STARGATE: The Army’s Psychic Spy Program
$20 million. 23 years. Three agencies. "Viewer" Pat Price saw Soviet submarines in real time. Then the program was declassified, mostly buried, and the question was never really asked: was it real?
Classification: GOVERNMENT — DECLASSIFIED | Confidence: PRIMARY DOCUMENTATION (CIA FOIA 2017)
“I Remember When the Government Admitted It Spied With Psychics”
I remember when the US government admitted, in plain English, that it had been training psychics to spy on the Soviet Union. November 1995. The front page of The Washington Post. “Pentagon Studies Paranormal Activity.”
The program had been called SCANATE, then SUN STREAK, then GRILL FLAME, then CENTER LANE, then STAR GATE — five code names over 23 years — funded by the CIA, the Army, and the DIA. North of $20 million spent. More than 40 people trained as “remote viewers.” And then, in 1995, the whole thing was declassified — except the CIA kept the bulk of the archive sealed. That archive didn’t come out until 2017, after a FOIA request dumped 12,000 pages to the CIA’s Reading Room. By then, most people had forgotten the original story. What was in the record is below. What wasn’t in the record is the more interesting question.
The 23-Year Arc — SCANATE to GRILL FLAME to STAR GATE
The program began in 1972 at the Stanford Research Institute (SRI) in Menlo Park — a lab that had been CIA-funded on side-projects since the late 1960s. Two physicists, Hal Puthoff and Russell Targ, noticed an anomaly in some equipment readings and began to wonder whether human perception could pick up information the instruments couldn’t. The CIA’s Office of Technical Service put up $50,000 to test the idea. The program was code-named SCANATE.
Within two years, funding shifted. The Army Intelligence and Security Command (INSCOM) took over and renamed the work GRILL FLAME. INSCOM ran it from Fort Meade, Maryland — the same installation that houses the NSA. Sessions were conducted at SRI in California and, later, at Science Applications International Corporation (SAIC) near Washington. By the mid-1980s the DIA had begun co-funding. The name changed twice more — CENTER LANE, then STAR GATE. Five titles, four sets of bureaucrats, one underlying methodology. The total cost, by the CIA’s own 1995 accounting, ran between $20 and $25 million across 23 years.
The Viewers — Puthoff, Swann, Price, McMoneagle
Four names dominate the public record. None were carnival performers — they were mostly former intelligence officers, retired military, and civilian contractors with consistent statistical results in controlled lab sessions.
Hal Puthoff — the physicist who co-founded the program at SRI. Not a viewer; he ran the lab.
Ingo Swann — the first viewer. A New York artist who, in 1971, wrote Puthoff a letter claiming he could influence the output of a shielded magnetometer. Puthoff tested the claim. The instruments, against all known physics, responded.
Pat Price — a retired California police commissioner. In a 1986 session, Price claimed to have “viewed” a Soviet submarine pen in Vladivostok and produced sketches. Months later, a US satellite passed over; the imagery reportedly matched Price’s description, including details with no public source. The session was kept classified until 1995.
Joe McMoneagle — Army Intelligence sergeant, the Army’s first acknowledged remote viewer. His work on the 1979 Iranian hostage crisis was used to brief a presidential commission. He wrote a 2002 memoir titled The Stargate Project, the program’s most common name. Puthoff went on to inhabit the same post-Cold-War fringe as the Project Blue Book advocates — a community that has never fully decided whether Stargate was a psyop, a real program, or both.
The Protocols — How The Army Made Remote Viewing Operational
The most clearly documented unit of the program is the protocol. By the late 1970s the methodology had been standardized. The viewer sat in a shielded room. A target — a coordinate, a person, or an object — was selected by an “experimenter” unknown to the viewer. The viewer was given only the coordinate (latitude and longitude, a six-digit grid, or a name). They described, sketched, answered questions. A second “monitor” with no knowledge of the target was present to prevent visual cues. Sessions were audio-recorded, timestamped, chained, archived.
Each target was judged blind by a separate evaluator against multiple decoys on a five-point scale. The standard cycle was 30 days.
