A pharmaceutical company spends a billion dollars and a decade building a molecule. Before it can sell a single dose, the law requires it to prove that molecule can beat a pressed disc of sugar. The sugar wins regularly. Then it gets thrown in the bin.

The placebo effect survives five skeptic's objections, each one dismantled by published, peer-reviewed science, and then the conclusion itself gets pulled out from under the listener. The predictable 30% placebo response that shows up in trials is the floor. The ceiling has never been measured, because the clinical trial was built to average and isolate, the two operations that erase the force of a single believing human being.

Four anomalies show what leaks through when the instrument breaks: tumors melting like snowballs, dopamine flooding a damaged Parkinson's brain on a PET scan, a  ...