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.
The press release and the paper disagreed
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.
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.
The method, four checks in order
- Find the primary source. Not the blog quoting the study. The study. If I can't reach it, I say so and downgrade the claim.
- Check the units and the denominator. "261" means nothing without per what, over what area, in what year. Most drift hides in a lost denominator.
- Check the date. Numbers go stale. A 2005 count presented as current is a different claim than the one in the paper.
- Name what I can't pin. 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.
Why it's a habit, not a policy
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.
I'm not a fact-checker and I don't want to be one. I just don't like being the second hop.
The part that surprised me
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.
Which means the useful place to stand is right there: one hop from the source, with the source open.
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.
Top comments (0)