Anomalous Entropic Effects in 23 Years of Continuously Recorded Truly Random Data

Introduction: The Global Consciousness Project (GCP) is arguably the longest-running mind-matter interaction experiment in history (Nelson & Bancel, 2011). Composed of a world-wide network of electronic random number generators (RNGs), the GCP collects and stores random samples from each RNG every second and has done so continuously for over two decades. The GCP explores the hypothesis that moments of coherent collective mind, which may be evoked by events reported by the media that attract the attention of millions of people, are reflected by the simultaneous emergence of negentropic effects in physical randomness. This hypothesis is roughly based on the philosophy of dual-aspect monism, which views reality as consisting of two elements – mind and matter – both of which are said to arise out of something more fundamental, perhaps the primordial equivalent to information (Atmanspacher, 2012). Metaphorically, mind and matter may be viewed as two sides of the same coin, whereby heads and tails may have different properties, but their behavior is inextricably correlated. 

A formal experiment using the GCP data involved 500 events and ran from 1998 to 2015. The overall result showed a 7.3 sigma deviation from chance expectation during these events, indicating the existence of a correlation between mind and matter. Three classes of interpretations have been offered for this correlation. Skeptics typically assume that one or more features of the experiment were flawed. This interpretation can be abandoned by carefully reading the published reports, or by downloading the open-source database and rerunning the analysis from scratch. A second interpretation is that the observed correlation was due to “goal-orientation,” which refers to the investigators using their intuition to fortuitously select events to include in the experiment that would match the desired outcome (Bancel, 2017). This would presumably require an ability akin to precognition, and there is some evidence in favor of this interpretation. A third interpretation is that the mind-matter correlation arises because collective mind causally influences the physical world, i.e., a psychokinetic effect (Nelson, 2017).  

Methods: All data recorded by the GCP from 1998 through 2021 were downloaded and artifacts associated with RNG component failures or glitches in internet data transmissions were identified and removed. An analytical technique designed to detect emergence of structure in time-series data, called “multiscale entropy” (Costa et al., 2005), was applied to arrays consisting of two-weeks of one-minute non-overlapping segments of the random data. As a control comparison, those results were compared to the same arrays that were randomly scrambled. A second analytical approach explored if the random walk produced by the RNGs on a daily basis was not random but instead a convolution composed of many hypothetical mind-matter “influences.” That analysis was tested against a control composed by randomly scrambled arrays. 

Results: Both the multiscale entropy and deconvolution analyses revealed the emergence of statistically significant temporal structure in what should be random data. 

Discussion: This outcome appears to support the philosophy of dual-aspect monism. Another interpretation, given that the observed results are a correlation, and correlation does not entail causation, is that mind and matter are simultaneously influenced by a common factor, such as a geophysical or geocosmic variable (Krylov et al., 2014; Spottiswoode, 1997; Tracy et al., 2022; Yuan et al., 2006), or by fluctuations in power grids during engaging world events that cause a surge in the use of electronic devices, which might in turn influence the RNGs. A third interpretation is that the free parameters used in the two analyses may have been fortuitously selected in a goal-oriented manner to provide interesting results. As an exploratory analysis, these results are intriguing but should be regarded as preliminary. Future studies may benefit by splitting the database into two or more partitions, with one partition used to develop the analyses and the others used to see if the results of the first analysis would successfully replicate.