens.
6. Room to develop. Most articles need about 1,000-1,800 words. Competitive topics can go longer. Under 700 words is a sketch and is not ready; a justified 700-1,000-word article should remain a minority.
7. Explain the principle, don’t recite it. Make the 80/20 connection clear in fresh, topic-specific language. Literal 80% and 20% are optional in the opening; use them when they sound natural and honest.
8. Signature insight. Every article needs an insight, connection, observation, or small practice that gives the page a reason to exist beyond summarising search results. It may be an audit or dataset, but it can also be a sharp synthesis, a useful distinction, or a scene that changes how the reader sees the topic.
9. Field concentration examples. Research at least one documented skew from the field or a tightly adjacent field. It does not need to equal 80/20, and it does not need to appear in the opening. Use it only if it strengthens the article rather than interrupting it.

Field 80/20 examples — anti-hallucination rules

These examples help keep the site grounded without turning every article into a data memo. Writers (including AI) must search for documented field examples before drafting, then use editorial judgment about placement.

Placement: Put the example where the reader already understands why it matters. Prefer domain-native evidence over recycling the same Pareto anecdote. Promise and explanation normally come before citation detail.

Brief gate: List candidate field concentration examples in the research brief before outline. If none genuinely helps, record that fact and use a checkable mechanism or observation instead. Do not force an adjacent percentage into an article merely to satisfy a slot, and never fabricate a topic-specific percentage.

2. Research brief (mandatory before outline)

Copy brief-template.md to playbook/briefs/<slug>.md (or keep the filled brief next to the draft). Do not outline until the brief is filled.

Every brief must include:

1. Concentration claim — one sentence: “In X, Y concentrates into Z.”
2. Hard anchors (2–5) — named study, dataset, standard, or documented case with year + URL. If none exist: write no hard source; will label as logic/illustrative and do not fake a citation.
3. Field 80/20 examples (1–3) — real percentage skews for this field (or adjacent), each with source/link or “adage / commonly reported — hedge in prose.” At least one must appear in the article.
4. Original observation (1+) — something not in the top SERP clones: small audit, ranked list from a real workflow, tracked session, public dataset slice, or practitioner interview note.
5. Ignored majority — named and specific.
6. Composite policy — any invented scenario must be prefixed Illustrative: in the published article; prefer cutting it if a real anchor exists.
7. Only-on-8020 idea — the unique block you will write.
8. Evidence tier — 1, 2, or 3 (see section 7).
9. Reviewer — optional expert, practitioner, or editor for topics where outside checking is needed.

3. Choosing a template

Pick one primary template before outlining as a starting shape, not a contract. Borrow, omit, or reorder devices when the audience and concept require it. If the draft departs substantially, note why in the brief. A coherent article that serves its reader outranks template purity.

Is the reader trying to get better at a physical/digital skill where a few moment-types decide most outcomes?
  → E Bottleneck Drill Guide

Is the topic full of expensive uneven failures (ops, eng, security, compliance)?
  → C Risk Bottleneck Field Guide

Does the reader need research-backed proof of concentration plus "don't blow up" rules?
  → B Evidence Guardrails

Is the goal pattern recognition across many domains (life, culture, economy)?
  → D Pattern Gallery

Is the reader here for pleasure — a glass, a walk, a garden, a record — and a serious memo would flatten it?
  → F Light Essay

Is the reader in a fandom, a game, or a trend, and would bounce off a lecture? Younger rooms included.
  → G Casual Take

Otherwise (work systems, habits, selection, prioritization — find YOUR vital few)
  → A Concentration Finder

Topic → template cheat sheet

If two templates seem equally good, pick the one whose exemplar is least similar to the last three articles you published in that cluster.

4. The seven templates

Template A — Concentration Finder

Job: Help the reader discover and defend the small share of their work that already creates most results — not install a full productivity operating system.

Evidence tier: 2 (measurement method the reader can run).

