Evidence for GCP

Since the inception of the GCP project in 1998 the continuous collection of trials from multiple devices has led to the creation of a database containing over 26 billion time stamped trials, as of May 2016 (Bancel, 2017). This allows researchers to look at various historical events that have occurred during this time and ask whether prior, during or after such events there were any noticeable changes in the random output of the devices. According to Nelson (2015) the answer to this is a clear ‘yes’. The findings from such research are often grouped under headings relating directly to the events that inspired them and include terror and tragedy, religion and ritual, concern and compassion, and celebrations.

Terror and tragedy

Attempts to register the influence of group consciousness at large events related to terrorist activity has shown significant effects. Perhaps unsurprisingly one of the most widely reported events examined was the terrorist attack on the World Trade Centre in New York, USA on 11 September 2001. When the data from the various REG units was examined it was found to have significantly deviated from random (see Figure 7.2; Nelson et al., 2002; Radin, 2002). The data also showed a substantial increase in structure which correlated with the most intense and widely shared periods of emotional reactions to the event (Nelson, 2002, 2014, 2015).

This was taken as evidence of a coherent global conscious response to the terrorist attack. Radin (2002) agreed, suggesting that this work pointed to some form of entanglement between mind and matter/machine. Furthermore, Nelson (2002) argued that such a pattern cannot be accounted for in terms of an electrical disturbance and/or

Figure 7.2 Cumulative deviation of GCP network up to II September 2001 (marked) and beyond (from Nelson et al., 2002, p. 6, with permission).

increased mobile phone use. However, not all agree with this outcome. For instance, May and Spottiswoode (2011) suggested that if the time window were altered making it either shorter or longer the effects would not have achieved significance. Nelson (2015) takes issue with this claim calling it ‘unacceptable post-hoc data selection’ (p. 286).

This, however, is not the only tragic event to show up in the GCP network. For instance, the funeral of Princess Diana provided a global event with the focus of many millions of people. Monitoring these events the GCP network again showed significant deviations from random chance (Nelson et al., 1998). Interestingly, a week later during the funeral of Mother Theresa, however, there was no clear deviation from random. It is of course difficult to control for the variety and nature of such naturally occurring events. Some may be more important, or deemed more important than others and as such the emotional response will differ. However, more recently Nelson (2011a) reported on an international event which involved the Israeli navy sending commandoes to stop a flotilla carrying humanitarian aid to Gaza in May 2010. This event caused the reported deaths of at least 10 pro-Palestinian activists and many more were injured in the skirmish. This sparked an international diplomatic crisis causing further strains on relationships between Israel and Turkey and led to condemnations by the United Nations and the European Union. During this crisis, examination of the GCP network showed a significant deviation from random chance (see Figure 7.3).

However, not all such events that have been examined have been shown to elicit significant effects. For instance, during 2009 there was a period of time when global concern peaked regarding a possible pandemic of the very deadly swine flu. When Nelson (2010) examined the data he did find a trend, in that there was a period of time when the data cumulatively showed a shift away from the random baseline, but there was no overall effect (see Figure 7.4).

The lack of an effect here may be because the effect itself was subtle and or the number of those involved may not have been sufficiently high, or the emotiveness not sufficiently engaging. It is certainly the case that the outbreak was not as serious as originally expected and the World Health Organisation officially declared the

Figure 1.3 Cumulative deviation of GCP network during a 6-hour period representing the Israeli attack on the Gaza flotilla (from Nelson, 2011, with permission).

Figure 1A Cumulative deviation during possible swine flu pandemic of 2009 (from Nelson, 2010, p. 7, with permission).

pandemic over in 2010. Hence, the nature of the event and the emotive response it evokes may influence the outcome.

