The Cold War produced many strange decisions. But few were stranger than this one: career intelligence officers, reviewing reports from Soviet bloc countries in the early 1970s, concluded that psychic research might be a national security problem.

The Soviets called it psychotronics. The study and potential weaponisation of psychic phenomena. Whether their programme was real research or elaborate disinformation — some combination of both is plausible — the American response followed a logic that was hard to argue against at the time. If there is even a small probability the enemy can gather intelligence with their minds, the cost of not investigating may be catastrophic.

This is not irrationality. This is asymmetric risk calculation. The same logic that funds pandemic preparedness and missile defence funds things that later look absurd if the threat never materialises. Or things that look prescient if it does.

The research anchored at Stanford Research Institute — an independent think tank that had split from Stanford University in 1970. Two physicists ran it: Russell Targ and Hal Puthoff. These were not mystics. Puthoff had published in mainstream laser physics journals. Targ had a legitimate research background. Both were intellectually open to unconventional possibilities. Their colleagues found this uncomfortable. The CIA found it useful.

The programme ran from approximately 1972 through 1995. It cycled through bureaucratic identities — GONDOLA WISH, GRILL FLAME, CENTER LANE, SUN STREAK — before arriving at the name that stuck: Project Stargate. Multiple agencies held it at different times. The Army. The DIA. The CIA. Careers were staked on it. Budgets were appropriated. The files are now declassified. The programme happened.

Popular accounts collapse the distinction between two different things. The controlled laboratory research and the operational military applications were related but not the same. Conflating them produces a muddied picture of both.

In the lab, a typical experiment worked like this. A monitor — usually a researcher — would sit with a percipient (the remote viewer) in a shielded room. A separate person, the outbounder, was driven to a randomly selected location from a pre-prepared target pool. The viewer, knowing only that the outbounder had gone somewhere, would describe impressions — visual, spatial, tactile, emotional — of wherever that person was. The session was recorded in full.

Later, an independent judge — blind to the actual target — would compare the viewer's transcript against the real location and several decoy locations, ranking them by correspondence. This rank-order matching methodology was designed to close specific loopholes. No unconscious signalling. No cold reading. No selective reporting. In a pool of five possible locations, chance predicts a correct identification roughly twenty percent of the time. The question was whether the hit rate was meaningfully higher.

In 1974, Targ and Puthoff published their results in Nature. One of the most prestigious scientific journals on earth. The peer reviewers were notoriously sceptical. The paper passed after extensive revision, with editorial commentary making clear that publication was not endorsement — but that the methodology appeared sound enough to warrant scrutiny. The results showed correct identification substantially above chance.

The protocols tightened over the following years. Double-blind designs were introduced. Sensory leakage — the possibility of viewers picking up subtle physical cues rather than accessing information by any unconventional means — was systematically addressed. The work eventually transitioned to Science Applications International Corporation (SAIC), where a newer generation of experiments with stricter controls ran through the late 1980s and early 1990s.

Methodological refinement is what science looks like when it takes a problem seriously. SRI took it seriously.

No honest account skips this part.

Ingo Swann was an artist and self-described psychic who became arguably the most important single figure in the programme's development. He did not arrive through neutral screening. He arrived with a prior reputation and strong personal conviction about his abilities. He was involved in early SRI experiments in ways that blurred the line between subject and collaborator.

He contributed substantially to developing Coordinate Remote Viewing (CRV) — the structured technique in which viewers are given only geographic coordinates and asked to describe the target location. His conceptual contributions were intellectually serious. His distinction between analytical overlay — the tendency of the conscious mind to interpret and corrupt raw perceptual impressions — and raw perceptual signal remains genuinely interesting as a cognitive framework, regardless of what one believes about the underlying phenomenon.

But when a subject helps design the protocol by which they are tested, the evidentiary value of subsequent results is complicated. This is not an accusation of fraud. It is a structural problem in experimental design. Swann appears to have been entirely convinced of his abilities and genuinely curious about them. His influence was creative and substantial. The historical record requires honesty about the degree to which the programme's star subject and its core methodology co-evolved in ways that create legitimate scientific questions.

Swann illustrates a broader problem that follows this field everywhere. The file drawer problem. Experimenter degrees of freedom. Which sessions get written up? Which ambiguous correspondences get scored as hits? These problems are not unique to parapsychology — they haunt nutrition science, pharmaceutical research, psychology. But they cut deeper here because the phenomenon being studied is already at the outer edge of what mainstream science considers worth investigating. Small biases in protocol execution can, in marginal effects, produce statistical artefacts that look like real phenomena.

The rigorous answer is pre-registration. Commit the methodology before the data arrives. The programme rarely did this systematically. That is a genuine weakness in the evidentiary record.

In 1995, facing a decision about whether to terminate Stargate, the CIA commissioned a formal evaluation of the entire programme. Two evaluators were appointed with explicitly contrasting starting positions.

