The archive analyzed in the open: every figure drawn beside its null, every method declared before the data arrive.
These are the structural analyses the literature earned the right to ask for: serial correlation in time, personal signatures across people, and the position effects that haunt every archive in this field. Each figure is computed live from the whole record and drawn with its chance expectation beside it, so nothing has to be taken on trust: when a curve sits inside its band, that is the honest result, and it is published just the same. The figures’ numbers come from the open endpoint, which serves aggregates only: no identifiers, no locations, no raw streams, and no group smaller than three people. The measurement itself is defined on the method page.
Does one moment lean on the next? The trial statistics assume no memory at all.
Each line is the pooled autocorrelation of trial sums, mean-centered within every session so a mere lean cannot masquerade as memory, at lags of one to twenty-four trials, about six seconds to two and a half minutes. The shaded band is where a memoryless coin lives. The Global Consciousness Project’s deepest structural finding was autocorrelation persisting around ten minutes, and von Lucadou’s model predicts that if anything inhabits these streams it lives in correlations rather than means, which makes this the right first place to look. The split matters as much as the curve: structure that appears equally on calm pages and during exercises is the browser’s physics; structure that follows attention is something to chase with the preregistered protocol below.
Princeton found each operator distorted chance in a personally characteristic way. Do our records differ more than luck allows?
The first figure is the count population: the distribution of all trial sums against the exact binomial curve a fair coin commands, the same lens PEAR used to find that operators bent the shape of chance, not just its center. The second holds one dot per person, lifetime lean against lifetime variance ratio, and beneath it the one number that makes “signatures” a testable word: Cochran’s Q, which asks whether people differ from the pooled coin more than sampling luck can explain. A significant Q on lean, variance, or serial correlation is the statistical shadow of what Radin’s neural networks and PEAR’s profiles reported: that the record carries the mark of its person.
The field’s most stubborn replicated pattern is the saddest one: effects that fade with familiarity.
PEAR saw serial position effects inside its series; the Munich registry watched strong first results decay under preregistered replication, an oscillation their model treats as the phenomenon’s own signature and skeptics read as regression to the mean. A permanent archive is the one instrument that can tell those stories apart, because it never forgets its beginnings. The first figure tracks deviation by position within a sitting: a warm-up, a fatigue, or nothing. The second compares everyone’s first sittings against their later ones, and does the same for the Moon’s scored trials in the order each person met them, where the aims were assigned in advance and no selection can hide. If the first-exposure effect is real here, it will stand in these bars; if it is regression, the bars will say that instead, and either answer joins the record.
Exploration in the open, confirmation only by preregistration.
Everything on this page is the exploration layer: computed from the archive as it stands, refreshed hourly, splits and sample windows stated, artifact diagnostics recorded beside every stream. Nothing here will ever be promoted to a claim by looking good; promotion happens one way, through protocols frozen in advance like the one above, with the confirming data collected after the freeze. That is the standard both sides of this field’s hardest argument ended up demanding, and it is cheaper to honor than to argue about. The instrument is described exactly, with its limitations, on the method page; the history that earned these questions is on the experiments page.