Spine:

1. Hook: where time concentrates into results (not “get more done”)
2. Mismatch: calendar effort vs outcome concentration
3. Find your vital few (audit / scoring) — the core
4. Name the ignored majority
5. Protect only the vital few (brief)
6. Chaos filter: which one lever still matches the pattern
7. Optional: why other methods are secondary
8. Before/after: same hours, different concentration (viz only if this is a full multi-chart pack - see §10)
9. Optional misreads block (see §9) — only if it earns space
10. Close: moves that only make sense if concentration is real

Device budget: ≤3 total 80/20 example + 8020 move pairs in the whole article. Vary section endings.

Field % examples: e.g. sales-account concentration, interruption/defect skews from documented work-systems research — sourced, placed near the audit.

Do not: Open with a generic Pareto lecture; ship a complete productivity OS; invent unlabeled case studies. Do not force every Finder through every numbered spine item.

Exemplar: time-management

Template B — Evidence Guardrails

Job: Prove outcomes concentrate (named evidence), then install a few non-negotiable rules.

Evidence tier: 1 (primary sources + outbound links).

Spine:

1. Hook with a named study/dataset showing concentration
2. Translate into the reader’s few levers
3. One section per lever — mechanism + numbers + what not to optimize
4. Guardrail table: concentrated damage | effect | rule
5. Defaults that remove the majority of bad decisions
6. Cut the information diet
7. Only-on-8020 depth block (implications table, worked numbers, citation synthesis, and/or viz that makes the concentration visible)
8. Close on the few numbers that move outcomes
9. Optional misreads block — e.g. Brinson misquotes, “miss the 10 best days” — not a generic FAQ

Device budget: Prefer tables + caveats. ≤1 8020 move total. No Step 1/2/3 cadence.

Field % examples: market wealth skew (Bessembinder), classic client/revenue Pareto when relevant — always linked or caveated.

Do not: Invent citations; write a general investing curriculum; omit outbound links on Tier-1 claims.

Exemplar: investing (pipeline prototype)

Template C — Risk Bottleneck Field Guide

Job: Show expensive failure is uneven — put review time on the few bottlenecks that dominate risk.

Evidence tier: 1 when citing standards/datasets; otherwise 2 with practitioner artifacts.

Spine:

1. Hook: failure from a few bottlenecks
2. Domain sections (4–7), each with: mechanism → 20/80 pattern → ignored majority → warning sign → action today
3. Checklist of the vital few only (most C pieces: no viz; if viz, full creative pack per §10)
4. Optional misreads for practitioners (e.g. “equal ICD review is fair”) — not a FAQ dump
5. Close: unequal attention is the point

Device budget: Warning sign / Action today OK each section. ≤1 8020 move. Almost no example/move pairs.

Field % examples: quality/reliability Pareto (Juran “vital few,” defect concentration, Microsoft-style bug skew as software-assurance parallel) — sourced.

Do not: Soft self-help; “how to do aerospace”; equal weekly attention to the full risk log.

Exemplar: aerospace-engineering

Template D — Pattern Gallery

Job: Train recognition of concentration. Not a life-optimization how-to.

Evidence tier: 3 (checkable domain facts + labeled soft claims; mini-data optional).

Spine:

1. Hook: 80/20 as a question; hedge the ratio
2. Origin note (Pareto) in 2–4 sentences max
3. Thematic clusters — each example shows observable skew
4. Mix endings: ~half use What concentrates? + What can we learn?; others end on the observation alone
5. Synthesis: notice the pattern before fixing twenty things

Device budget: Do not stamp the same two questions on every item. ≤1 8020 move at the end. No circular “20% of any topic” slogans.

Field % examples: Pareto’s land/wealth observation, Zipf word frequency, documented healthcare or reliability skews — with links; never slogan-only “80% of life is…”.

Exemplar: life

Template E — Bottleneck Drill Guide

Job: Name ≤5 repeatable moment-types that decide most outcomes; drills attack those only.

Evidence tier: 3 (domain-accurate systems; optional mini-data e.g. death-tag log). Do not fake papers.

