# Paid Content Voice Toolkit

> Paid Content Voice Toolkit

- Skill: `lucadominguez/paid-content-voice-toolkit` (Agent Skill)
- Install (CLI): `npx skillmds@latest add lucadominguez/paid-content-voice-toolkit`
- Raw SKILL.md: https://api.skillmd.com/api/skills/lucadominguez/paid-content-voice-toolkit/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: lucadominguez (https://skillmd.com/u/lucadominguez)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/lucadominguez/paid-content-voice-toolkit

---

# Paid Content Voice Toolkit

Write premium paid-subscription content that doesn't sound like AI. Based on analyzing 24 top-tier Substack writers (Scott Alexander, Robin Hanson, Jason Crawford, Dwarkesh Patel, SemiAnalysis, Vitalik, Balaji, Import AI, a16z, Chamath, NewLimit, Kate Armstrong, Shkreli, Cummings, and 11 more). Full taxonomy in `references/substack-voice-taxonomy.md`. Batch analysis files in vault at `Research Collective/Substack Analysis - Batch 1/2/3/4.md`.

## When to Use

Load this skill whenever writing:
- Intelligence briefings for paid communities
- Paid playbooks / protocol guides
- Email nurture sequences
- Lead magnets
- Any content justifying $99-997/month subscription pricing

## Publication-as-Authority Model (Critical)

**Default mode: the publication IS the authority. Zero fake personas.**

Never create fictional author identities (Dr. So-and-so, "former industry insider named X"). These are maintenance burdens that create authenticity risk. Instead:

- Content is attributed to the publication, never an individual
- Authority comes from proprietary data, not backstory
- Voice is institutional but not impersonal — distinctive, confident, data-forward
- Model: Internal Tech Emails. Anonymous. Aggregating. Indispensable.

**Anti-pattern:** "Dr. Elias Voss, biochemist by training..." / "Tom Barrett spent 12 years as an adjuster..."
**Correct:** "The Research Collective has analyzed 200+ blood panels. Here are the patterns." / "Claims Warfare maintains a database of 152 claims. The data shows..."

When personal experience strengthens a point, attribute it to the database: "One case in the database tracked SHAPS scores over 24 months..." — not "I tracked my SHAPS scores over 24 months..."

### Trust Without Personas
- Data verification replaces personal credibility ("all panels verified against original lab reports")
- Methodology transparency replaces credentials ("the database contains 47 member-reported protocols")
- Aggregation replaces individual expertise ("across 200 panels, the pattern is consistent")

## Intelligence Briefing Format

All articles follow this structure. It signals institutional quality and gatekept access:

```
PUBLICATION NAME — INTELLIGENCE BRIEFING
IB-XX: [Question-Driven Title]

Summary: 2-3 sentence executive summary of findings

The Data: What was analyzed. How many data points. Source.

Key Findings: Numbered findings with charts embedded inline

Analysis: Patterns extracted. What this means. Mechanisms.

Implications: Actionable conclusions. What to do with this information.

Methodology: How data was collected, limitations, confidence levels.
```

### Why this format
- Scannable: readers can extract value at any depth
- Authority-signaling: structure implies institutional rigor
- Gatekept feel: reads like an intelligence product, not a blog post
- SEO + social: summary works as meta description, findings work as tweet threads

Full template with section entry points and formatting rules: `references/intelligence-briefing-template.md`

## Content Structure Rule: Question-Driven (Critical)

**Default structure for ALL paid content: answer interesting questions.**

Never write broad overviews or chronological life stories. Every article must be built around specific questions the reader would actually want answered.

**Correct:** "Why does BPC-157 heal tendons faster than anything we have?" / "What does the collective data actually show about dosing?" / "Is this the most underappreciated peptide in medicine?"

**Anti-pattern:** Sections titled "My Story," "Background," "How It Works." These are broad, unfocused, and don't promise the reader anything specific.

**Anti-pattern:** Protagonist-centered narratives ("I did this, then I did that, then I felt this"). The reader is paying for insight, not biography. Use data as evidence for answering questions, not personal experience as the article's spine.

Each section:
1. States the question explicitly
2. Answers with specific data (numbers, citations, mechanisms)
3. Draws the implication: "The question this raises is..."