Ray Hyman was a psychologist and prominent sceptic of parapsychological claims. Jessica Utts was a statistician at the University of California, Davis, willing to evaluate the data on its own terms without a prior commitment to the conclusion. The contrast was deliberate. What they found, and where they diverged, is the most important document in this entire debate.

Utts's finding was direct. Applying the same statistical standards used in any other area of science — in pharmacology, in psychology, in epidemiology — she concluded that psychic functioning appeared to have been established. The effect sizes across the studies she reviewed were consistent. Not large. What social scientists would describe as falling between small and medium. But reliable. Replicable under proper experimental conditions. Not the result of chance.

More important than any single study was what Utts observed across studies. Similar effect sizes had been replicated in multiple independent laboratories around the world. Different experimenters. Different subjects. Different countries. Different methodological designs. The consistency of magnitude across independent replication is, in conventional scientific reasoning, one of the strongest arguments for a real phenomenon. A single laboratory producing unusual results can be explained through a dozen local mechanisms. Multiple independent laboratories producing results of similar magnitude requires a different kind of explanation.

Utts also examined precognition — the more radical claim that remote viewers might perceive future targets before they are selected. Several experimental designs had tested this specifically, providing targets to viewers only after the description was already recorded. These designs eliminate certain information-leakage channels that could explain standard remote viewing results. Utts found that precognitive designs worked at least as well as contemporaneous ones. If taken at face value, this has significant implications for any mechanistic account of what might be happening.

Hyman's counter-evaluation did not dispute the statistics. He agreed the results were unusual and could not be dismissed as simple methodological failure. His disagreement was philosophical. The data, he argued, was not yet sufficient to overturn the prior probability that psychic phenomena do not exist. Extraordinary claims require extraordinary evidence. What counted as extraordinary enough was the real point of disagreement.

This was not a dispute about numbers. It was a dispute about prior probabilities and the burden of proof. How anomalous must results be before they can overturn a strongly held prior belief? This is a genuine epistemological question. It has no clean answer. And it maps onto some of the deepest unsettled disagreements in the philosophy of science.

Parallel to the laboratory programme, a military operational unit developed — primarily through the Army and later the DIA. Military personnel were recruited and trained in remote viewing techniques from the late 1970s onward. They were given real intelligence problems. The location of hostages. Soviet military installations. Activities at undisclosed sites.

Joseph McMoneagle is generally considered the most operationally successful remote viewer the programme produced. A Warrant Officer with a background in military intelligence, McMoneagle was involved in an early experiment in which viewers were asked to describe a large, unusual Soviet facility at Semipalatinsk before satellite imagery had been obtained. The descriptions provided were later compared against imagery and reported to show significant correspondences — including an unusually large building associated with what analysts believed might be a new weapons system.

These operational claims resist clean evaluation. There is no independent blind judging. There is no pre-specified protocol. The degree of retrospective confirmation bias possible when matching verbal descriptions against complex sites is considerable. A viewer who describes "a large structure" at a military installation has described something consistent with almost any military installation.

McMoneagle has spoken and written about his experiences at length. His account is neither triumphal nor dismissive. He describes an ability that was real but inconsistent. Useful but unreliable. Present but not fully controllable. This pattern — striking when it worked, inexplicably absent when it didn't — appears in virtually every serious practitioner's account. Effect size variability and intraindividual inconsistency of this kind are not what you would expect from a conventional perceptual channel. They may or may not rule out a genuine phenomenon. What they rule out is a reliable intelligence tool.

The CIA drew exactly this conclusion in 1995. Following the Utts and Hyman evaluations, Stargate was terminated. The stated reason was not that the phenomenon was proven false. It was that operational utility was too limited and too inconsistent for continued investment. Intelligence produced through remote viewing could not be verified until after the fact. You cannot act with confidence on information you cannot verify. As a spy tool, it remained fundamentally ambiguous.

If the statistical evidence is taken seriously — not as proof, but as requiring explanation — what theoretical frameworks are available? This is genuinely speculative territory. Intellectual honesty requires saying that plainly before entering it.

The most frequently invoked physics connection involves quantum non-locality. Quantum entanglement — in which measurements on separated particles show correlations that cannot be explained by local hidden variables — is established physics. It won part of the 2022 Nobel Prize. Some researchers have asked whether some analogous non-local coupling might exist at biological or cognitive scales. The physicist Evan Harris Walker proposed formal models. Puthoff wrote about connections to zero-point energy and vacuum fluctuations.

Most physicists regard these proposals as involving category errors. Importing quantum phenomena from subatomic scales to macroscopic cognitive processes requires justification that has not been provided. The fact that entanglement is real at the quantum level does not mean something analogous operates in the brain. The proposals are not without intellectual content. But unproven is different from disproven, and the distinction matters for how seriously they deserve to be held.

