The Global Consciousness Project

Roger Nelson has announced on the GCP Facebook page that his book on the Project is due to be published next month, under the title "Connected: The Emergence of Global Consciousness". I presume this will be the English version of the book published in German last year. It's to be published by ICRL Press.

The book is now available from Amazon in the USA:
https://www.amazon.com/Connected-Emergen...936033356/

And also in theory in the UK, though amazon.co.uk currently says "Usually dispatched within 1 to 2 months."

Apparently there will be an electronic version as well.

There is a short interview with Roger Nelson about the Global Consciousness Project on Carlos S. Alvarado's blog:
https://carlossalvarado.wordpress.com/20...er-nelson/

Also, I see that Roger Nelson's book about the project, Connected, is now available through Amazon in the UK as a paperback. No electronic version yet, though. (I may wait for that, as I already have a pile of books waiting to be read.)

Here's a review of Roger Nelson's book by Robert Charman on the SPR website:
https://www.spr.ac.uk/book-review/connec...r-d-nelson

Unfortunately the reviewer has got the impression that the Global Consciousness Project is looking for more 1s than 0s in the output of the random number generators during globally significant events. In fact the statistical measure that's generally been used essentially reflects correlations between different RNGs rather than deviations from the mean.

Disappointing, the review ends by referring to an analysis of the GCP 9/11 data by May and Spottiswoode, and saying these authors concluded that "the small deviations and small p values are easily attributable to variations to be expected in long streams of random sequences according to mainstream statistical theory." As I read it, the conclusion of the study was that for that particular event a different choice of the time period would have produced as non-significant result. But the study can't be judged on a single event. It has to be judged on the whole series of 500 events, for which the statistics are hugely significant. Whatever the explanation for the results is, it's not remotely plausible that it's just chance variation.

One hopes that these aspects are made clear in the book, though I haven't seen it yet.

I see that Roger Nelson's book is now available from Amazon on Kindle. There is a rather generous free preview, including the first five chapters (out of 28):
https://www.amazon.co.uk/Connected-Emerg...936033356/

Courtesy of the SPR Facebook page, here's what's described there as a review of Roger Nelson's book, but is really an extended description of the project with a one-paragraph review near the end. It's by John Walker at Fourmilab, who hosted the second random number generator in the GCP network:
"If you're interested in exploring further, Roger Nelson's book is an excellent introduction to the rationale and history of the project, how it works, and a look at the principal results and what they might mean. There is also non-formal exploration of other possible effects, such as attenuation by distance, day and night sleep cycles, and effect sizes for different categories of events. There's also quite a bit of New Age stuff which makes my engineer's eyes glaze over, but it doesn't detract from the rigorous information elsewhere."
https://www.fourmilab.ch/fourmilog/

I've been thinking about this a bit more, and it seems to me that the graphs produced by Peter Bancel - showing the averaged time-variation of the data before, during and after events - aren't really the clinching evidence for a psi-mediated selection mechanism that they might appear.

Here are the graphs:
[Image: SelectionGraphs.jpg]

Both graphs show the time-varying cumulative correlation data. The one on the left is for events where there was flexibility in the choice of start and end times (the events have all been stretched to a duration of 24 hours before averaging), while the one on the right is for 24-hour events starting and ending at midnight, where there was no such flexibility. The key point is that the graph on the left shows that the increase in the cumulative signal during the event is almost exactly cancelled out by decreases before and after the event. But the graph on the right shows no systematic change before or after the event.

The interpretation is that for the events with flexibility, psi has allowed the experimenter to choose start and end points so as to produce a positive effect within each event, at the cost of matching negative contributions before and after. The signal itself remains an ideal random walk, but a favourable period of time has been selected. For the 24-hour events this cannot happen. Instead, presumably, psi has allowed the experimenter to choose favourable days for 24-hour events, and to avoid unfavourable days.

One question that might be asked is why there shouldn't also be an element of choosing whether to define events on favourable days in the left-hand graph, as well as just choosing favourable start and end points. If there were such an element, the negative contributions before and after the event would only partially cancel out the positive contribution from the event itself. But as the graph shows, this didn't happen.

Another question is whether we can actually produce a graph that looks like the left-hand one by defining suitable rules for the selection of start and end points. I doubt we can. Perhaps the simplest rule would be to specify a start period and an end period for each event, and to select start and end points according to the location of the minimum and maximum respectively of the cumulative curve during these periods. I reckon that would produce something of the form below, which doesn't look much like the experimental data:
[Image: SelectionGraphTheoretical.jpg]
There's also a more quantitative objection to the psi-mediated selection mechanism. In the simple model just suggested, the selection within the start and end periods produces a contribution to the cumulative statistic that is proportional to the square roots of the durations of those start and end periods. That's because of the fundamental fact that the deviation of a random walk scales like the square root of time. So the same square-root scaling will also apply to more sophisticated models of selection based on the behaviour of the signal within fixed start and end periods.