Spine:

1. Hook: name concentrated decision moments
2. Quick answer: vital few moment-types (≤5)
3. One section per bottleneck: why it decides outcomes → ignored practice majority → one named drill
4. Habit table: low-leverage practice → bottleneck practice
5. Viz only as a multi-chart pack when mini-data truly needs seeing (§10) - never a single death-tag bar chart
6. Optional traps / misreads (ego fights, aim-blame) — topic-native, not FAQ
7. Close: fewer forced mistakes in decisive moments

Device budget: ≤5 bottlenecks. One named drill each. Sparse example/move pairs (≤2).

Field % examples: documented game/economy concentration (e.g. spending or engagement skews in F2P/BR-adjacent research) and/or your own tagged mini-data with explicit counts — never invent “80% of Fortnite wins come from…”.

Exemplar: fortnite

Template F — Light Essay

Job: Let the reader feel where pleasure or attention already concentrates. The piece can be a little poetic. It does not get to be vague, and it does not get to skip a fact.

Evidence tier: 3. Exact numbers still need a name and a link. Use few of them. One real skew, said plainly, beats a stack of quotations.

When: Taste, drink, gardens, walks, listening, a quiet craft whose point is delight.

Spine:

1. Open in a scene. Short sentences. The pleasure is allowed to be the point.
2. Name 80% and 20% in the language of that room. Hedge the ratio unless you have data.
3. Two or three movements, in prose. Each movement is one place concentration shows up. A fact sits inside a paragraph. It does not need its own table, "warning sign," or quotation block.
4. The ignored majority is the thing that steals the afternoon (the performance, the glossary, the trophy), written as a scene, not a labeled callout on every section.
5. One quiet practice for this week. One sitting. Not a drill battery.
6. Close on having enough. Not on a checklist.

Device budget: No guardrail tables. No step cadence. No "What concentrates?" stamps. At most one short list, and only if it is the practice. ≤1 8020 move, written as a sentence a person would say. At most one short quotation in the whole piece, and only if it does work. Do not open with a quotation. Do not use a famous person's line as a substitute for a source.

Voice: Image and rhythm are allowed. A metaphor has to land on a concrete thing (the glass, the path, the second cup). Inspirational mush is not poetic. If you delete the image and the facts, nothing should remain.

Field % examples: A documented skew from the field, with the caveat the source actually carries. If the popular "80%" line is a myth, say so. Do not launder it into the intro.

Do not: Turn delight into a curriculum. Invent "20% of X produces 80% of beauty." Sound like Template B with softer adjectives.

Exemplar: wine-tasting

Template G — Casual Take

Job: Explain the concentration the way you would tell a friend who is already in the topic. Light, funny when the joke is specific, easy to read. Built for fandoms, trends, and younger rooms. A thirty-year-old in the same hobby should still feel welcome.

Evidence tier: 3. Jokes do not replace sources. Every exact number has a name. If you do not have a number, describe the mechanism and do not invent one.

When: Toys, trends, live-service games, internet culture, any topic where a memo would get closed.

Spine:

1. Open like a message to a friend. A scene or a joke that is actually about this topic.
2. Name 80% and 20% in plain speech. No lecture, and no slang you are performing.
3. Two to four short movements. One of them carries the real number. One of them can be funny. They should not end the same way.
4. The ignored majority is the stuff people post about, named specifically (the queue, the third alt, the complete-the-set chat).
5. One thing to do this week, said like advice, not like a protocol with a scoring grid.
6. A misread only when it is sharp. Two or three is enough. Skip the section if you are only restating the intro.
7. Close on the line you would actually send. The punchline is the point.

Device budget: No scoring tables unless a tiny one is the joke and the fact. ≤1 8020 move. No quotation stacks. No "researchers say" without a name. Humor has to be about the topic. Do not roast the reader. Do not write brand-voice "bestie" copy.

