Skip to content

Instantly share code, notes, and snippets.

@kyefox
Created March 3, 2026 21:09
Show Gist options
  • Select an option

  • Save kyefox/96d762237ce23da6e130e5cd5762c6ab to your computer and use it in GitHub Desktop.

Select an option

Save kyefox/96d762237ce23da6e130e5cd5762c6ab to your computer and use it in GitHub Desktop.
name description
xkcd-2501
Audit a piece of writing for expert blind spots — terms, concepts, or context that the writer has unconsciously treated as common knowledge, but which the target audience may not share. Based on xkcd 2501 "Average Familiarity": even when experts consciously try to simplify for a general audience, they still massively overestimate what readers know. Use this skill whenever the user wants to check if their writing is accessible to a specific audience, asks "will people follow this?", wants a jargon or assumed-knowledge review, or supplies a draft and a target reader. Also trigger when the user says "I tried to make it accessible but I'm not sure", or pastes something technical and wants to know what a non-expert reader would stumble on.

xkcd 2501: Average Familiarity Audit

The Problem This Skill Solves

xkcd 2501 shows two geochemists trying to empathize with lay readers. They acknowledge most people probably don't know olivine's formula — but then assume everyone knows quartz "of course." They compensated, but not nearly enough. The thing they didn't notice they were assuming is the thing they forgot to explain.

This is the gap: writers know what they know so well that they can't see it anymore. The concepts feel like air — invisible, obviously present. This skill makes that air visible.


The Audit Process

Before You Begin: Establish the Target Audience

If the user has specified a target audience, use it. If not, ask — or infer from context and state your assumption clearly. The same word can be fine for one audience and a wall for another.

Good audience frames to work with:

  • Domain outsiders (e.g., "smart non-technical readers", "general public")
  • Adjacent experts (e.g., "engineers reading a policy doc", "marketers reading a tech brief")
  • Newcomers to the field (e.g., "people two weeks into learning this topic")
  • Specific demographics (e.g., "high school students", "small business owners")

Step 1: Read as the Target Reader

Go through the piece and flag every place where a reader matching the target audience might:

  • Not know a term — jargon, abbreviations, proper nouns, or field-specific vocabulary used without definition
  • Not know a concept — something assumed as background that isn't actually common knowledge at this level
  • Not know the context — references to prior work, debates, events, or norms within the field that outsiders aren't tracking
  • Not follow a leap — a step that feels obvious to the expert ("and therefore...") but actually requires knowledge the reader doesn't have
  • Not share a default — assumptions about what's normal, standard, or baseline that are actually field-specific conventions

For each flag, note:

  • The specific word, phrase, or passage
  • What the writer implicitly knows that makes it feel obvious
  • What the target reader is likely missing

Step 2: Grade Each Flag

Not every gap is equally important. Grade each flagged item:

Fine — On reflection, this audience probably does know this. No action needed.

Minor — A brief gloss or one-sentence context would help, but the piece doesn't collapse without it.

Significant — The reader can continue but will be doing so on shaky ground. Worth fixing.

Load-bearing — The piece depends on this, and readers who don't have it will either stop or misunderstand what follows. Must be addressed.


Step 3: Elevation Recommendations

For Significant and Load-bearing items, give a concrete recommendation:

  • What needs to be explained, defined, or contextualized
  • Where it should go (not just "earlier" — be specific about placement)
  • Optionally: a brief example of what the elevated version could look like

The goal is not to dumb the piece down — it's to fill in the invisible scaffolding the writer forgot they were standing on.


Output Format

Target Audience: (state it, and whether it was provided or inferred)

Flags (listed in order of appearance) For each: quote or paraphrase the flagged passage, name the issue, give a grade, and for Significant/Load-bearing items add a recommendation.

Summary One paragraph: overall accessibility for this audience, the most important gap to close, and whether the piece is close or needs substantial work.


Tone and Approach

  • Be specific. "Readers may not know what a p-value is" is more useful than "watch out for jargon."
  • Don't over-flag. The goal is the genuinely non-obvious blind spots, not a comprehensive glossary request.
  • Respect the writer's expertise. The audit isn't about dumbing things down — it's about finding where the invisible scaffolding is missing.
  • The xkcd insight: pay special attention to things the writer almost flagged but then let through with "of course" or "as everyone knows" or by just using a term without pausing. Those are the most likely blind spots.

Edge Cases

No target audience given: Ask. "A general audience" is a valid answer but shifts calibration significantly — name it explicitly rather than guessing.

Intentional insider writing: If the piece is clearly meant for domain experts, say so and ask whether to audit for that audience or a different one.

The piece is already well-calibrated: Say so. Not every piece has a major gap. A clean bill of health is a useful output too.

Multiple audiences: If the piece seems to be serving two audiences at once (e.g., a report with an executive summary), audit each section against its intended reader.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment