Is remote viewing better than chance?
Every session on Social RV is done blind, scored by a blind judge, and added to our running experiment. This page shows what that experiment says so far, and exactly how we measure it.
The question, as a coin flip
Remote viewing is the practice of trying to describe a hidden target, a photo the viewer has never seen, using only their mind. If that is impossible, sessions on this platform should score no better than random guessing. If it works even a little, the data should drift away from chance as sessions pile up.
We turn every session into something like a coin flip. Each session is scored blind against the real target and nine decoys, earning a rank from 1 (best) to 10 (worst). Ranks 1–5 count as landing in the top half, ranks 6–10 in the bottom half. If remote viewing is not real, each session is a fair coin: 50% heads, 50% tails. The question becomes simple: are we seeing more heads than a fair coin allows?
What the data shows
The dashed line marks 8,442 sessions, which is what a pure 50/50 coin flip predicts for each half.
Sessions above the coin-flip expectation
+276
Odds of this happening by pure luck
about 1 in 89,665
In plain terms: if remote viewing were pure guessing, a lead this large across 16,884 sessions would come up about 1 in 89,665 if we reran the whole experiment from scratch. Scientists usually treat anything rarer than 1 in 20 (p < 0.05) as statistically significant. Our current result is extremely significant.
Standout viewers
The platform-wide numbers above average everyone together, including first-timers. These are the viewers whose personal record deviates most from chance, among the 130 viewers with at least 20 judged sessions.
Odds are each viewer's one-tailed binomial p-value for landing in the top half (ranks 1–5) more often than a fair coin predicts. With many viewers on the platform, a few low p-values are expected by luck alone, so treat this list as candidates worth watching rather than proof about any individual. Counts include each viewer's private sessions, but only these aggregate totals are ever shown.
How every session gets its score
The numbers above only mean something if the scoring cannot cheat. Here is the blind ranking pipeline every session goes through, in five steps.
Your target is drawn from the pool
Every session starts the same way: a target is picked for you from the target pool, and you never see it. You record your impressions blind, then upload your work. Your drawings, notes, and PDF pages go to the judge exactly as you submitted them.
Nine decoys join from the same pool
Nine decoy targets are drawn at random from the same pool your target came from. Together with the real target they form a lineup of 10 candidates, and any one of them could plausibly have been the target you were viewing.
The lineup is shuffled
Before the AI sees anything, the 10 candidates are shuffled into a random order. The judge is never told which one is real. From its point of view, every candidate is equally likely.
The AI ranks all 10
A multimodal AI studies your session next to every candidate and orders the full lineup from best match to worst. A separate verification pass then checks the reasoning against each ranked target, and the ranking is redone if it fails.
Your rank is the result
Your score is simply where the real target landed in that blind ranking. #1 means your session pointed at the real target more strongly than at any decoy. Pure chance averages around #5 to #6, so ranks 1 to 3 suggest a strong hit.
Why the judge stays blind
A judge that knows the answer can talk itself into finding it. Because our AI judge must rank the real target against nine plausible decoys with no hints, the real target only lands at #1 when the session genuinely describes it better than the alternatives. This mirrors the independent blind judging protocol developed at Stanford Research Institute in the 1970s. We simply run it instantly, on every session, with a multimodal AI.
Perfect pick
Your session matched the real target more strongly than every decoy.
Strong hit
The real target beat almost the whole lineup. Well above chance.
Above chance
Better than the random average, but not a decisive match.
Around chance
The session did not distinguish the real target from the decoys.
Want the full detail, including how the judge's reasoning is verified? Read the complete blind ranking guide.
A note on the decline effect
Psi research has a well-documented quirk: effects that look strong early in a study often shrink as more data accumulates. Researchers call this the decline effect, and it shows up in other fields too. Possible explanations range from early publication bias and regression to the mean to genuinely unexplained properties of psi itself.
We take it seriously. Because every judged session on the platform feeds the same running analysis, we can watch whether our significance holds, strengthens, or fades as the dataset grows. The significance-over-time chart in the advanced section below is exactly that watch, updated continuously and shown unedited whichever way it moves.
How this fits the wider research
We are not the first to run this experiment. Decades of laboratory research, from the SRI remote viewing program to the ganzfeld telepathy studies, have reported small but persistent above-chance effects. Statisticians and physicists including Jessica Utts, Dean Radin, and Peter Bancel have analyzed these datasets in depth. Read the originals and judge for yourself:
The ganzfeld telepathy experiments
Bem & Honorton (1994), Psychological Bulletin
Meta-analysis of ganzfeld studies reporting roughly 32% hit rates where chance predicts 25%, one of the most replicated protocols in psi research.
An assessment of the evidence for psychic functioning
Jessica Utts (1995), report for the CIA-commissioned AIR review
Statistics professor Jessica Utts reviewed the US government remote viewing program and concluded that "psychic functioning has been well established" by the standards applied to any other area of science.
A perceptual channel for information transfer
Puthoff & Targ (1976), Proceedings of the IEEE
The original SRI remote viewing paper that introduced blind independent judging against candidate targets, the same protocol our AI judge automates.
Meta-analysis of free-response studies, 1992–2008
Storm, Tressoldi & Di Risio (2010), Psychological Bulletin
Updated meta-analysis of ganzfeld and other free-response ESP studies finding a small but statistically significant effect across decades of data.
Searching for global consciousness: a 17-year exploration
Peter Bancel (2017), Explore
Physicist Peter Bancel's rigorous statistical analysis of the Global Consciousness Project's random number generator data, including a careful look at alternative explanations.
Dean Radin's selected psi research
Dean Radin, Chief Scientist at IONS
A curated collection of peer-reviewed psi publications spanning presentiment, meta-analyses, and mind-matter interaction experiments.
Are you a researcher?
If you would like to analyze this data further, reach out and we can arrange access to the raw dataset: per-session blind ranks, timestamps, and session types, suitable for independent analysis.
Join the community
Start your remote viewing journey today and contribute to the advancement of consciousness research!