Stargate: Twenty Years of Government Remote Viewing

The popular version is that "the CIA proved psychic spying works" or, alternatively, that "the CIA proved it was all nonsense". Both claims cite the same document: the 1995 American Institutes for Research evaluation. Reading the primary files is a lesson in how a single dataset supports two irreconcilable summaries.

The programme in outline

• 1972–1986, SRI International. Physicists Harold Puthoff and Russell Targ run laboratory trials, initially with Ingo Swann and Pat Price. Funding from CIA, later DIA and Army.
• 1986–1995, Fort Meade. Operational unit, at times fewer than six viewers, tasking from multiple agencies.
• 1995. Programme transferred to CIA, evaluated by the American Institutes for Research, terminated and declassified.

The AIR evaluation, in its own words

Two reviewers were commissioned. Jessica Utts, a statistician, concluded that the laboratory effect size was consistent and statistically robust across replications. Ray Hyman, a psychologist, agreed the statistics were not the result of obvious error but argued that a methodological flaw not yet identified remained more likely than a novel channel of information.

Utts: the effect is "far beyond what is expected by chance." Hyman: "the case for psychic functioning seems better than it ever has been" — while still concluding the evidence was insufficient.

The programme was closed on the operational finding that the intelligence produced was too vague to be actionable, which is a different judgement from "the effect does not exist".

The best-documented operational episodes

Files describe viewers producing correct structural descriptions of targets they had no conventional access to, alongside a large volume of descriptions that were wrong, generic, or unverifiable. The declassified record does not contain a clean, pre-registered operational success.

Where the methodology is genuinely weak

1. Judging. Matching a sketch to a target site involves human judgement; blind judging was used but panel selection varied.
2. Target pools. Early SRI target sets were criticised by David Marks for potential cueing through transcript ordering.
3. Selection of reported cases. Operational anecdotes circulated publicly were chosen after the fact.
4. Sensory leakage. Rarely fully excluded in the earliest trials; tightened later.

::: callout The honest summary: a twenty-year, tens-of-millions-of-dollars effort produced a laboratory statistical anomaly that two competent reviewers could not dismiss, and an operational record too noisy to use. Neither "proven" nor "debunked" fits. :::

Stargate is the most expensive negative-but-not-quite-negative result in the history of anomalous cognition research. The documents reward reading precisely because they refuse to deliver the headline either camp wants.