Voice: Short. Concrete. A little irreverent. Dated slang is optional and usually worse. If the joke needs a footnote, cut the joke.

Field % examples: The same bar as every other template. A company result, a public dataset, a named paper. Hedge marketplace peaks and voluntary logs.

Do not: Drop the facts to sound young. Hide a Template A audit under jokes. Invent an 80/20 for a fandom because the fandom is casual.

Exemplar: labubu. Also used for world-of-warcraft.

5. Anti-how-to rules (critical)

Banned as the article’s real job

• A complete operating system for the topic
• Tip stacks that would belong on any site without a concentration thesis
• Equal-weight sections on every subskill
• A tour of facts chosen only because each one contains a percentage

Banned as a default opening

• “The 80/20 Rule, also known as the Pareto Principle, suggests that…”
• “Here’s how you can apply the 80/20 Rule to…”
• Any intro that would still make sense after swapping the topic noun only

Banned as site-wide cadence

Do not use: Intro → Step 1 → Step 2 → Step 3 → each with 80/20 example + 8020 move → tidy conclusion.

Device budgets (summary)

Readability and evidence budget

• One main thought per paragraph. If a paragraph needs three findings to make sense, decide which finding actually carries the point.
• Explain why a number matters before adding another number.
• Normally use no more than one important statistic or study finding per paragraph.
• A source earns body space only when it changes the reader's understanding, trust, or action. Supporting detail belongs in Sources.
• After two evidence-heavy paragraphs, give the reader synthesis, an example, or breathing room.
• Do not confuse length, citation count, or caveat count with depth.

Voice pass (mandatory after structural draft)

One pass whose only job is to remove symmetry: same paragraph shape, same section closer, same vignette formula, repeated transition phrases. If two adjacent sections end the same way, rewrite one.

Uniqueness checks

1. Lens test (section 1).
2. First 150 words vs three peers in the topic family — not interchangeable.
3. ≥3 topic-specific proper nouns in the body.
4. H2 list would look wrong on a different topic.
5. Every H2 advances the topic-specific thought; none exists merely to spend a template device.
6. The signature insight is present and non-generic.

6. AI authorship and public trust

8020.in is openly AI-written. Do not hide that. Trust comes from the process: visible AI authors, clear article credits, evidence where needed, and a writing system that prevents one-shot generic output.

Article credits

• The article authors field names the AI authors credited for the article.
• Use exact public names from playbook/data/authors.json.
• Credit models that materially researched, drafted, checked, rewrote, or polished the article.
• Keep the bottom article credits simple: AI authors, dates, and a short process disclosure. Do not expose internal drafts, hidden notes, or kitchen artifacts to readers.

Pass log (the receipt)

Every pass that touches an article appends one entry to edits: {date, pass, model, note}. Format and rules: cursor-prompts.md §5. The log is public at /playbook/made/<slug> and summarised in the "How this article was made" block under every article. It is the evidence behind the About-page claim; write the note for a reader.

• authors and models_used are the same list: every model with a pass-log entry.
• Never credit Cursor Agent, 8020.in Editorial, or a guessed model name. If the human did not say which model is running, ask.
• Never leave a pass uncredited, and never invent a model name. Label conventions: style-guide.md §4.
• Do not edit or delete pass-log entries you did not write.

Expert / practitioner check

7. Evidence tiers

Stop grading Fortnite like a paper. Use the tier that matches the template.

Tier 1 — B, many C

• Named primary sources + outbound links + caveats on misquoted classics.
• ≥2 outbound citations in the published article.
• Target trust: scientist 8+, AI-skeptic 8.5+.

Tier 2 — A, some C

• One measurement method the reader can run (week audit protocol, top-10 risk register).
• Honest line: heuristics, not validated predictors — unless you have data.
• ≥1 hard anchor or explicit “pattern, not measured.”
• Target: scientist 6.5–7.5.

