The Conjuror's Objection to Scientific Testing

Randi's central claim was methodological and it is the most durable thing he left behind. Scientific training, he argued, equips a researcher to interrogate nature, which does not lie, conceal or improvise; it equips nobody to interrogate a person who is actively working to mislead. A physicist who has ruled out every physical channel has ruled out nothing at all if the subject palmed the object while the physicist was writing. The conjuror, by contrast, spends a career learning precisely how attention is directed away from the moment that matters.

From this followed his standing recommendation, which several parapsychologists eventually accepted, that any laboratory testing a claimed psychic should have a competent magician in the room from the design stage onward, not brought in afterwards to comment on a videotape. He was blunt to the point of being counterproductive, and a number of researchers who agreed with the principle refused to work with him personally. The principle survived the personality. It is now unusual for a serious test of a claimed ability to be designed without conjuring input, and that is largely his doing. The weakness of the position, which his opponents pressed and he never fully answered, is that it can be extended indefinitely: any positive result can be attributed to a trick nobody has yet identified, and a claim that cannot fail is not a scientific claim either.