That means we can estimate how the size of the effect, expressed as a Z value per event, should vary with the duration of the event (call it N). That depends on whether the start and end periods have fixed lengths, or whether they grow in proportion to the duration of the event. If they have fixed lengths, Z is inversely proportional to the square root of N. If they grow in proportion to the duration, Z is independent of N. Those cases are in contrast to a field-type effect, for which Z would be proportional to the square root of N.

It so happens that in 2015 (in "Evidence for Psi", edited by Broderick and Goertzel) Bancel tested the dependence of Z on N for events whose duration was 12 hours or less (thus, fortuitously, excluding the 24-hour events for which he now believes the mechanism of psi-mediated selection is different). He called this a signal-to-noise test. The result was that he rejected what he then described as the "data selection hypothesis", in which Z was independent of N. The result of the statistical test would translate to a (one-tailed) p value of .0037. That rejected hypothesis would be equivalent to the start and end periods growing in proportion to the duration of the event. The alternative, where the start and end periods had fixed lengths, would have been even more strongly rejected. (The 2015 paper was based on 426 events, and therefore represented about 85% of the complete series.)

Peter Bancel's conclusion at that time was that both a simple selection hypothesis and a straightforward global consciousness field hypothesis had to be rejected:
"The analysis of data structure rejects the simple selection hypothesis at a reasonably high level of confidence. The signal-to-noise analysis provides the most clearcut support for this conclusion. ...Tests for a loophole to circumvent the XOR no-go suggest that a straightforward conception of proto-psi global consciousness is also not tenable. ...
The analyses, then, provide good arguments for rejecting both simple models and we are forced to look elsewhere for an explanation."
https://books.google.com/books?id=KVyQBQ...&lpg=PA274

Going back to the discussion earlier this year about Peter Bancel's suggestion that the GCP results were consistent with a selection process, in which the start- and end-points of the events were chosen to increase the value of the test statistic - I've been looking at the question of how the effect size per event, Z, depends on the duration of the event.

To summarise what was said previously, there are three possibilities:
(1) With a selection model of this kind, if the intervals from which the start- and end-points are chosen don't vary with the total duration of the event (for example, if the start- and end-intervals are an hour long, regardless of the whether the event is two hours or two days), then Z will scale as the inverse square root of the event duration.
(2) With a similar selection model, but allowing the start- and end-points to be chosen from intervals that increase in proportion to the total duration of the event, then Z will be independent of the event duration. This is the same scaling that is to be expected for a selection model in which - instead of adjusting the start- and end-points - whole events are either included or excluded according to their Z values (for example, if only the events with Z values in the top 50% are included).
(3) If instead of a selection process, there is a "field" effect, causing some kind of steady perturbation of the statistics for the full duration of each event, then Z will be proportional to the square root of the event duration.

As far as I know, Bancel's most recent conclusion is that the results are owing to start- and end-point selection in most cases, but that for events one day long, starting and ending at midnight, there was selection of whole events. That means that - apart from whole-day events - we should expect Z either to be independent of duration, ot to vary like the inverse square root of duration.

It should be emphasised that Bancel proposed that these selection effects were psi-mediated experimenter effects. However, the same scalings would be expected if the selection were accomplished by conventional means.

Previously I referred to earlier work by Bancel, in which - in contrast -  he found statistically significant evidence of a field effect, corresponding to possibility (3) above. A field effect would result in Z growing in proportion to the square root of event duration.

However, Bancel's finding wasn't quite that straightforward, because he had to exclude longer events (in his analysis, longer than 12 hours), for which the effect (viewed as a steady "field" effect) was weaker. The analysis was also complicated by the inclusion of some events for which non-standard statistical measures were used, whose behaviour would be expected to be different. Also, only 426 of the full series of 500 events had taken place when he did the analysis.

I've looked at the variation of Z with the event duration just for the events for which the "standard" statistical analysis was used (that is, for each second the outputs of all the random event generators were added together, and then the squares of these sums were added together over the whole duration). I've also excluded a few events which were made up of several disjoint time intervals, for which the notion of choosing single start- and end-points isn't applicable. (I've also excluded the 13 events identified by Bancel as having been insufficiently specified before the data were examined.)

