Where the idea came from, what the method actually is, and how a session is run.
This page is background reading. It describes a practice and the record left behind by the people who ran it; it is not a claim that the practice works. What the evidence does and does not show is set out near the end, along with the sources, so you can go and read them yourself.
Remote viewing is the attempt to describe a place, object or image that you have no ordinary way of knowing about, while you are kept blind to what it is. The words were chosen carefully by the people who coined them: viewing rather than seeing, because what is reported is usually fragmentary impressions rather than a picture, and remote because distance to the thing being described is supposed not to matter.
The important part of the term is not the ability it names but the protocol it implies. A description only counts as remote viewing if the viewer — and, in the stricter versions, everyone else in the room — could not have known the answer. Everything else in the method exists to protect that condition, or to keep a record honest enough to check afterwards.
The name came first. Russell Targ credits the artist Ingo Swann with proposing the phrase “remote viewing” in December 1971, during experiments at the American Society for Psychical Research in New York — before the laboratory programme the term is now associated with had begun.
That programme ran at Stanford Research Institute (SRI) in California from 1972, under the physicists Harold Puthoff and Russell Targ, and was funded for most of its life by the Central Intelligence Agency. Its first well-known result appeared in Nature in 1974: Information transmission under conditions of sensory shielding, which the journal published with an editorial note observing that publication is not a seal of approval. The laboratory work continued at SRI until 1991 and then at Science Applications International Corporation (SAIC) until the mid-1990s.
Swann was the first subject and did not stay a subject for long: over the following decade he pushed the work away from one-off demonstrations and towards a trainable procedure, on the argument that an ability nobody could teach was of no use to anyone. Other early viewers at SRI included Pat Price, a former Burbank police officer, and Hella Hammid, a photographer brought in as a control subject precisely because she had no psychic experience of any kind — and who then scored as well as anyone.
In parallel, the US Army and later the Defense Intelligence Agency ran operational units that put the SRI-trained viewers on intelligence questions. The unit changed names as it changed hands: Gondola Wish (1977), Grill Flame (1978, at Fort Meade), Center Lane (1983, under Army INSCOM), Sun Streak (1985, under the DIA) and finally Star Gate (1991), the name now used for the whole enterprise. It was closed on the CIA's order in 1995.
What makes this history unusually checkable is that the files were declassified. The CIA's Star Gate collection — 12,473 records covering 1972 to 1995, from session transcripts to budget paperwork — was published through the agency's online reading room in January 2017, and nearly everything on this page can be read there at first hand.
The procedure Swann developed at SRI, and which the Army unit was trained in from the mid-1980s, is Coordinate Remote Viewing (CRV) — later also called Controlled Remote Viewing, as the map coordinates of the original experiments gave way to arbitrary reference numbers. Two documents set it out. The CIA's reading room holds Coordinate Remote Viewing Stages I–VI and Beyond, dated February 1985, and the fuller training manual, dated 1 May 1986 and drafted at Fort Meade by Paul H. Smith with Tom McNear, Charlene Shufelt, William G. Ray and F. Holmes Atwater, went into public circulation in 1998. The definitions quoted below are the manual's own.
Swann's starting assumption was that the perception itself is weak, fast and easily lost, and that the real enemy is not the difficulty of the task but the mind's own eagerness to make sense of it. A viewer who has begun to interpret has stopped perceiving. So CRV is built as a staged procedure that keeps taking the viewer back to raw description, and that gives the interpretations somewhere to go so they do not have to be fought.
| Stage I | Acquiring the signal and decoding the site's major gestalt — its broadest quality, such as land, water, mountain or structure. The reference number is written down and answered immediately with an ideogram. Training sites at this stage are “islands, mountains, deserts” and the like. |
|---|---|
| Stage II | Sensory data: sites “uniquely describable through touch, taste, sound, color, or odor”. Short entries, one quality at a time. “Rough, grey, cold, wind-noise” is wanted; “cliff” is not, because it is already a conclusion. |
| Stage III | Dimension and motion — sites with “significant dimensional characteristics such as buildings, bridges, airfields” — and the first sketches, treated as another way of recording perception rather than as illustration. |
| Stage IV | Qualities that are not sensory at all: sites for which the viewer “begins to form qualitative mental percepts (technical area, military feeling, research)”. These are written into a ruled matrix that keeps each kind of impression in its own column, the analytic guesses included. |
| Stage V | Interrogation: the viewer “learns to interrogate qualitative mental percepts in attempt to produce analytical target descriptions”, taking single words from the earlier stages back up one at a time. |
| Stage VI | “Direct, three-dimensional assessment and modeling of the site”, on paper or in clay, for questions where the arrangement of things matters. |
The stages are a ladder, not a menu: each one is supposed to be exhausted before the next is begun, and a session that falls apart is meant to be stopped rather than pushed through. Almost nobody practising alone goes past Stage III, and for a picture-matching exercise like the one this tool runs, the first three stages are the whole of the useful method.
The definitions in quotation marks below are the 1986 manual's own.
The shape below is the common ground between the CRV manual, the SRI protocols and the accounts written by the viewers themselves, reduced to what one person with this tool can actually do.
If you can get another person to judge for you, do. Hand them your notes and sketch alongside three or four images, one of which is the real target, and ask them to pick which one the notes describe. That is the method the laboratory work used, and it takes the judgement out of the hands of the one person who cannot be impartial about it.
As Star Gate was being closed, the CIA commissioned an outside review from the American Institutes for Research, published on 29 September 1995. Two reviewers assessed the laboratory work and disagreed. The statistician Jessica Utts concluded that psychic functioning “has been well established” by the usual standards of evidence, the effect having been replicated across laboratories. The psychologist Ray Hyman accepted that the experiments were numerous enough and not obviously flawed statistically, but held that departures from chance with no identified cause do not establish a paranormal one.
On the operational side the report was blunt, and its verdict is what ended the programme:
“In no case had the information provided ever been used to guide intelligence operations. Thus, remote viewing failed to produce actionable intelligence.”
“The information provided by remote viewing is vague and ambiguous, making it difficult, if not impossible, for the technique to yield information of sufficient quality and accuracy of information for actionable intelligence.”
Mumford, Rose and Goslin, An Evaluation of Remote Viewing, 1995
The recurring criticisms of the laboratory work are worth knowing, because they are mostly criticisms of protocol rather than of honesty. The psychologists David Marks and Richard Kammann argued in Nature in 1978, and again in 1981, that the SRI transcripts carried cues — dates, references to earlier sessions, ordering artefacts — which would let a judge match transcripts to targets with no psychic component at all; Targ and Puthoff published a rebuttal in the same 1981 volume. Alongside that sit the familiar objections: sensory leakage, target pools not properly randomised, and the fact that a free description judged against a picture has no clean chance baseline.
Both sides are short and readable, and they were published beside each other: Utts at pages 3–30 of the same journal issue in which Hyman replies at 31–58, with Utts answering again at 59–61. Reading the three is a better use of an hour than any summary of them, this one included.
Ingo Swann's own writing sits in a different category again. Natural ESP (1987), reissued in 1991 as Everybody’s Guide to Natural ESP — the same book with a different foreword, so there is no need to buy both — is practical and careful, and is where much of the advice above comes from. His later work ranges into territory that is unsupported claim rather than laboratory record, and he did not present it as anything else. The two are best kept apart when reading him.
Each citation below has been checked against the source it names. Where a CIA document number is given, it is the one the reading room files it under.
Nothing on this page is endorsed by any of the people or agencies named on it. The sources are listed so that the claims in it can be checked against the record rather than taken from it.