Page 31 - Deep Learning

14 Introduction

the invading German armies and all seemed darkness, that England should
34
continue to resist nazi expansion, whatever the cost. The consequences of
this stance and churchill’s ability to transform it into a national policy cas-
caded through the war years and beyond, however unlikely Mr. churchill may
seem in the role of butterfly. From a complex systems perspective, there is no
conflict between levels of description, no contrast between attributing histor-
ical change to societal movements or to the actions of Great Persons. It is true
that the Allies won World War II by outproducing the fascist states, but it is
equally true that the Allies won because Winnie was stubborn. All social sys-
tems are sensitive to externalities; economic markets are thrown into disarray
by inventions and wars, and entire civilizations can become destabilized by cli-
mate changes. In The Landscape of History, historian John Lewis Gaddis writes
that the focus on complex systems in the physical and biological sciences has
“brought those disciplines closer than they once were to what historians had
been doing all along.” 35
our ability to predict the trajectory of social and political systems is
widely acknowledged to be limited. Even in the two most analyzed types of
social systems, elections and stock markets, prediction is notoriously difficult
and imprecise. Anybody who could predict the future course of the market
would gather riches without limit, but the only people who make money on
this possibility are the authors of books that advocate systems for how to make
such predictions. In The Box, a history of the standardized shipping container,
Marc Levinson writes: “Perhaps the most remarkable fact about the remark-
able history of the box is that time and again, even the most knowledgeable
experts misjudged the course of events. The container proved to be such a
dynamic force that almost nothing it touched was left unchanged, and those
changes often were not as predicted.” Political analysts have fared no better.
When the Berlin wall came down on november 9, 1989, it came as a surprise
to everybody in the West, including intelligence analysts, policy experts and
social scientists. The rise and fall of popular movements, social practices, reli-
gious commitments and fashions are not predictable to any interesting degree
in spite of the ambitions implicit in the term “social science.” 36
My purpose is not to deny the successes of predictive social science. Like
natural systems, a social system can be locally predictable over short periods of
time, but even then only if the system is not seriously impacted by an external
event. Instances of predictability in social systems are not unreal, but they are
rare. turbulence is the normal state of human affairs.
The lesson of the complex systems revolution is that we live in and through
complex, unpredictable and irreducibly turbulent systems. This statement is as