Most companies have real expertise and no way to publish it repeatably. They produce campaigns, not programs: a burst of posts, an empty topic list, silence. Duolingo solved the supply problem structurally — the questions users already send in become the topic queue, a named credentialed expert answers them on a fixed cadence, and the submission address at the bottom of every instalment refills the queue.
The second half is less obvious and buys more trust than the answers do: marking in public where the evidence stops. Defining the unit before publishing a count. Naming the motivated-reasoning risk before answering a contested question. Saying plainly that nobody knows.
This skill is a graph. Scan the descriptions, follow only the [[wikilinks]] you need.
The intake loop
- [[references/reader-questions-as-pre-validated-demand]] — an inbound question is demand someone already paid to express; run it verbatim.
- [[references/the-inbox-is-the-cta-and-the-backlog]] — the submission address is both the call to action and next week's editorial queue.
- [[references/credentialed-byline-and-named-outsiders]] — the named credential is the cheapest proxy for accuracy, and hiring an outsider proves you know your limits.
Turning the audience into a supply line
- [[references/promote-corrections-to-their-own-instalment]] — corrections are the highest-intent signal an audience produces; never patch them silently.
- [[references/solicit-tactics-credit-them-add-the-mechanism]] — users find affordances your telemetry cannot see; credit them by name and explain why each works.
Content only you can produce
- [[references/publish-your-aggregate-error-data]] — your error logs are a frequency table a competitor cannot replicate without your volume.
- [[references/one-worked-derivation-beats-ten-rules]] — one end-to-end trace shows rules composing, which a rule list structurally cannot.
Epistemic honesty as a credibility asset
- [[references/define-the-unit-and-recut-by-frequency]] — publish the counting rule with the count, and re-cut inventory statistics by usage before believing them.
- [[references/invert-a-confident-belief-with-a-primary-source]] — reversal travels, but only the artifact on the page makes it citable instead of contrarian.
Boundaries
This skill owns expertise-driven, question-fed publishing. It does not own content whose payload is your own usage data — the annual report, the state-by-state map, the spike post — and it does not own whether a number was validly measured in the first place. Pixel craft, motion, and design tokens route out of this pack entirely.
Sibling skills
- [[../duo-voice/SKILL]] — the register these instalments are written in, and how a brand voice survives a credentialed byline.
- [[../duo-growth/SKILL]] — what happens to a reversal or a data post once it leaves your site.
- [[../duo-culture/SKILL]] — candor as an operating norm; marking the limits of your evidence starts internally.
- [[../duo-experimentation/SKILL]] — the discipline behind any number you are about to publish.
Sources
- blog.duolingo.com — the Dear Duolingo advice column, crowdsourced-exceptions and study-tips posts, learner-error analyses, and the myth-debunking and etymology posts cited node by node.