Today's essay: the imagination that lets you build the tool is the same one that watches it fail at 2 AM — fully rendered, before a line is written. When it fails in your head, it fails as a death. Here are 1,000 real serious drug-safety reports from the FDA. First the version you fear. Then the version that actually gets logged.
Built on today's essay: “The Same Muscle” (clinicians.build, Jul 12 2026)
Data: FDA Adverse Event Reporting System (FAERS), 2026 Q1 intake extract · outcome mix of serious-coded reports · queried in MIMI Labs
Fourteen percent is not comfort — a death is a death, and 30,927 of them sit in a single quarter's extract. The point is calibration, not reassurance. But read the field honestly: this interactive deliberately selects the 218,055 cases carrying a coded serious outcome; another 179,169 cases in the quarterly intake had no outcome code and are outside the grid. Spontaneous reporting also captures serious and fatal events more reliably than mild or uneventful ones. The ledger you'd reach for to steady yourself carries the same catastrophe bias as the 2 AM imagination.