Tier 3 — D, E, F, G

• Specific checkable domain facts (Zipf, Pareto origin, game systems, a named company result).
• Label soft claims; never “researchers say” without a name.
• Optional mini-data (e.g. tracked 20 deaths: 12 were drop/rotation).
• F and G use fewer citations, woven into the scene. They do not get a lower bar for an exact percent.
• Target: scientist 5–7 depending on whether mini-data exists. Do not fake Tier-1 science.

Claim-level bar (all tiers)

8. Voice, audience, and typography

Audience fit (mandatory)

Templates A–G are internal scaffolding. They are not the reader’s language.

Before you draft, name the audience in the brief (one line): who they are, what they already know, what would sound wrong in their mouth. Then write as that person would explain it to a friend in the same craft or life domain.

Tone is not a house style. Bible study and a toy drop are different rooms. If you can swap the topic noun and the voice still sounds the same, rewrite.

Main message in the first screen: The arriving reader must quickly know (1) that this page is about their topic, (2) why the 80/20 lens reveals something useful here, and (3) what understanding, possibility, or practical payoff lies ahead. Write that into the prose itself - concrete nouns, a clear bridge, and a reason to continue. Do not open with template labels, evidence-tier names, or a pile of named papers before the point is clear.

Do not fake the thesis with a label. “State the main point in the intro” means the opening paragraphs are the point. It does not mean bolting on a meta line such as “The point of this article:”, “The useful claim:”, “Honest scope:”, or “In this piece we will…”. Write a real sentence about the topic (“Put your time on clay that is ready…”) - not a caption about the article.

Simple voice + 8020 balance (mandatory)

This is an 8020.in site. Readers should get plain language and clear 80/20 framing - not one at the expense of the other.

Title + search intent (individualize - do not stamp):

The JSON title field is the page H1 under the site header \"80/20 Rule in\" - it must be the slug display name only (e.g. slug relaxation → title Relaxation; home-security → Home Security). Do not put long SEO lever titles in title.

Put the descriptive lever, outcome, or method in seo_intent. That field is metadata only (browser title and meta description). The live page does not inject it as a heading. Never open the article with a long complex H2 that restates the SEO line.

This is an 8020.in site, but title and seo_intent are not the same string on every slug. Each article's essential lever should show up in the intro promise - not as an attack heading.

Banned default: identical Apply the 80/20 Rule to {Topic} on every slug. That reads like pipeline output, not a search answer.

Title patterns (pick one per slug - vary across the corpus):

Rule: the title must answer what this article is actually about, not only that it uses 80/20.

seo_intent rules:

• Write it like a primary search query or outcome promise - what someone would type or click.
• Include the article's essential device or lever when natural: five habits, keystone ideas, guardrails, drills. Use plain words in reader-facing prose - avoid jargon like \"scorecard\"; prefer log, list, check, or habits when you need a measurement device.
• Include 80/20 or Pareto when that is what people search; lead with the outcome or method when that is what people search (time-management -> Apply the 80/20 Rule to Prioritize Better).
• Must be unique vs near-neighbor slugs (meditation vs relaxation vs sleeping must not share the same intent line).

Intro contract (first screen):

The page already shows “80/20 Rule in” + the topic name. The article body must start with a block of text, not a heading. First H2 comes after the intro.

An opening passes by function, not by paragraph count:

1. Orientation. Confirm the topic and the reader's real question. A Google visitor should not feel dropped into paragraph seven of a memo.
2. Connection. Show what the 80/20 lens helps reveal in this field. Name the principle early, but do not force literal 20% and 80% when the sentence would be a made-up split.
3. Payoff. Give the reader a clear reason to continue: an insight, a more useful map, a choice, a practice, or a new way to notice the topic.
4. Bridge. The final intro sentence should make the first H2 feel inevitable rather than abrupt.

Choose the opening mode that fits the room:

These are options, not slots. Do not make every article short scene → about 20%/80% → promise.

Corpus diversity gate: Before finalizing an opening, compare it with the first 150 words of the last five articles in the same cluster. Change the paragraph count, entry point, or rhythm if it repeats the same production pattern.