A second theoretical line draws on the work of physicist David Bohm and neurophysiologist Karl Pribram, who explored the idea of the universe as holographic — with information about the whole encoded non-locally in every part. Bohm's holonomic brain theory is speculative but was explored seriously in academic contexts. Remote viewing, on this account, would represent not transmission through physical channels but something closer to direct access to an underlying information structure that is everywhere at once.

The most scientifically conservative explanation requires no new physics at all. Some researchers have proposed that remote viewing, if real, might involve an extraordinarily refined form of unconscious probabilistic inference — the brain integrating extremely subtle environmental signals in ways that produce accurate impressions without any unconventional mechanism. This would demand revision of our understanding of human perception. It would leave the physics textbooks intact.

None of these frameworks is established. All are speculative to varying degrees. What they share is the attempt to think rigorously about mechanism — which is precisely what distinguishes serious theoretical inquiry from wishful thinking.

When the Stargate files were released in the mid-1990s, something predictable happened. The careful, contested, methodologically complex laboratory research became, in popular culture, something closer to confirmed fact. Or, in sceptical media, confirmed nonsense. Neither was accurate.

A cottage industry of civilian remote viewing trainers emerged in Stargate's wake — former military viewers, people who had trained with them, people who claimed to have trained with people who had trained with them. Some are credible, explicit about limitations, and honest about what the original research does and does not support. Others have made commercial and predictive claims that the research record does not support at all.

The distance between "statistically significant effects observed under controlled conditions with trained subjects in a research programme" and "learn to locate missing persons and predict stock markets in a weekend workshop" is vast. Much of popular remote viewing culture has collapsed that distance entirely.

This has damaged the scientific credibility of the field more effectively than any sceptic's argument. When practitioners make falsifiable predictions — and the predictions fail — they create noise that makes it harder to evaluate the more modest, carefully documented laboratory phenomena. The signal-to-noise problem in this field is not only technical. It is sociological. It is commercial. Every sensationalised claim by a weekend instructor is an argument against taking the laboratory data seriously, even though the two things are largely unconnected.

The same logical error runs in the other direction. Dismissing the SRI and SAIC laboratory research because of what remote viewing culture has become is judging the evidence by its loudest distortion. Targ and Puthoff's 1974 Nature paper is not answered by pointing at a YouTube channel.

Remote viewing does not sit alone. It sits inside a larger set of questions that are becoming more urgent, not less, as neuroscience and physics develop.

The hard problem of consciousness — why subjective experience exists at all, why there is something it is like to be a perceiving creature — remains entirely unsolved. Mainstream science has mapped neural correlates of cognitive processes with increasing precision. The explanatory gap between neural firing and phenomenal experience remains as wide as it was when David Chalmers named it in 1995.

Chalmers proposed that consciousness might be a fundamental feature of reality — panpsychism in one of its more careful contemporary forms — rather than an emergent product of sufficiently complex computation. If something like this is true, the question of remote viewing changes shape entirely. The question is no longer how information travels through physical channels to a brain. It becomes how consciousness, as a fundamental feature of reality, relates to the information structure of events at a distance.

Neuroscientist Giulio Tononi's Integrated Information Theory proposes a mathematical framework in which consciousness is identified with a specific property of information integration. It has attracted serious academic attention and pointed criticism in equal measure. Physicist Roger Penrose and anaesthesiologist Stuart Hameroff developed Orchestrated Objective Reduction, proposing that quantum processes in neural microtubules are relevant to consciousness. This remains highly contested. It also remains in print, cited, and debated in serious scientific venues.

None of this proves remote viewing is real. What it establishes is that confident assertions about the absolute limits of perception — about what minds can and cannot do — are made against a background understanding of consciousness that remains deeply, genuinely incomplete. The territory at the edge of mind is not mapped. The people who insist it is mapped are speaking from habit, not knowledge.

The classified files are open. The data exists. The statistical arguments are tractable. The theoretical frameworks, speculative as they are, exist in print and in peer-reviewed journals. What is absent is not evidence. What is absent is the institutional willingness to pursue the question with the full rigour it deserves.

Neither incurious dismissal nor credulous enthusiasm is an adequate response to what the Stargate research actually produced. The Utts findings have never been seriously refuted — only disputed on philosophical grounds about prior probabilities. The Hyman objection — that replication under truly standardised conditions has not been achieved — remains valid. It is also, two decades after Stargate closed, a standing indictment of every institution that could fund that replication and chose not to.

The programme was terminated not because the phenomenon was proven false. It was terminated because a real intelligence agency could not build reliable operational workflows around results that were consistent but inconsistent, statistically meaningful but individually unpredictable. That is a statement about operational utility. It is not a statement about truth.

Whatever remote viewing ultimately turns out to be — genuine anomalous cognition, an extraordinary amplification of unconscious inference, a persistent statistical artefact produced by methodological flaws not yet identified, or something stranger than any of these — the question is one of the rare places where the limits of current knowledge become clearly visible. That is precisely where inquiry should press hardest.

The files are open. The question is whether anyone will press.