That leaves 415 events in total. I divided these according to their duration into 5 groups of roughly comparable size - up to and including 2 hours (n=78), 4 hours (n=64), 6 hours (n=101), 20 hours (n=69) and the rest (n=103). These are shown as a scatter plot below. The duration is the average within each group, and the error bars indicate the standard errors of the mean.

[Image: ZvDuration.jpg]

Obviously the dependence of Z on duration is more complicated than Bancel's earlier finding of support for a "field"-type behaviour would suggest. That would correspond to an increase of Z with duration. In fact, for the first three groups there is a decrease of Z with duration, and it appears consistent with the inverse square root dependence predicted by Bancel's start- and end-point selection model number (1) (I've added an inverse square root of duration line passing through the first point, for comparison).

However, for events longer than 6 hours the value of Z is much larger, and clearly must be accounted for by a different mechanism. It's not obvious to me what the explanation for that is.

In case anyone is wondering whether the midnight-midnight whole day events, for which Bancel proposed a different selection mechanism, behave any differently from events of similar duration whose start- and end-points can be chosen, here is a version of the chart in which they are separated out (shown as a red box):
[Image: ZvDuration2.jpg]
The answer is maybe, but the numbers are too small for the difference to be statistically significant. As the scaling behaviour suggests the end-point selection mechanism doesn't apply to these longer events, I'm not sure that we should expect a difference, or what it would mean if we found it.

Interestingly, the average Z value for the events longer than 20 hours with adjustable end-points is very similar to that for events longer than 6 and up to 20 hours. Although the overall effect size for the GPC (just over 0.3) would normally be classified as small, for these two groups of longer events, making up nearly a quarter of all the events, it is nearly 0.6, which would normally be described as medium to large.

In a recent Psi Encyclopedia entry on "Observational Theories of Psi," Brian Millar discussed briefly whether the Global Consciousness Project data indicate experimenter psi, and then commented:
"The conclusive piece of evidence would be an inability by other people to replicate the GCP effect: a cell phone app being prepared independently may settle the matter definitively."
https://psi-encyclopedia.spr.ac.uk/artic...eories-psi

Presumably this refers to "Entangled," a mobile phone app that was being developed by Adam M. Curry and was due to be launched three years ago:
https://noetic.org/blog/entangled-the-co...sness-app/

I see that funding of £14,000 of a target of £80,000 was raised through Indiegogo, and that there was a fair amount of publicity about the impending launch in 2016, but apart from a comment that June on Twitter that they were "a little behind," I can't work out what happened.

I wonder if anyone knows more?

Roger Nelson also kindly sent me a link to a short paper by him entitled "Evoked Potentials and GCP Event Data", which draws a parallel between the transient variations of electrical potential in the body following stimuli and the time-varying correlations during events in the GCP data. The idea is that this may suggest an alternative interpretation of the  results found by Peter Bancel, which he interpreted as evidence of a psi-mediated experimenter effect related to the selection of the start and end points of the events:
http://global-mind.org/papers/pdf/event....ntials.pdf

Here is a talk by Roger Nelson given at the Society for Scientific Exploration's Conference last year, in which he elaborates on this idea of comparing evoked potentials with the GCP data:


The part I definitely disagree with is the characterisation as an example of the gamber's fallacy of Peter Bancel's argument that psi-mediated selection of the start and end points of events, to maximise the signal during the events, would result in a smaller signal immediately before the start points and immediately after the end points. I'm sure Bancel is correct on that general point.

However, as discussed above, the detail of the time-variation doesn't seem consistent to me with the end point selection scenario. And I think some of the data plots given by Nelson in this talk reinforce that impression.

Here are a couple of exploratory attempts to see whether the GCP network is being affected by the coronavirus pandemic.

One, by Roger Nelson, looks at the periods when the US stock market was trading, during the week of 11-17 March when the Dow Jones Industrial Average fell by about 15%:
http://global-mind.org/papers/pdf/Dow.Mar11to16.pdf

The other, by Bryan J. Williams, looks at the day - 11 March - when the WHO declared a pandemic (this is also the first day that Nelson looked at):
https://www.facebook.com/photo.php?fbid=...1307644480

Both emphasise that these are exploratory investigations (unlike the formal GCP hypothesis series, which ended several years ago). But they do show some apparent deviations from chance behaviour, particularly on 11 March. If I understand correctly, these are mostly in the opposite direction from that seen on average for the main series of pre-specified "global events". That is, instead of the outputs of the different random number generators being slightly more correlated with one another than would be expected, they are slightly less correlated.