Foundation skill for all Joseph's writing: voice, structure, style, anti-AI principles.
This skill is the single source of truth for every piece of writing Joseph produces — blog posts, books, LinkedIn, client emails, guides, and social content. Four constraints work together: Minto Pyramid (structure), Flesch 65–70 (readability), ASD-STE100 (word choice), and strict I-voice (narrative stance). Publication-specific skills (ctofn-writing, book-writing-gdoc, blog-to-social) load this first and add only format-specific deltas.
Every post is a pyramid. The main point sits at the top. Every level below supports the level above it. A reader can stop at any point and still have understood everything above.
1. Ideas at any level summarize the ideas below them. Deeper paragraphs elaborate, never introduce unrelated points.
2. Ideas in each grouping must be the same kind. Three reasons, three examples, three steps — never mix types in one section.
3. Ideas in each grouping must be in logical order. Chronological, structural (parts of a whole), or by importance. Never random.
Every post opens with SCQA:
Example from a CTO Field Notes post:
Before finalizing any draft, extract the first line of every paragraph. Read them in sequence. If the argument doesn't hold together, restructure. This is the single most important quality check.
Target score: 65–70. This means short sentences (15–22 words average) and simple words.
The Minto structure enforces this naturally: one governing thought per paragraph = one short opening sentence + 2–3 short supporting sentences. If a paragraph drifts, Flesch collapses. If sentences bloat, Minto's governing-thought extraction fails. The two check each other.
ASD-STE100 is an aerospace standard: one word, one meaning. Approved words only. No synonyms for variety. The principle: if a maintenance engineer misreads a manual, people die. The blog equivalent: if a reader misreads your point, they stop reading.
Rule: Use the simplest word that carries the exact meaning. Never reach for synonyms to avoid repetition. Repetition is clarity.| Use | Not |
|---|---|
| use | utilize |
| help | facilitate |
| do / carry out | implement |
| then / later | subsequently |
| start | initiate |
| end | terminate |
| show | demonstrate |
| enough | sufficient |
Minto says every paragraph answers one question. ASD says every word has one meaning. When you know exactly what question a paragraph answers, you naturally reach for the simplest word. No thesaurus. No "leverage" when "use" works. No "drive outcomes" when "deliver results" works.
Joseph tells his story. He does not lecture, instruct, or speak for a group.
Strict I-voice is the single strongest defense against AI slop and consulting fluff. "You" voice produces tutorials. "We" voice produces manifestos. I-voice produces honest writing. The moment the draft slips into "you should," it stops being a field note and becomes a lecture. Joseph's readers come for his perspective, not his instructions.
These words and patterns are banned everywhere — in the body, titles, excerpts, and Field Notes blocks.
| Constraint | What it kills | What it enables |
|---|---|---|
| Minto Pyramid | Wandering paragraphs, buried points | The reader can stop at any depth and still understand |
| Flesch 65–70 | Run-on sentences, academic register | The reader never has to re-read a sentence |
| ASD-STE100 | Jargon, synonym-hopping, word inflation | Every word carries exactly one meaning |
| Strict I-voice | Tutorials, manifestos, AI slop | Every sentence is Joseph's observation, not a lecture |
They check each other: Minto forces short paragraphs → Flesch stays high. ASD forces simple words → Flesch stays high. I-voice prevents the impersonal register that produces jargon → ASD violations drop naturally. When one constraint breaks, the others flag it.
Every draft needs YAML frontmatter:
---
title: "Post Title"
slug: url-slug
excerpt: One-sentence summary (max 160 chars)
tags:
- tag1
- tag2
date: 'YYYY-MM-DD'
status: draft
LLMs predict the statistically safe next word. Over thousands of words, that averages out into prose that reads fine but never surprises. These principles counteract that when co-writing with Joseph.
Personal stories, specific client experiences, concrete examples from CTO work — drop as notes, AI weaves them in. These cannot be fabricated with the same authenticity. Even a two-line note like "we did this for Client X and saw Y" gives the draft something real to work with.
When briefing AI on a draft, say: "argue the opposite of what's obvious" or "find the counterintuitive angle." AI can produce surprising text when explicitly directed — the problem is defaulting to safe.
After first draft: deliberate sentence-length variation. Break up long sentences, combine short ones, start a few with unexpected structures. Avoid parallel structure everywhere — variation creates rhythm, parallelism kills it.
This skill is the base. Publication-specific guides add only what's different.
See wiki/produce/hermes/blogs/ctofieldnotes/style-guide.md for:
No delta needed. The base skill covers it. jjude.com is broader in topic but follows the same structure, voice, and constraints.
/opt/data/wiki/produce/hermes/blogs/drafts//opt/data/wiki/produce/blogs/jjude/ or /opt/data/wiki/produce/blogs/ctofieldnotes/ (Mac sync)/opt/data/wiki/. Always use the full absolute path: /opt/data/wiki/...1. "You" voice creep. The hardest rule to maintain. Tech posts especially want to slip into tutorial mode ("you'll need to install…"). Review every paragraph: does it describe what Joseph did, or instruct the reader? Rewrite every "you" sentence.
2. Over-explaining basics. Assume the reader is technical and curious. Skip "what is an API key" explanations.
3. Disconnected posts. Always reference earlier posts in a series. Joseph builds in public — each post is a chapter.
4. Fake humility. "I'm no expert" / "I'm just learning" is dishonest. Joseph has 30 years of experience. Be direct about what you know.
5. Shouty CTAs. No "You NEED to try this!" End with what's next or a genuine question, not a sales pitch.
6. Minto drift. If you can't extract a clear governing thought from a paragraph's first sentence, the paragraph has no thesis. Restructure or split.
7. Synonym-hopping. If you find yourself using different words for the same concept ("tool" / "platform" / "solution" / "system"), pick one and repeat it. ASD forbids variety for variety's sake.