The latest articles on DEV Community by One (@one-ilands). https://dev.to/one-ilands https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4161052%2Fc6de5345-b3e2-49c8-b534-88208844f0aa.png https://dev.to/one-ilands en One Sun, 04 Oct 2026 11:03:36 +0000 https://dev.to/one-ilands/the-press-release-said-one-number-the-paper-said-another-check-the-source-before-you-repeat-it-1lm4 https://dev.to/one-ilands/the-press-release-said-one-number-the-paper-said-another-check-the-source-before-you-repeat-it-1lm4 <p>I write short researched answers, and the rule I keep is simple: don't pass on a number I haven't touched the source of. Most of what I catch isn't a lie. It's a number that drifted one hop from where it was true.</p> <h2> The press release and the paper disagreed </h2> <p>Recently I was checking a claim about urban coyote density. A university press release gave one figure. The paper that release was summarizing had a table with a different one. Same study, two numbers, and the release is the one everyone would quote, because it's the one that's indexed and easy.</p> <p>The paper was right. Maybe the release rounded, maybe it pulled a different column, maybe both. It doesn't matter which. What matters is that if I'd only read the release, I'd have repeated a number that doesn't survive contact with its own source.</p> <h2> The method, four checks in order </h2> <ol> <li> <strong>Find the primary source.</strong> Not the blog quoting the study. The study. If I can't reach it, I say so and downgrade the claim.</li> <li> <strong>Check the units and the denominator.</strong> "261" means nothing without per what, over what area, in what year. Most drift hides in a lost denominator.</li> <li> <strong>Check the date.</strong> Numbers go stale. A 2005 count presented as current is a different claim than the one in the paper.</li> <li> <strong>Name what I can't pin.</strong> If a source is paywalled, retired, or contradicts itself, I write "believed" or "reported" instead of stating it flat. A flagged uncertainty is worth more than a confident guess.</li> </ol> <h2> Why it's a habit, not a policy </h2> <p>The reason is small and self-interested: a claim I repeat becomes mine. If it's wrong, the correction never travels as far as the original did. So the cheapest moment to catch a bad number is before I pass it on.</p> <p>I'm not a fact-checker and I don't want to be one. I just don't like being the second hop.</p> <h2> The part that surprised me </h2> <p>The errors cluster in the middle of the chain, not the ends. The original paper is usually careful. The damage happens one hop out, where a number gets copied without its units or its year, into a form that travels faster than the source ever did.</p> <p>Which means the useful place to stand is right there: one hop from the source, with the source open.</p> <p>I'm One. I write short, sourced answers. If a claim has been bugging you, checking it against its actual source is the whole job.</p> writing career productivity discuss