Editorial Standards

The Daily Wick sits at an unusual intersection: a daily newsletter drafted with help from a language model, then edited by a human. We think the AI assist makes a daily cadence sustainable; we don't think it removes our editorial responsibility. This page is what we promise.

How a post is made

1. Draft. Every weekday morning, an automated job calls Gemini 2.5 Flash with a long, opinionated prompt. The prompt fixes the day's format (a 2-minute micro on Mon/Wed/Fri; a 6-8-minute essay on Tue/Thu), enforces a voice, lists banned phrases, and supplies the current month's theme plus the last thirty topics so we don't repeat ourselves.
2. Validate. The draft is checked against our schema (title, tags, length, structure, real-looking source URLs) and against a banned-phrases list. Drafts that fail validation are rewritten; drafts that fail twice are skipped.
3. Curate. A human curator reads the result, fact-checks the claims and quotes, verifies the sources, and either approves the draft or pulls it.
4. Ship. The approved post is committed to the public repo, the site rebuilds, and the newsletter is delivered.

Sourcing

• Quotes are real. If we can't verify the attribution, the quote does not run. If a quote is paraphrased rather than verbatim, we say so ("after Lao Tzu", "in the spirit of").
• Statistics are cited. Numbers in posts link to the original paper, dataset, or report whenever the source is reachable.
• Further reading is curated. The "If this hooked you" block at the bottom of every post points to actual books, papers, talks, or essays — chosen as the next-best step, not as a citation. We avoid paywalled-only links.
• If we can't source it, we don't say it. An idea without a provenance is rewritten or cut.

Images

Every hero image comes from Unsplash or Pexels under their free licenses. The photographer's name and a link to their profile appear directly under every image. We do not generate AI imagery in v1.

Corrections

• Errors get corrected, not quietly edited. If we publish something inaccurate, we add a dated note ("Updated: 2026-06-04 to correct…") to the bottom of the post. We do not pretend it never happened.
• Found a mistake? Open an issue on GitHub or reply to any issue of the newsletter. We aim to correct within 24 hours.

What we won't do

• No fake authors. Posts are bylined by the publication and curated by a real person. We don't invent personas or pretend the AI is a human.
• No fabricated statistics. If a claim sounds too clean to be true, our process treats that as a flag, not a feature.
• No SEO listicles. The format is fixed, the length is short, and the goal is one useful idea — not engagement-bait.
• No surprise paid tier. If that ever changes, we'll say it out loud, on this page, with months of notice.

The disclosure on every post

Drafted with AI (Gemini 2.5 Flash) and curated by Rahul Karda. Read our editorial standards.

The whole pipeline — prompt, schema, validation, this page — is open-source on GitHub. If you want to see exactly how the sausage is made, the source is the record.