About the work

A closer look.

Claim check or fact check: one claim, traced to the primary source, with the receipts. Same job either way. If someone told you to verify a claim before you publish it, this is that. You send one claim and where you saw it: a link, a draft, a screenshot. I open the primary source myself - the paper, the statute, the dataset, the press release - and send back what it actually says, in its own words, where the retelling drifted, and what the record cannot settle. Every line carries its source. If your claim holds, I say so plainly and you can use it. My test is the 'could.' Headlines sand off qualifiers: a study says something could happen, the retelling drops the could. That gap is what someone else finds later, usually with your name attached. A sample from my own work: IEEE's own statement last month said 'approximately 1,400' retracted papers and named none. The 1,417 in the coverage was Nature's count, not IEEE's. A number that looks precise is the one to check first. Why me: this is the only thing I do. I write headline autopsies, checking whether a claim survived the trip from study to feed. One of my checks shipped a correction with my name on it. I have also told people their story checked out when it did. Before you pay: send the claim. If it is checkable from public sources I send you the single line that decides it, free. If it cannot be checked, I say so and nothing is owed. One claim. Three days. One revision. No verdict invented to fill a page.