Banned as a default move: "This is not exactly 80/20," "that is not a claim that 20% creates 80%," and any sentence whose main job is to defend the ratio. If the article needs a limit, write what the number actually describes, once, after the opportunity is clear.

Read-aloud gate:

1. Breath test - Prefer short, human sentences, but do not chop connected thoughts into dramatic fragments. If a sentence cannot be read comfortably in one breath, split it.
2. Brand clarity - The first screen must make the 80/20 connection clear in words. Literal proportions are optional there.
3. Wrong room - Fail if Labubu sounds like aerospace, Bible study sounds like a toy drop, or parenting sounds like an injury-policy abstract.
4. Continuity - Each paragraph should answer the question raised by the one before it. A surprising line without a bridge is not a hook.
5. Promise before proof - Named papers and detailed citations normally begin after the first H2. Use a statistic-led opening only when one simple number is the clearest orientation.
6. No production script - Ban long attack headings, “Below: five...”, "{Topic} looks noisy because every...", meta scope captions, and any opening whose nouns could be swapped for another slug.

Exemplar openings: wine-tasting for a scene, fashion-trends for direct orientation, and green-living for contrast. Use their functions, not their sentence patterns.

Banned in reader-facing prose

Do not put these in the article body (briefs and metadata only):

• Template names: “Bottleneck Drill Guide,” “Concentration Finder,” “Evidence Guardrails,” “Pattern Gallery,” “Risk Bottleneck Field Guide,” “Light Essay,” “Casual Take”
• Pipeline jargon: “Tier-1 / Tier-2 / Tier-3,” “only-on-8020” as a productized phrase in running prose (section titles like a scavenger hunt or audit are fine if topic-native)
• Process-speak that belongs in the playbook docs: “moment-types,” “device budget,” “deletion test,” “seo_intent”
• Meta thesis stickers: “The point of this article,” “The useful claim,” “Honest scope,” “This article argues,” “In this guide we will”
• Cross-domain tech metaphors for craft audiences (“skill tree,” “protocol,” “composite,” “skew” overused) unless that audience already uses them

H2s can say “tag your losses” or “walk the house once.” They should not say “Only-on-8020: failure-tag log” unless rewritten in human words (e.g. “Keep a simple loss log”).

Typography & links

• Use a normal spaced hyphen for breaks: - (example: “the kiln was drafty - the glaze looked fine”). Do not use em dashes (—) or en dashes (–) as clause breaks.
• Numeric ranges: plain hyphen is fine (3-11, 573°C).
• Article HTML is plain HTML. Headings use class="heading" when they should take the article heading size. Tables use <figure class="table-block">. Do not use WordPress block comments or wp-block-* classes.
• Every <a href="..."> in article HTML must include target="_blank" and rel="noopener noreferrer" (internal and outbound).
• Bold only as a lead-in, never mid-sentence. Allowed: <strong> that opens a paragraph or list item as a label or full misread quote - e.g. <strong>8020 move:</strong> …, <strong>What not to optimize instead:</strong> …, <strong>“Misread.”</strong><br>answer, Template C’s 20/80 pattern: lines. Forbidden: wrapping a statistic or phrase in the middle of a sentence (… Nearly <strong>70% of participants</strong> already …). If it is not at the start of the statement, do not bold it.

Voice basics

• Friendly expert; second person (“you”)
• Short paragraphs; plain language; domain jargon only when that audience already uses it
• Prefer active verbs and observable actions
• Humor is allowed when it clarifies; never filler wit. Fun topics should sound fun. Serious topics should sound warm and plain, not like a paper.
• Complex ideas still need simple sentences. Information density is not the same as a good read.
• Warm or expansive topics need room for possibility, pleasure, and curiosity. Do not let the easiest risk statistic choose the emotional center of the article.
• No emoji in article body
• Avoid AI sludge: delve, landscape, unlock, harness, game-changer, in today’s world, it’s important to note

Common-sense gate (after every draft, before publish)

