COVID-19 has revealed many unrecognised problems. I have become aware that medical surveys are becoming less scientific. This perhaps always has been the case with non-medical surveys, notably illustrated by contradictory partisan pre-election polls in the USA, some of which used landline phone calls (landline users do not reflect the general population). Electronic, mostly social media–based, surveys are often worthless. Survey purposes should be explicit, the target population should be explicit, the sampling procedures should be explicit and robust, the questions should be ruthlessly focused, the follow-up procedures should be explicit and appropriate, the results should be capable of …