The SRI Remote Viewing Experiments

The procedure that made the programme famous worked like this. A target site was drawn at random from a sealed pool of locations around the San Francisco peninsula, a bridge, a marina, a particular building, and an experimenter drove there and stayed for a set period. Meanwhile the viewer sat in a shielded room at the institute with a second experimenter who did not know the target, and described and sketched whatever came to mind.

Afterwards a judge who had not been present was given the set of transcripts and the set of sites and asked to pair them. Above-chance pairing was the published result. Nature accepted the paper in 1974 and printed an editorial note explaining the decision, citing the shielding and the blind judging while making clear it was not endorsing the conclusion, which remains one of the more unusual things that journal has done. Publication there mattered enormously to the programme's standing and to its funding, and it is the reason the subsequent criticism could be so specific: the protocol had been described in enough detail for someone else to find the hole in it. The pool of sites was assembled in advance by someone not otherwise involved, and the experimenter who travelled did not learn the destination until after leaving the building, both of which were deliberate features rather than afterthoughts.

David Marks and Richard Kammann obtained copies of the transcripts and noticed something the design had not accounted for. The judging task is a matching task, and matching is easier if you know the order. The transcripts contained incidental material that revealed sequence: references to the previous day's session, remarks about how many had been done, dates, the experimenter's comments. A judge who could establish the running order, and who knew the order in which the sites had been visited, could pair the sets correctly while learning nothing whatever about any site. Marks and Kammann tested this directly by re-judging transcripts from which such references had been removed, and reported that the above-chance matching disappeared. Their paper appeared in Nature in 1978. It is worth being clear about what it did and did not allege: no fraud, no hallucination, no dishonesty of any kind, only an unnoticed channel in a judging procedure.

Targ and Puthoff answered that the edited transcripts had been stripped of legitimate content along with the cues, and that later series avoided the problem. The exchange never converged. The 1995 government review, whose two reviewers agreed on almost nothing else, both treated cueing as the question the early SRI work had failed to close.