Religion and ritual

According to Nelson (1997) a large organised global meditation event called ‘Gaiamind’ produced positive results. Another more recent religious event that

Figure 7.5 Changes in the GCP network during the Papal visit (top jagged line) compared to a control period (bottom jagged line) (from Nelson. 2019, p. 4, with permission).

attracted a lot of media attention was the week-long pilgrimage of Pope John Paul II to various sacred religious sites in the Middle East in March 2000 (Nelson, 2013, 2019). According to Nelson (2019) the data from the GCP network showed a persistent and non-random trend that was not evident when a later ‘control’ dataset was extracted (see Figure 7.5).

Such a finding is consistent with the data from a field REG approach which used two REG devices in an effort to detect possible changes in random output due to a nearby religious ceremony (Kokubo, 2016).

Concern and compassion

Nelson (2013) has also reported on the peace demonstrations during February 2003, aimed to show worldwide support for a peaceful resolution to the conflict in Iraq and the Middle East. During these demonstrations Nelson (2013) found that data from the GCP network was clearly random until about 11am, when crowds of people gathered in cities such as Rome, London and Berlin. Then the data showed a clear departure from random which continued for the rest of the day.

Celebrations

A good candidate for identifying a priori widespread engagement in a global celebration is New Year’s Eve. Everybody knows when it occurs, it happens every year and there is widespread engagement and media coverage. When Nelson (2013) reported on 10 years of GCP data from 1998 to 2008, the prediction was that the random variance of the REGs would decrease as midnight approached and then increase again after midnight. Averaged over all time zones and across all 10 years the data clearly showed a non-random pattern, which confirmed this prediction. According to Nelson (2013, 2014, 2015) the pattern exhibited across the devices

Figure 7.6 Cumulative changes (represented by the jagged line) in the GCP network from 1998 to 2015 along with the probability levels indicating that such changes are distinct from chance (from http://noosphere.princeton.edu with permission).

shows that the effects are real and robust, yet small. Because of the small effect size, it may well be the case that single events do not always produce a clear effect. However, Nelson (2013, 2014, 2015) has argued that a more comprehensive picture can be obtained by examining the cumulative effects of various trials from the GCP network since its inception (see Figure 7.6).

The jagged line in Figure 7.6 shows the cumulative sum of deviations from 500 tests compared to chance expectation with smooth lines to indicate statistically significant levels of 0.05, 0.001 and 10"⁶ (or 0.000001). According to Nelson (2014) the pattern to date shows an overall effect with odds of about one in a million that the correlations are due to random chance fluctuations. Such findings have led Nelson (2015) to argue that that amount of data collected and the pattern it exhibits provides highly significant evidence for something influencing the devices such that the output is no longer random. Indeed, Nelson (2014) reported that over the previous 15-year period positive correlations in the network that match predictions have been found in about two out of three cases. Though the effect for single trials tends to be small the composite results are robustly significant and deviate from chance expectations by seven standard deviations. Furthermore, resampling the output not linked to global events shows the expected random output. Hence, the non-random output is not something inherent in the output of the devices. The general consensus of opinion is that the findings from the GCP provides a clear indication of a correlation among globally distributed systems and human consciousness (Bancel, 2011, 2014; Bancel & Nelson, 2008; Nelson, 2001, 2002, 2010, 2011a, 2013, 2014, 2015; Nelson et al., 1998; Nelson et al., 2002; Radin, 2002). The implications of such correlations suggest that consciousness may create order though it is not clear yet what factors may influence this.

Mediating factors

Trying to understand what factors may influence or mediate the cohesiveness of the GCP network will not only help to shed light on the possible processes involved but also benefit attempts to model or explain such structure. To date researchers have examined the distance between devices and from the device to the event, the size of the event and the emotional response it produces as well as the timing of the event.

