Global Consciousness Project - Is it real?

[robbiehouston]

According to the Wikipedia article on the Human Consciousness Project (which is trolled by skeptics, but whatever), the following criticisms were raised:

Based on an exploratory analysis of 'highly statistically significant' experimental results, the GCP has suggested changes in the level of randomness may have occurred during the September 11, 2001 attacks at the times of the plane impacts and the building collapses, and over the two days following the attacks.[17] Moreover, the GCP has identified similar 'anomalies' in the EGG data hours and even days before the attacks; while the GCP does not claim a causal relationship,[18] such changes—if genuine—would seem to imply either subconscious mass precognition, or backwards causality.[19]
Independent scientists Edwin May and James Spottiswoode conducted an analysis of the data around the 11 September 2001 events and concluded there was no statistically significant change in the randomness of the GCP data during the attacks and the apparent significant deviation reported by Nelson and Radin existed only in their chosen time window.[20] Spikes and fluctuations are to be expected in any random distribution of data, and there is no set time frame for how close a spike has to be to a given event for the GCP to say they have found a correlation.[20] Wolcotte Smith said "A couple of additional statistical adjustments would have to be made to determine if there really was a spike in the numbers," referencing the data related to September 11, 2001.[21] Similarly, Jeffrey D. Scargle believes unless both Bayesian and classical p-value analysis agree and both show the same anomalous effects, the kind of result GCP proposes will not be generally accepted.[22]
In 2003, a New York Times article concluded "All things considered at this point, the stock market seems a more reliable gauge of the national—if not the global—emotional resonance."[23]
According to The Age, Nelson concedes "the data, so far, is not solid enough for global consciousness to be said to exist at all. It is not possible, for example, to look at the data and predict with any accuracy what (if anything) the eggs may be responding to."[24]
Robert Matthews called it "the most sophisticated attempt yet" to prove psychokinesis existed, but cited the unreliability of significant events to cause statistically significant spikes, concluding "the only conclusion to emerge from the Global Consciousness Project so far is that data without a theory is as meaningless as words without a narrative".[25]



What are y'all's opinions on the issue?

[robbiehouston]

Independent Analyses


Here is a list of independent analyses

[Arouet]

I emailed Dr. Nelson awhile back and he was kind enough to respond.

Hi Arouet,

We record the sum of 200 bits (we call that a "trial") once per second
at all the sites. That is 86400 trials (17280000 bits) each day from
each egg in the network. The number of eggs grew from a few to 65 over
the first few years, and for the last 10 years that number has been
relatively steady. You can do the math if you like, but one simple way
to think of the data quantity is that the archive is about 30 Gb of
completely compressed (1 byte per trial) data.

The data as a whole are indistinguishable from expectation in the first
four moments of the appropriate statistical distribution (binomial). We
construct a control data background distribution using resampling of the
98% of data not in the formal event periods This is described in the
"background" menu item, which links to
Introduction to the Science Go to the "Control Data
section 2/3 down the page.

Best,
Roger

On 5/24/2012 9:33 PM, Arouet Arouet wrote:
> Hi Dr. Nelson. I’ve become familiar with your work through the Skeptiko
> website. I’m hoping you can answer a quick question for me. I’m curious
> as to the total numbers of 1s and 0s from the beginning of the
> experiment until now: are they at expectation? How far do they deviate
> when looked at from the perspective of the entire data set? I’m not
> talking about the sum of the events you’ve looked at but rather the
> entire stream from all eggs from beginning to end. I’ve tried to find
> this info on the site, but can’t seem to find it. If there is a link
> that shows the entire numbers (I’m sure its a pretty big number) that
> would be great too.
> Thank you so much in advance!

--
Roger D. Nelson
Director, Global Consciousness Project
rdnelson@princeton.edu Global Consciousness Project -- consciousness, group consciousness, mind

I'm not sure if he quite answered my question though but I didn't want to bug him (or seem too stupid to admit that I wasn't quite sure if he answered me or not!). What I've been trying to find out is if the total number of 1s and 0s are at expectation over the entire run of all the numbers. My hypothesis is, if there are a bunch of events that mess up the randomness in one direction or the other then the overall results after all this time should be significantly away from expectation. If not, then can we really say that any of the patterns that have been measured are anything more than variance? How would the total numbers even out over that mammoth sample if it is filled with non-random results?

[Johann]

Global Consciousness Project -- consciousness, group consciousness, mind

They have a whole page dedicated to discussing the Wiki article on them.

- Johann

[Johann]

Originally Posted by Arouet

I emailed Dr. Nelson awhile back and he was kind enough to respond.



I'm not sure if he quite answered my question though but I didn't want to bug him (or seem too stupid to admit that I wasn't quite sure if he answered me or not!). What I've been trying to find out is if the total number of 1s and 0s are at expectation over the entire run of all the numbers. My hypothesis is, if there are a bunch of events that mess up the randomness in one direction or the other then the overall results after all this time should be significantly away from expectation. If not, then can we really say that any of the patterns that have been measured are anything more than variance? How would the total numbers even out over that mammoth sample if it is filled with non-random results?

