This is indeed a brilliant exposition, and an article that is important to read.
(Many thanks for showing it to me!)
The pursuit of the impossible, absolute precision in complex statistical modeling (whether in the thoroughly human world of economics or, as first described in this piece, in weather forecasting) carries it's own inherent and inescapable risk of 'long tailed events', massive, even exponentially growing, breakdowns in the models.
Wild 'unpredictable' events thoroughly at odds with the understanding of 'regularity' that is massaged out of the models normally....
The point of criticality, (the unpredictable point-in-time at which massive change cascades through the system... say, for example, the point at which one more grain of sand on a sand pile causes a landslide), was described and defined in Economic theory by the brilliant Hiram Minsky... and that point-in-time where the unpredicted cascade of events starts is now know as a "Minsky Moment".
This article added to my understanding of Chaos Theory in one especial way... where in an early section of the article it described the radically different results that Edward Lorenz received in 1961 when running a weather forecasting model with two different starting points (same data set --- *except* for the vanishingly small differences between inputing his numbers to an accuracy of SIX, or to THREE decimal places --- an initial difference in independent variables of less than a tenth of a percent).
Tracking the cause of the wilding divergent results from his model runs down to the (vanishingly small) *exclusion* of tiny bits of data was truly an eye-opener.
We humans become fools when, in our hubris, we attach too much confidence in our ever-more-complicated models attempting to mimic reality which themselves are dependent upon 'smoothing' of the data, and expecting predictability when independent data differences *within* the margin of error of our models can cause grossly and disproportionately different end results.
Humbling.
That's all I can say... humbling.
(And a caution: put not too much faith in the reliability of 'experts'... there is always a place for common sense and caution.) |
