We can say that the GCP/EGG project has produced a large, consistent
database of random
numbers in which subtle but significant structure occurs. Over
the past four years, we have been engaged in a
major analysis project that is producing substantial new results,
as well as a rigorous re-analysis of previous findings.
We have had some working meetings to discuss and
dissect these results, and some of our efforts work toward
models that can integrate and accommodate empirical facts that
are anomalies from the perspective of standard (mainstream)
scientific models. The GCP data, and the most recent
analyses provide what we can call "meat" that is ready to be
chewed and maybe digested into viable models.
For example, we now are able to connect the anomalies with
consciousness and intention based on solid findings and
arguments.
Many questions remain -- indeed they proliferate, but we
think it is fair and reasonable to draw some conclusions.
Here are some of the pieces of evidence that can form a
basis for modeling.
-
The formal analysis series shows a cumulative probability
against chance of about 10e-6, or odds against chance of
a million to one.
Put another way, the composite score
over many types of events exceeds Z = 4.7 (Nearly 5 Sigma,
the standard often set in physics).
-
The result shows a broad effect in one dominant statistic we
refer to as the network variance or netvar,
with an average Z = 0.3. This has been the default standard
analysis for most events.
-
The small average Z-score implies that individual events
cannot be expected to reliably show significant departures
from expectation. We need, depending on the type of event, a
few dozen examples to develop reliable statistics.
-
The netvar statistic is the average pair-product of reg
trials: < z(i) z(j) >. That is, the main effect is driven by
correlations among the eggs, and the significance level is
about 4 Sigma.
-
We have developed a second, fully independent statistic
called the covar, < z(i)^2 z(j)^2 >. It
is also significant at about 3 Sigma.
-
The two measures are significantly correlated, both
tracking the overall trends associated with the events in
similar ways. But there are important differences as well,
and we think these will be powerfully instructive.
-
There is a negative distance correlation between pairs of
Eggs on both Netvar and Covar measures. Pairwise
correlations have a slow dropoff over several thousand
kilometers, going to zero after about 10,000 Km.
-
The estimated number of people engaged in the events
matters. Effects for "big" events are larger, at a modestly
significant level, on all measures.
The following links provide access to some of the most
intriguing recent findings.
-
It seems likely that different kinds of events will have
different levels and qualities of effect. While
categorization is necessarily subjective, we can see
interesting variations by subset.
-
A logical exercise intended to help understand the source of
strong and persistent negative deviations points to
autocorrelation at the bit level
as a promising model for a deep-lying mechanism of effect.
-
Examination of the differing implications of the
XOR operation
imposed on two different types REG devices gives some support
to the autocorrelation model.
MORE TO COME -- send me an email to remind me ...
And as for the proliferating questions? Stay tuned.
|