Editorial Guidelines
How we write here, and why.
The rule
Nobody who comes to this site should feel unwelcome because they didn't know something coming in.
That's it. That's the editorial guideline. Everything else follows from it.
What that means in practice
If you use a word that a high school graduate wouldn't know, you explain it. Right there. In the same sentence or the next one. Not in a footnote. Not in a glossary. Not behind a link that says "learn more." Right there, where the reader needs it, the moment they need it.
We don't assume you know what lambda calculus is. We don't assume you know what SSO stands for. We don't assume you've read Church's 1936 paper or Kahneman's book or the man page (manual documentation) for chmod (a command that sets file permissions). If those things matter to the point we're making, we explain them. If they don't matter to the point, we don't mention them.
This is not dumbing down. This is the opposite. Dumbing down is leaving things out because you think the reader can't handle them. We put everything in — the lambda calculus, the formal proofs, the technical architecture — and then we make sure every reader can follow every step. That's harder to write. It's worth it.
No walls
A wall is a term, a symbol, or a concept dropped on the page without explanation. It divides readers into two groups: those who already know it and sail past, and those who don't and feel like the page wasn't written for them.
Walls are everywhere in technical writing. They're so common that people who build them don't notice. A security engineer writes "RBAC" (Role-Based Access Control) without a second thought. A data engineer writes "idempotent" (meaning you can run it twice and nothing changes). A mathematician writes ∃ (a backwards E that means "there exists") and assumes you took the same classes they did. Each one is a small door that closes on someone.
On this site, we don't build walls. When a concept needs a technical term, we introduce the term and its meaning together:
Wall
"The system uses SSO with externalbrowser authentication."
No wall
"The system uses SSO — Single Sign-On, the 'Log in with Google' pattern where one login gets you into multiple systems."
Wall
λgross.(¬∃n : |gross| = n)
"For the function 'gross,' there exists no n such that the cardinality of its meanings equals n."
No wall
λgross.(¬∃n : |gross| = n)
Then a table that walks through every symbol: λ means given. ¬ means not. ∃ means there exists. The vertical bars mean count of. Every piece, in plain English, before you ask the reader to hold the whole expression in their head.
The first version tells you what it means if you already speak the language. The second version teaches you the language. We write the second version.
The test
Before publishing any page, ask one question: could someone with a high school diploma and no technical background read this page start to finish without feeling lost?
Not "could they understand every nuance." Not "could they implement what's described." Could they read it without hitting a wall that makes them feel like the page isn't for them?
If the answer is no, find the wall and tear it down.
What we do explain
- Every symbol in a formal expression
- Every acronym, the first time it appears on a page
- Every technical term that isn't common English
- Every concept that requires domain knowledge
- Why something matters, not just what it is
What we don't explain
- Common English words
- Arithmetic
- How to use a web browser
- Things the reader just learned three paragraphs ago on the same page
The line is: would a thoughtful person who graduated high school and reads the news feel comfortable here? We're not writing for children. We're writing for adults who happen to have different expertise than we do. Respect them enough to include them.
Why this matters for AI content
AI-generated prose has a specific failure mode: it writes for the author, not the reader. It mirrors the vocabulary of whoever prompted it. If Bill says "presigned URL" (a temporary download link), Claude writes "presigned URL" and never thinks to explain it, because Claude is pattern-matching on Bill's language, not modeling a first-time reader's confusion.
This is a form of false termination (the AI stopping as if the job is done when it actually isn't). The output looks correct — grammatical, well-structured, confident — but it has silently excluded part of the audience. The AI doesn't know it's building walls because it can't see the reader. The human has to catch it.
That's why these guidelines exist. The AI writes fast. The human checks for walls. Neither works alone.
The deeper point
This site is called YOU++. The thesis is that anyone can do more with AI as a thinking partner. Anyone. Not anyone with a CS degree. Not anyone who's read the right papers. Not anyone who already knows what lambda means. Anyone.
If the site itself is only readable by people who already know this stuff, then the thesis is a lie. The content would say "everyone is welcome" while the language says "but only if you're already in the club."
We don't do that. The door is open. The page is readable. The wall isn't there.
Disclosure: This page was generated by Claude (Anthropic) under Bill's direction. The editorial standard it describes was applied to this page while writing it, which is either good practice or recursion depending on your perspective.