popular-science

Popular science - write an accessible technical article

Turn technical work into an article a curious, busy non-specialist reads to the end and is glad they did. This skill is the writer; its critic is the popular-science adversary in devils-advocate:adversarial-review - draft here, review there, revise.

• Register - a smart reader in another field: respect their intelligence, never their prior knowledge
• The bar - the best of the field (Quanta, Scientific American, NYT/FT visual journalism, Our World in Data); write to it, do not settle below it
• The spine - hook → nut graf → ladder of abstraction → sourced claims → arc-back kicker; every section below is one vertebra
• Full craft canon - best practices, anti-patterns, visual standards and named reference sources live in references/craft-canon.md (the single source of truth, shared with the popular-science adversary); read it before drafting
• Structure templates - pick one before drafting: references/structures.md (WSJ formula, inverted pyramid, ladder/narrative arc, braided, martini glass)
• Exemplar teardowns - five best-in-class articles torn down on the spine: examples/teardowns.md

When to use

• Writing up an experiment, result, finding, design or README for a non-specialist or a public audience
• Rewriting a dense internal doc into a readable article, blog post, explainer or story
• Not for: an internal spec or reference doc (that is Modus Secundis / technical-documentation), a journal entry (the journal plugin), or the SOTA conclusions doc (the hypothesis skill)

The spine - structure every article on this

• Hook - the opening earns the second sentence: a scene, a number, a paradox, a person. A reader decides in two sentences
• Nut graf - early, not always right after the lede: the story in a nutshell - what it is, and why it matters to the reader
• Ladder of abstraction - climb both ways: scenic detail (show - zoom in on the concrete thing) and summary meaning (tell - zoom out to what it means). Good writing constantly ascends and descends it
• Sourced claims - a bare claim is weak; the named research team and the hard figure sell it. Pair every empirical assertion with (Author, year) or an honest "our model finds", and with the demonstrated number, never an adjective ("2.0% vs 6.2% across 476 couples", not "about a third")
• Concreteness - every abstraction gets one anchor: an image, a number the reader can see, a person making a choice
• One idea per sentence; short paragraphs - one beat each; break the wall of text (white space is oxygen on a screen)
• Visuals - see below; a figure earns its space or it is chartjunk
• The kicker - end on a line that lands: arc back to the opening image or question (callback / bookend, full circle with a twist), then conclusions, then next steps where the work continues. The best endings echo the beginning in an essential but surprising way