Distance

An interesting question is whether the correlations seen in the GCP network depend on the precise location of the various devices. Current understanding informed by the inverse square law would suggest that the intensity of any effect should change as a function of the square of the distance from the source. That is, as the distance between the source of the effect and the REG devices registering it increases the effect itself should decrease or weaken. However, as Nelson (2013) has pointed out, it is not always easy to precisely identify the spatial location of an event. For example, the Asian tsunami on December 26 2004 produced disastrous impact locally; however the response to the news reporting was literally global. Hence, it can be difficult to determine with any degree of precision the locus of an effect. It is not clear if data should only be collected from the location of the physical event or extended to capture the reaction to spreading news reports. Nevertheless, Nelson and Bancel (2011) have reported suggestive evidence of a linear regression effect with regards to distance. That is, the correlation between pairs of devices within the network decreases as the distance between them increases. Indeed, Nelson (2013) reported that the inter-device correlations decreased as the distances increased from 8,000 to 10,000 km, providing strong support for spatial structure in the data. However, Nelson (2015) points out that nonrandom output can still be found between devices that are separated by large distances during ‘global events’. There is also agreement that more work is needed to explore this in order to ascertain whether it applies uniformly or only to certain kinds of events. As such, any model attempting to describe the data would need to incorporate such structure in order to describe the data adequately.

Size of the event

It should be possible to separate events in terms of their size, based on estimates of the number of people attending or involved. In this way it should be possible to compare large global based events to more local ones. An intuitive expectation would be that larger more global events would produce more robust effects in the GCP network. Indeed, Bancel and Nelson (2008) reported that events engaging a very large number of people do produce significantly more deviations in the GCP network than minor events with fewer people. Nelson (2013, 2015) confirmed this, stating that such comparisons show that larger effects are found with global scale events compared to smaller local events. This would suggest that the number of people involved in an event would influence its cohesiveness. Furthermore, that these people would be required to respond in a similar way at a similar time for this coherence to occur and then influence the GCP network.

Figure 7.7 The GCP effect as a function of high (left), medium (middle) and low (right) levels of emotion (adapted from Nelson. 2008).

Emotive response

Several emotions are identifiable at various levels in the world events, and subjective ratings can be made with good reliability. While not all events can easily be assigned to a specific class or category, such as ‘fear/anxiety’, ‘positive feeling’ and ‘compas-sion/love’, all yield adequate samples to assess the presence of a high, medium or low emotional response. Bancel and Nelson (2008) suggested early on that events may require a strong emotional response from a large proportion of the population for an effect to clearly emerge. Indeed, when Nelson (2008a) examined the data he found that high emotional events tended to produce much greater effects than those that elicited either medium or low levels of emotion (see Figure 7.7).

Hence, the stronger or more intense the emotion related to the event the stronger the effect on the GCP network. Interestingly, it is the events that seem to provoke the extremes of these emotional responses, whether positive or negative, that produce the most powerful effects. That is, events with either strong positive or strong negative valence produce significant effects whereas neutral non-emotional events do not (Nelson, 2008a). According to Nelson (2008a, 2008b, 2013) the largest contributions come from events characterised by high levels of fear or compassion and love.

Timing

In terms of timing it is not always clear when to look for an effect or how long an event may last. This may be because, as Nelson and Bancel (2011) point out, an event need not be constrained by the moment of occurrence but also includes the spreading reactions to it. For instance, in response to most events human reactions will initially grow, persist whilst interest is maintained or fed by mainstream media and then possibly spike, and thereafter dissipate as attention moves on to other things. Given that the GCP reflects the level of emotional engagement of the people it is difficult to say precisely at what time a certain number of people will be engaged and as such it may only be possible to approximate the timing of an event. This has important implications as Bancel and Nelson (2008) point out; changing the time window that defines an event may influence the outcome. Nevertheless, Nelson (2013) suggests that data correlations should correspond to the human response to an event, rather than the event itself. As such, any effect may take some time to appear and would likely persist for a short period of time whilst people attend to and focus on the event and then disappear as interest and attention wanes. Indeed, Nelson (2013) reports that correlations between devices grows to significance between 45 minutes to an hour from the start of an event. The correlation has a broad peak of approximately two hours and then begins to diminish. This suggests that the effect may not always be immediate but may take time to build up to a level that is detectable. It may then remain stable for a brief period of time whilst interest in the event is maintained. This pattern is likely to be influenced by the type of event and the level of response.