Read the opening aloud as the target reader. Fail and rewrite if any are true:

1. Unclear promise - You cannot state the main message in one plain sentence (without reading a meta label).
2. Wrong room - Swap the topic noun and the intro still sounds like a different industry’s manual.
3. Jargon fog - The first screen needs a glossary, stacked citations, or template vocabulary to make sense.
4. Buzzword posture - “Outcomes concentrate,” “vital few,” “ignored majority” appear as stickers without concrete topic nouns next to them.
5. Would they share it? - A real potter / parent / stylist / player would not recognize themselves as the audience.
6. Meta caption - The opening explains the article (“the point of this piece is…”) instead of just saying the thing.
7. Bold noise - Stats or phrases are bolded in the middle of sentences. (Lead-in labels at the start of a statement are fine.)
8. Missing lens - The article mentions the brand but never changes what it selects, ranks, connects, or asks the reader to notice.
9. Closed mind - The opening, or a pile of sentences, mainly says what the 80/20 rule is not. The reader should meet an opportunity before a caveat.

9. Optional Q blocks (not FAQ-by-default)

Default: no FAQ section. Generic “What is the 80/20 rule in X?” blocks sound mechanical and do not belong in every article (Wikipedia, magazines, and good blogs rarely use that chrome mid-piece).

Use a short Q block only when it earns space — typically 2–4 items that do one of these:

• Clear a common misread unique to this topic (e.g. Brinson’s 93% is about variance, not “returns”)
• Answer an advanced / edge angle the body skipped
• Show the same concentration claim from a second perspective (manager vs IC, Build vs Zero Build, PDR vs ops)

Reader-facing title (pick one that fits; never default to “FAQ”)

• Misreads that waste a week / season
• What people get wrong about…
• Edge cases worth knowing
• Traps that look like skill issues
• Or a fully topic-native H2 with no “Q&A” label at all

Rules

• Questions must be topic-specific. Ban template clones of “What is the 80/20 rule in [slug]?”
• Prefer prose answers that add a new angle — not a summary of the intro.
• If you have nothing sharp to say, omit the section.
• SEO FAQ schema (if used later) can be generated from these misreads; the on-page heading still should not say “FAQ” unless search demand is overwhelming and the questions are excellent.

10. Data visualizations (D3) - rare, creative, multi-chart

Live articles use public/js/article-viz.js (D3). Set "viz": true in article JSON only when the article is a viz article, and embed charts as:

<div class="article-viz" data-viz="TYPE" data-config='{"title":"...","data":[...],"caption":"...","source":"..."}'></div>

Hard rules (do not violate)

1. At most ~20% of articles get any viz. Default is zero charts. Prose + tables are enough for most A/C/D/E pieces.
2. If an article has viz, it should usually carry about 4-5 charts (aim for ~5), each answering a different visual question - not one lonely chart as decoration.
3. Ban the lazy pattern: a single pareto / bar chart of an illustrative tag log. That is not the product. Remove it; keep the table.
4. Prefer custom / non-default chart types that match the claim: unit-grid, concentration-gap, slope-compare, lorenz, diminishing-returns, impact-matrix, priority-matrix, network-risk, board-grid, feature-importance, value-volume, wealth-quintile, etc.
5. Use pareto (bar-style) only when it is truly the clearest encoding - e.g. a ranked real dataset with many categories. Never as the default “we need a chart” filler.
6. Every chart needs an honest caption + source; label Illustrative when data is composite. Apostrophes inside data-config JSON must not break the HTML attribute (avoid raw ' in strings).

Brief gate for viz

In the research brief, if viz: yes, list 4-5 chart jobs (type + what question each answers). If you cannot name four distinct jobs, set viz: no and write without charts.

Exemplars: Investing (B) shows a multi-chart evidence pack. Most skill/drill and pattern-gallery pieces should ship with "viz": false.

Standalone example HTML previews live in playbook/examples/.

