Magnets, Means, and Errors

Did you ever wonder what the world would be like if a notable person had never been born?  Did you ever realize the futility of that musing and wonder how it is that the work of one person could shape the whole of humanity in profound ways far beyond their intentions?  Well, today we celebrate a birthday.  No, not mine!  And my how I loved turning 50 this past week and the great celebration that attended the event.  No, the birthday in question was 190 years earlier – April 30, 1777 to be exact.  Like me, he was intrigued with mathematics.  Like me, he taught about magnetism and the universal principles associated therewith.  And like me, he was fascinated by light and the manipulation of light through lenses to understand its essence.  But unlike me, he thought that, “the world would be nonsense, the whole creation an absurdity without immortality.”

The birthday we celebrate today was none other than Johann Carl Friedrich Gauss.  And the reason why I care about his birthday is singular.  He gave humanity one of its most toxic cognitive forms of bondage.  And together with his French collaborator (of sorts) Adrien-Marie Legendre, Gauss is unconsidered at our collective great peril.

In an effort to predict astronomical movements – principally the orbits of planets – Gauss developed a computation known as the Gaussian gravitational constant which is built on a mathematical notion of least squares.  This algebraic notion – that observable phenomenon can lead to forward predictions and thereby minimize measurement error – was innocent enough for its immediate application.  When one is trying to figure out where a planet is going to be in three days hence, this math had its utility.  But, like other astronomical innovations, Gauss’ work unleashed a toxic divination impulse that has become the root of our modern scientific inquiry and the bane of humanity.  The notion of prediction based on linear regression.

Now some of you hate math and, acknowledging that, let me explain something.  As the abject failure of pundits and analysts have shown in the 2016 U.S. Presidential election, if you measure consensus assumptions, your conclusions are entirely wrong.  In 2006 and 2007, I correctly described the conditions and the timing of the Global Financial Crisis in 2008[1].  Was I forecasting an outcome using predictive analytics?  No.  I was merely observing irrefutable documented behaviour in an occult industry and critiquing the system level convergence that was certain.  From mass pandemics (the Asian bird flu) to resource shocks to social paroxysm (the Egyptian multi-coups), the “trained” and the “expert” are left agape when linear regression behaviour is punctuated by disequilibrium events.  Regrettably, education’s obsession with the scientific method have taught regression but have assiduously ignored its dominant fallacy – that we know the variables that matter and we recognize that which is significant.  Elementary statistics teach us that interrogatory inquiry presupposes:

1.      Known variables;

2.      Known scale in which these variables operate;

3.      And Measurement Error.

Interestingly, the same discipline teaches us the error of untested assumptions about normalcy, kurtosis, skewness, and orthogonality.  However, the modern education system and the scientific method upon which it is built fails to account for these in every instance diminishing the efficacy of social and technical interventions.  Put another way, in the world of obsessive prediction and outcomes, we rely on our elementary algebra which seeks to solve for y in the classic linear predictive model:

y= mx +b

where y is the expected outcome; x is the variable(s) we think have an association with an observable; m is the scale or range in which x operates; and, b is unexplained variance.  This formula presumes that we know both the association between observations and effects (an entirely fallacious assumption), we understand the scale in which variables operate (an entirely untested assumption), and that the remainder is “unexplained”.  We don’t hold the possibility that the entire ontology projected in regression may in fact be prima facie false.

Here's the problem.  We are not conditioned to ask any of the fundamental assumptions that underpin the error of statistical divination.  We want to “know” what’s going to happen.  We want to “control” outcomes by manipulating variables.  But what we constantly ignore is the fact that the human analog experience does not happen on 2-dimensional scatterplots through which lines can be drawn.  Every struggle you’ve had; every emotional pain; every sense of loss; is based largely on the fact that you projected a plane in a dimension around which you built a narrative.  Often those narratives involved others.  But they had their narratives, their frameworks, their projections.  And just because a dot showed up in your world and a dot showed up in mine doesn’t mean that the lines that you connect and the lines I connect go in the same direction.  In fact, it’s certain that they do not.  So one day, using the exact “same” data, we arrive at different conclusions.  And then we expend amazing energy trying to re-narrate what we didn’t understand in the first place.

Gauss gave us a 2-D god-complex in a multi-dimensional world.  And as long as we subscribe to either of those features (the error of 2-D or the god-complex) we insure nothing but pain for our existence.  Our obsession with the “future” is nonsensical.  Making up stories and myths about immortality – a prerequisite for Gauss – implies that the present is insufficient.  And with that measurement error, all other measurements (and the measurers) will be disappointments. 

---

.[1] Martin, David E. “Social Contingent Liabilities and Synthetic Derivative Options” EUPACO-2, Brussels.  15, May 2007.