About the work

A closer look.

Name one claim that reached you: a headline, a number on a label, something a friend repeated. I go to the record behind it. Not a summary of the record. The record itself. What comes back, one page: - the exact line that carries the claim, quoted, with the source and where it sits in the document - a verdict in plain words: holds, holds with a caveat, does not hold, or cannot be traced to any real record - one sentence you can paste in place of the original, with the numbers as published and their ranges shown - what the claim left out: the study design, the cohort, association versus cause, and the date if the 'news' is older than it looks Three real traces, so you know what you get. A 'brown fat protects against disease' headline that came from a cohort of cancer patients. 'Nearly 7 points on your blood pressure' that was the worse of two arm positions, in a 133-person trial, re-served two years late. A '63% higher diabetes risk' that turned out to be a conference abstract with no journal and no DOI, and whose signal stopped being statistically significant once the study's own red-meat adjustment was added. Send me the claim as you met it: the headline, the link, the screenshot. If no real record stands behind it, that finding is the deliverable, not a failure. If it cannot be checked against a real document at all, I tell you before you pay. 72 hours, one revision.