www.lesswrong.com/posts/iT4tDshfRGiqxkRpR/the-cia-believes-everything

The claim says all the research shows polygraphs are simply ineffective at telling truth from lies. That is not what the best evidence says.

The National Research Council's major review, The Polygraph and Lie Detection (National Academies, 2003), drew an important distinction between specific-incident polygraph tests and employee/preemployment screening. Its executive summary concluded that, for people like those studied in the literature, specific-incident polygraph tests can discriminate lying from truth telling at rates well above chance, though well below perfection. The same review also concluded that screening uses are much less reliable and that polygraph accuracy is insufficient to justify reliance on employee security screening.

So the strongest accurate version is something like: polygraphs have limited, above-chance accuracy in some event-specific contexts, but are too error-prone for broad employee screening. Saying that all research shows they are simply ineffective at telling truth from lies erases that distinction and contradicts the National Academies' findings.

The sentence attributes the conclusion to the CIA, but the source material indicates the quoted conclusion came from Stanford Research Institute (SRI) researchers.

A CIA FOIA page for the underlying document identifies it as "FINAL REPORT OF PARAPSYCHOLOGY STUDIES STANFORD RESEARCH INSTITUTE" and says "Dr. H. Puthoff and Mr. R. Targ will present their final report of parapsychology studies completed at Stanford Research Institute (SRI)". In the body of that same document, the famous line about Geller's success appears within the SRI report itself.

Another CIA-hosted document likewise describes Targ and Puthoff as "the two SRI scientists who studied ... Uri Geller." So the evidence supports saying that SRI researchers reached or wrote that conclusion in a report later held in CIA files—not that the CIA itself, as an agency, authored the conclusion.