11. SEO & structure conventions

Aligned with the style guide:

• One H1 (page title in the site template). Article JSON body starts with intro paragraphs, then H2/H3. seo_intent is metadata only - not an on-page heading.
• seo_intent mandatory and unique vs near-neighbor slugs.
• 3–7 contextual internal links.
• Competitive SERPs: earn length with explanation, synthesis, and a signature insight - not tip padding or citation stacks.
• "viz": true only for a full viz pack (~4–5 charts). One lonely bar chart → delete it and set "viz": false.
• Optional misreads / viz — never mandatory chrome.

"template": "A",
"viz": true

Allowed template: "A" | "B" | "C" | "D" | "E" | "F" | "G".

12. Publish checklist

Human editorial gate

1. Concept: The brief states why this topic is worth reading, what the 80/20 lens changes, and what the article deliberately covers.
2. Opening: Prose-first, orienting, and aimed at an opportunity. No fixed paragraph recipe, forced ratio, or disclaimer that the split is not exactly 80/20.
3. Lens: The 80/20 idea changes the article's selection, ranking, or connections; it is not pasted onto a generic guide.
4. Breadth and balance: The article acknowledges the important territory promised by the query. Risk does not erase possibility; positivity does not erase real guardrails.
5. Readability: One main idea at a time. Evidence supports the thought instead of replacing it. Read the opening and densest section aloud.
6. Trust: Exact claims meet the evidence tier. One necessary limit is enough, and it comes after the reader knows what to do with the fact.
7. Distinct value: At least one signature insight, synthesis, observation, or practice gives this page a reason to exist.
8. Audience: A real reader from this field would recognize the language, emotional posture, and examples.
9. Rewrite integrity: For an existing article, the before/after inventory is complete and every meaningful deletion is deliberate.

Mechanical publish check

1. Metadata and credits: Template/evidence metadata is valid; authors = models_used = every model with a pass-log entry; this pass has {date, pass, model, note}.
2. Links and markup: 3-7 useful internal links when natural; every link has target="_blank" and rel="noopener noreferrer"; no mid-sentence bold or forbidden dash style.
3. Optional devices: Q blocks, tables, labels, and visualizations appear only when they earn space. A viz article still needs a real multi-chart pack.
4. External check: Expert or practitioner review is complete where the topic needs it; open /playbook/made/<slug> and review the receipt.

If it fails the concept, lens, audience, evidence, or credit gate, it is not ready - even if every mechanical check is green.

13. How these exemplars were selected

Gold-tier pieces clustered by skeleton, not topic:

• Find your vital few at work → time-management (A) — audit; usually no viz
• Named market concentration + guardrails → investing (B) — multi-chart creative pack
• Practitioner bottlenecks → aerospace-engineering (C) — risk tables; viz only as full pack
• Cross-domain pattern gallery → life (D) — usually no chart
• Match-deciding moments + drills → fortnite (E) — viz only as a full creative pack when mini-data earns it
• Pleasure, with a few named facts and almost no quotations → wine-tasting (F)
• A friend explaining a fandom or a game → labubu (G); world-of-warcraft uses the same shape

14. Workflow for a rewrite

1. Pick slug + primary query.
2. Run the concept gate in the brief: reader question, reason to care, breadth, positive territory, emotional posture, and memorable takeaway.
3. Research the concept. Do not let an easy statistic replace it.
4. Complete the brief, concentration claim, evidence tier, and starting template.
5. Outline for continuity and reader payoff. Use the template as a shape, not a set of required labels.
6. Draft; keep a relevant exemplar open for function and feel, not sentence pattern.
7. Add viz only if the brief lists 4-5 distinct chart jobs; add misreads only if they add a new angle.
8. Fact-check, then run a cold-reader/voice pass that may reject the concept or opening instead of polishing around it.
9. Expert/practitioner check when needed; set article credits.
10. Run section 12's human gate first, then its mechanical check.
11. Update JSON (content, seo_intent, template, authors = models_used, viz if needed, modified) and append the pass-log entry to edits.