The intent was to make a psi phenomenon look, on paper, like any other controlled intelligence method — chain of custody, defined input, defined output, independent judge. The protocol was designed by people who understood that credibility depended on producing a paper trail an outside auditor could follow. That the GAO did eventually follow it is one of the program’s more interesting facts. Compare this to MKUltra, where the program’s own record-keeping was destroyed in 1973 by direct order.
The Hits — Cases That Worked
In 1984, during planning for a US hostage-rescue attempt in Tehran, Pat Price was tasked with locating the hostages and identifying safe houses. His sessions reportedly identified the building where the hostages were held — one the rescue team had not previously known about. The mission failed for unrelated reasons (a helicopter collision in the Iranian desert).
In 1979, McMoneagle was tasked with viewing a Soviet Tu-22M “Backfire” facility. McMoneagle sketched an aircraft on a triangular airfield. Satellite imagery later confirmed a matching facility. (Critics note that Tu-22Ms were already known to exist and the airframe was generic.)
In 1988, a viewer was tasked with locating a Chinese embassy being constructed in Bujumbura, Burundi. The viewer produced coordinates and a description two weeks before the embassy was officially announced. The accuracy was verified by the Army’s evaluation team.
None of these are airtight. Each has a counter-explanation — lucky guessing, prior information leaked through the experimenter, confirmation bias. But the program ran 1972–1995 with the same five-to-eight-person core team, and the hit rate, by the program’s own internal scoring, was roughly 15–20 percent above chance. By research standards, that is small. By intelligence-work standards — where the alternatives are satellite passes and human spies — that is interesting.
The Misses — When It Failed
In 1995, the CIA commissioned an external evaluation. Two reports came out. The first, by Jessica Utts of the American Statistical Association, concluded that the laboratory data showed a small but statistically significant effect — better than chance, smaller than the advocates had claimed. The second, by Michael Mumford and a team at the American Institutes for Research (AIR), evaluated operational use and concluded that the field record was not significantly better than chance at producing actionable intelligence.
A separate 1995 GAO report — GAO/NSIAD-96-2, “Stargate Program” — confirmed the program’s existence and its $20-million-plus budget over 23 years, but declined to validate the operational claims. The GAO noted, with understatement, that “the value of the program for intelligence applications is questionable.”
The hit rate in operational use was low. The 1984 hostage-rescue attempt failed. The Soviet submarine base was a single dramatic session in a long arc that produced nothing actionable. The Chinese embassy prediction was an outlier. The program’s defenders tend to skip past this. The 1995 evaluations are uncomfortable for both sides.
The Burying — The 1995 Closure and the 2017 Partial Release
In late 1995, the program was shut down. The CIA’s official line: operational value was insufficient to justify continued funding. The viewers were dispersed. The files sat, mostly inaccessible, for the next 22 years.
In 2017, a FOIA request forced the CIA to release the bulk of the archive — over 12,000 pages of session notes, memos, contracts, and internal correspondence, now at the CIA’s Reading Room under “Stargate.” Searchable, dense, and — as far as this archive can determine — not systematically read by any academic historian. The most interesting parts are the internal correspondence, where program managers argue with their evaluators about what the work means.
What remains classified is, by definition, the part of the record that did not come out. The early 1970s work that seeded SCANATE is still partly sealed. The late-1980s funding paper trail is partly redacted. The names of certain program officers are still blacked out. The same gap is documented in Operation Northwoods — a fully declassified program whose existence was unknown for 35 years.
What It Meant — The Program’s Legacy in the Simulation Argument
Stargate is the most complete declassified record we have of an attempt by the US government to operationalize a non-local information channel — to read a target with the mind alone, at distance, without instruments.
If the small operational effect is real, the simulation hypothesis has to account for it. A simulation that allows non-local information access — by viewer, by oracular dream, by cryptographic anomaly — is not the cartoon “video game” version. It is a version in which the simulation has properties that include a channel the operators did not expect the simulated entities to exploit.
The viewer sat in a shielded room, with a chain of custody, with a blind evaluator, with the same protocols a chemist would use for a hot experiment. And sometimes, on the program’s own internal scoring, the viewer produced information that should not have been available. Most of the time, they did not.
This is what we know. The full record is still partly classified. The program ran 23 years. It cost north of $20 million. It was shut down in 1995. The archive was released in 2017. The question — was it real? — has never been answered. The simulation archive notes that the question rarely is.
Sources & Further Reading