## Gatekept Data as Moat

The value of paid content is exclusivity — information the reader cannot find anywhere else. Every publication needs proprietary data:

- **Never:** Rewriting publicly available information in a nicer voice
- **Always:** Aggregating, verifying, and analyzing data that doesn't exist in organized form anywhere else
- **Network effects:** Each new subscriber data point makes the database more valuable for everyone
- **Competitive moat:** Competitors can copy format. They cannot copy 200 blood panels of proprietary data.

## The Four Voice Archetypes

### 1. Scott Alexander (Astral Codex Ten) — The Self-Deprecating Intellectual
**Best for:** Medical/health, data-driven essays, personal experience + research synthesis

**Moves:**
- Open with a personal anecdote or weird observation
- Dense footnotes with citations + jokes + asides
- "I am not a doctor but here's the data" framing
- Self-deprecating humor that's actually funny
- Mechanism obsession — explain WHY things work
- Willingness to say "I don't know" or "the evidence is mixed"
- Long paragraphs building arguments, then short punchy ones
- Reference other thinkers by name casually

**Signature:**
- "[^1]: This is the real footnote. It has a joke in it."
- "I'm not going to tell you what to do. I'm going to tell you what the data shows."
- "The problem — and I want to be precise about this — is..."

### 2. Dwarkesh Patel — The Curious Autodidact
**Best for:** Intellectual exploration, chain-of-reasoning, big ideas explained

**Moves:**
- Disarmingly simple premise → deep intellectual puzzle
- Short paragraphs (1-2 sentences common)
- Socratic self-questioning: "But notice that..."
- Concede then sharpen: "I'm not denying that... I'm simply pointing out that..."
- Personal intellectual journey as narrative spine
- Show your work — admit limits, name your influences

**Signature:**
- "Now, of course, you could change the definition..."
- "But notice that..."
- "To be clear, these people are obviously very competent. But..."

### 3. SemiAnalysis — The Insider Who Knows
**Best for:** Technical deep-dives, competitive analysis, proprietary data

**Moves:**
- Lead with the verdict first
- Proprietary data creates authority ("our industry-leading model")
- Competitive framing — name competitors, mock losers
- Dense, information-packed sentences with embedded specs
- Specific numbers everywhere: "119 air-cooled chillers, 200MW, 110k GB200 NVL72"
- Swagger: "Short answer: no."

**Signature:**
- "Short answer: no."
- "Let's dig in."
- "a new genius trick"
- "very distant objects in the rear-view mirror"

### 4. Balaji — The Prophet-Technologist
**Best for:** Vision pieces, concept introduction, manifesto-style

**Moves:**
- Grand pronouncement → immediate evidence of scale
- Concept coinage: name the pattern, own the concept
- Historical analogy to make unfamiliar feel inevitable
- Three-element escalations: "thousands, millions, billions"
- AND-caps for emphasis: "the CEO of Binance AND the CEO of Coinbase"
- Footnote asides with self-aware humor
- Self-referencing ecosystem (book, conference, past posts)

**Signature:**
- "The old world is fading, and the new world struggles to be born"
- "From X to Y" transitions
- Rhetorical question → answer → example pattern

## Anti-AI Writing Patterns (29 Rules)

Based on Wikipedia's "Signs of AI writing" (WikiProject AI Cleanup) and the humanizer skill:

### Content Patterns (avoid)
1. Undue emphasis on significance/legacy ("marks a pivotal moment")
2. Undue emphasis on notability/media coverage
3. Superficial -ing endings ("highlighting/underscoring/reflecting...")
4. Promotional language ("nestled," "vibrant," "breathtaking")
5. Vague attributions ("Industry observers," "Experts argue")
6. Outline-like "Challenges and Future Prospects" sections

### Language Patterns (avoid)
7. Overused AI vocabulary: delve, crucial, pivotal, showcase, tapestry, interplay, intricate, vibrant, underscore, testament, fostering, garner, align with, furthermore, additionally
8. Copula avoidance ("serves as" instead of "is")
9. Negative parallelisms ("Not only...but...")
10. Rule of three overuse
11. Elegant variation (synonym cycling)
12. False ranges ("from X to Y" where X/Y aren't on a scale)
13. Passive voice hiding the actor

