Editorial Policy

Most coverage of a news event tells you what happened. There is usually a second, more useful question — why does this work the way it does, and what does it actually change? — that goes unanswered because it takes longer to research and does not perform as well as a headline.

That second question is our entire remit. A typical DrafterDaily article takes a single story, identifies the mechanism underneath it, walks through it in plain language, gives concrete examples, and states honestly what the story does not prove. We publish a small number of pieces a day rather than a large number, because the format does not work at volume.

DrafterDaily runs an automated editorial pipeline. We would rather describe it accurately than let you infer it.

• Story selection. Two independent processes scan news, filings, papers and trend data each day and propose candidate stories. Before anything is written, each candidate is checked against everything we have published in the last fourteen days and rejected if it covers the same underlying story — even under a different headline or angle. We would rather publish fewer pieces than restate our own coverage to fill a slot.
• Drafting.Articles are drafted by AI against a fixed editorial brief: identify the mechanism, give concrete examples, state what the evidence does not establish. We think this is a legitimate use of the technology, in the same way that a newsroom's use of research databases or transcription software is legitimate.
• Verification. Load-bearing factual claims — figures, dates, quotes, study findings, named organisations — are checked against a source before publication. Where a claim cannot be verified, the article is held back rather than published with the claim in it, and the reason is logged. No process catches everything; when something does get through, our Corrections Policy is how we deal with it.
• Oversight and accountability. The pipeline is designed, monitored and adjusted by a human editor, who reviews what published each day and holds the corrections inbox. Where the system produces something that does not meet the standards on this page, it is corrected or removed and the process is changed so it does not recur. Automation is a production method here, not an excuse — everything published under our name is our responsibility, and we answer for it the same way any publication does.

What the process will not do:pad word count for search rankings; generate near-duplicate articles about the same story to occupy more search results; invent quotes, statistics, sources or studies; or present a summary of another outlet's reporting as our own reporting. Volume is not the goal — the format does not survive being scaled up, and we do not try.

• Where a primary source exists — the filing, the paper, the dataset, the transcript — we work from it rather than from coverage of it, and link to it.
• Where we rely on another outlet's reporting that we cannot independently confirm, we name that outlet in the body text and attribute the finding to them explicitly.
• We do not treat anonymous claims or unsourced social media posts as fact.
• Numbers are given with their source and their date. Where two credible sources disagree, we say so rather than picking the more dramatic figure.
• Links to sources are live in the article. If a source is paywalled we say so.

A lot of what we write starts from work another organisation did first — a wire service's document review, a regulator's filing, a company's earnings release, a research team's paper. We think that is normal and healthy, but it carries an obligation to be straight about it.

So, plainly: we are not a news-breaking operation. We do not have reporters, we do not work sources, and we do not claim scoops. What we do is take a story that already exists and work out the mechanism underneath it — how the thing actually functions, what the numbers do and do not support, and what follows from that.

When an article rests on someone else's reporting, that outlet is named in the body text and credited for the finding, not buried in a link. Our headline describes our analysis, not their scoop. If we ever find we have nothing to add beyond restating what another outlet already published, that is a piece that should not run — and we would rather publish less on a given day than pad the count.

We cover health and financial subjects because the mechanisms there are frequently misreported. We hold that coverage to additional rules:

• We explain published research and public data. We do not give medical, financial, legal or tax advice, and nothing on this site should be treated as such.
• We report effect sizes and sample sizes, not just conclusions, and we try to name the study's limitations.
• We distinguish clearly between a preprint, a peer-reviewed paper, a press release and a company claim.
• We do not write about individual securities, tokens or treatments in terms that could be read as a recommendation to buy, sell or take them.
• Where a topic warrants professional advice, we say so in the article.

• Advertising is sold separately from editorial. Advertisers have no input into what we cover, no advance sight of articles, and no ability to have coverage changed or removed.
• Affiliate relationships never determine coverage. We do not write an article because a product has an affiliate program, and we do not soften criticism of a product we earn from. See our Affiliate Disclosure.
• We do not accept payment for editorial coverage, sponsored posts presented as editorial, guest posts placed for backlinks, or paid link insertions into existing articles. Requests of this kind are declined.
• If we ever have a material interest in something we write about, it will be disclosed in the article itself.

We correct errors openly and quickly. The full process, including how to report a mistake, is in our Corrections Policy.

Analysis pieces are sometimes updated when significant new facts emerge. When we make a substantive update we change the update date on the article and note what changed. We do not silently rewrite the substance of a published piece.

Errors, disagreements and complaints all go to the same place: drafterdaily@gmail.com. Every message is read. Corrections are prioritised over everything else.