Hey Arouet, you're back!

They have done pooled analyses of all events they have tested from 1998 to 2012, and the odds against chance stand at more than 12 million to one (p=2.144-e12).

If what you are asking is whether they have pooled the results from all bits ever aqcuired then I think it is the wrong question; a non-chance event is equally likely to spike one way or the other. There should not be any significant correlation, however, with all the selected events and net variance. For an example, Radin and Nelson have done an analysis of earth-based tremors and ocean tremors, and only the land earthquakes reach a highly significant result. Seems like the fish don't respond to them the same way.

- Johann

[Arouet]

Originally Posted by Johann

If what you are asking is whether they have pooled the results from all bits ever aqcuired then I think it is the wrong question; a non-chance event is equally likely to spike one way or the other.

That is the question I am asking and while perhaps one might say that a non-chance event could spike one way or another, given that non-chance events are of different durations, I would still think that over time there would be a significant divergeance from variance.

If you have constant non-random patterns being inserted into a random stream, from what I understand over a large sample (and you don't get much larger than the entirety of the GCP stream!) you would expect - I believe - to see a significant deviation from chance expectation.

[Nemo85]

I have never taken the time to read the literature about Global Consciousness, but for the record it was debated in the last issue of Journal of Scientific Exploration at the end 0f 2011, some of the articles from that issue are freely available: "journal of scientific exploration" global consciousness - Google Scholar

[Trancestate]

Originally Posted by Johann

Hey Arouet, you're back!

They have done pooled analyses of all events they have tested from 1998 to 2012, and the odds against chance stand at more than 12 million to one (p=2.144-e12).


- Johann

It seems to me the odds for this probability are considerably higher: over 466 billion to one. I got them by plugging 2.144 and -12 into the Scientific Notation to Decimal Converter to get the decimal probability. Then I used Google to calculate the reciprocal and plugged the numbers 4.66418 and 11 back into the Converter.

Doug

[Johann]

Originally Posted by Trancestate

It seems to me the odds for this probability are considerably higher: over 466 billion to one. I got them by plugging 2.144 and -12 into the Scientific Notation to Decimal Converter to get the decimal probability. Then I used Google to calculate the reciprocal and plugged the numbers 4.66418 and 11 back into the Converter.

Doug

Ah, well it's all still quite a bit higher than even "well over a million to one" - which is what they say on their site.

Still, are you sure about your calculation? p=.05 is the standard for "significance", and that indicates 20 to 1, or a 5% chance. To take the GCP number and convert it, you simply push the decimal place up two and divide that from 100.

So, 2.144-e12 in normal notation is .000000000002144, which in percent is .0000000002144%, which divided by 100 is 46,617,910,447.7612 to 1.

46 billion to 1 sounds more reasonable, though you are free to point out where I erred.

Thanks, - Johann

[Trancestate]

Originally Posted by Johann

Ah, well it's all still quite a bit higher than even "well over a million to one" - which is what they say on their site.

Still, are you sure about your calculation? p=.05 is the standard for "significance", and that indicates 20 to 1, or a 5% chance. To take the GCP number and convert it, you simply push the decimal place up two and divide that from 100.

So, 2.144-e12 in normal notation is .000000000002144, which in percent is .0000000002144%, which divided by 100 is 46,617,910,447.7612 to 1.

46 billion to 1 sounds more reasonable, though you are free to point out where I erred.

Thanks, - Johann

Johann,

You're right about the probability's decimal conversion (in this case, moving the decimal point 12 places to the left). But after that I believe all you need to do is divide 1 by the decimal to get its reciprocal, or 1/x number. Here are some examples:

Decimal....Conversion....Reciprocal (1/x)....Odds
.10 ............. 1/.10 .............. 10 ............ 9 to 1
.25 ............. 1/.25 .................4 ............ 3 to 1
.50 ............. 1/.50 ................ 2 ............ 1 to 1

So, to express the GCP probability as a reciprocal or 1/x number, simply divide 1 by that probability:

1/.000000000002144 = 466,418,000,000

It's true that the GCP claims at the bottom of its results page that "The overall result is highly significant. The odds against chance are much greater than a million to one."

Summary of Results, Intro

But that was written a long time ago and hasn't been updated. Similarly, the GCP claims on its home page that "The probability is less than one in a billion that the effect is due to chance", but I think that was also written quite a while ago.

Global Consciousness Project -- consciousness, group consciousness, mind

Sorry for picking on you, but I think the odds against chance for the GCP results are orders of magnitude higher than your first estimate of 12 million to 1. They're far too high to be ignored, and it's astonishing that, to my knowledge, no one in the skeptic community has gone beyond hand waving to debunk the GCP's methodology.

Doug