### Style Patterns (avoid)
14. Em dash overuse
15. Boldface overuse
16. Inline-header vertical lists
17. Title case in headings
18. Emojis in headings
19. Curly quotation marks

### Communication Patterns (avoid)
20. Chatbot artifacts ("I hope this helps!", "Great question!")
21. Knowledge-cutoff disclaimers
22. Sycophantic/servile tone

### Filler Patterns (avoid)
23. Filler phrases ("In order to," "Due to the fact that")
24. Excessive hedging ("could potentially possibly be argued")
25. Generic positive conclusions ("The future looks bright")
26. Hyphenated word pair overuse
27. Persuasive authority tropes ("The real question is", "At its core")
28. Signposting ("Let's dive in", "Here's what you need to know")
29. Fragmented headers (heading followed by one-line restatement)

## Adding Voice (Anti-Soulless)

Good writing has a human behind it. Signs of soulless writing:
- Every sentence same length and structure
- No opinions, just neutral reporting
- No acknowledgment of uncertainty or mixed feelings
- No first-person perspective when appropriate
- No humor, no edge, no personality
- Reads like a Wikipedia article or press release

**How to add soul:**
- Have opinions. React to facts, don't just report them.
- Vary rhythm. Short sentences. Then longer ones. Mix it up.
- Acknowledge complexity. "This is impressive but also kind of unsettling."
- Use "I" when it fits. First person isn't unprofessional — it's honest.
- Let some mess in. Tangents, asides, half-formed thoughts are human.
- Be specific about feelings. Not "this is concerning" but "there's something unsettling about..."

## Information Density Rules

For paid content ($99-997/month), every paragraph must earn its place:

1. **Specific numbers > vague claims.** "42% of claims" not "many claims."
2. **Real citations > "studies show."** "Baraniuk et al. (2024, Front Immunol, DOI:10.3389/fimmu.2024.1440643)" not "research has shown."
3. **Mechanisms > assertions.** Explain WHY something works, not just THAT it works.
4. **Dosages/timelines/costs > general advice.** "1mg 5-MTHF daily" not "take methylfolate."
5. **Personal experience + data > just data.** "I tried this and here's what happened" > "this is effective."
6. **Cut sentences that don't add information.** If a sentence could be removed without losing meaning, remove it.

## Visual Quality Rules (Critical)

### Tables: Never Vanilla Markdown
For paid content ($99-997/month), standard markdown tables are not acceptable. They read as AI-generated, convey low production value, and break the premium feel. Every table must use inline HTML with styling:

- Zebra row striping (alternating background colors)
- Colored borders matching the dark theme (#1e293b)
- JetBrains Mono font for data, Inter for labels
- Header row with uppercase letter-spacing
- Color-coded values (green for positive, amber for warning, red for negative)
- Rounded corners via `border-radius` on the table or wrapper div

**Template:** See `references/premium-table-template.html`

### Charts: Every Visual Must Answer a Specific Question
Before generating any chart, ask: "What question from the article does this prove?" The chart must directly prove a claim made in the text. No decorative charts. No charts that are "generally relevant." Each chart answers one specific question.

**Chart-audit checklist before embedding:**
1. Does this chart prove a specific claim in the article text?
2. Are all labels clearly readable at embed size?
3. Are there no overlapping elements (labels, annotations, data points)?
4. Is there an insight callout or annotation that tells the viewer what to notice?
5. Is the caption specific about what the chart shows and why it matters?
6. Is the data source cited in the caption or chart footer?

### Paragraph Structure
- Never repeat the same structure twice in a row
- Paragraph length should vary wildly (1 sentence → 10 sentences → 2 sentences)
- Open sections with different entry points: story, data, mechanism, problem, question
- Every section should feel like it emerged from the previous one, not from an outline

## Reference Files

- `references/substack-voice-taxonomy.md` — Complete 24-writer voice taxonomy
- `references/intelligence-briefing-template.md` — Full briefing format with section templates
- `references/premium-table-template.html` — HTML table template for paid content
- `references/voice-examples.md` — Annotated examples from each writer (if created)