All things “new” fascinate us. Of late, the business world’s growing excitement about Big Data and its analytic modeling seems to turn up surprising results in interesting places. Predictions mesmerize us, they offer us control in the midst of uncertainty and fool us to believe we understand things more completely than is possible. The models used to predict an outcome are often confused with underlying mechanisms responsible for the outcome. Models fuel discovery and yet we get cocky when we rely wholeheartedly on a built model’s power and accuracy. Risk doesn’t disappear and its infrequent appearance merely challenges our ability to prepare adequately and only in hindsight differentiate the early warning signs. This is what int he trade we call differentiating signal from the noise and is the focus of Nate Silver’s book.
Power to Predict
In finance, or physics circles the fascination around models is anything but new. Isn’t the primary purpose of analysis and model building discovery or greater understanding of causal relationships and interactions? Observing physical properties of planets helps us make sense of their movements and explain other observable phenomenon. The notation and models provide insights into other activities and data collected in other settings. These scientific modeling techniques when introduced into social science formed the basis of understanding economic behavior and a framework for a series of policies governing the money supply to welfare. Once operating in obscurity, the mathematically trained analysts and modelers impact on society continues to ripple into ever-widening arenas difficult to miss.
Michael Lewis earned his living as a quant on Wall Street. His dual talents manipulating numbers and words led to his successful book Liar’s Poker. Complexity found a voice and Lewis continued to seek out and tell more stories about the quants in multiple settings. Perhaps it was the popular success of MoneyBall, that attracted the popular interest. I admit I’m an ardent fan. Michael Lewis and his wonderful story telling ability around number problems, shared how the Oakland As made the playoffs using statistics for competitive advantage. Among the collected stats, the story revealed those overlooked by scouts the Oakland As valued, making it possible for them to compete effectively against baseball teams with much larger budgets.
In Presidential Elections, during 2008 the baseball stats model maven Nate Silver demonstrated how a command of statistics can improve the quality of a candidate’s campaign. By 2012, his success garnered him personal attention as author of the New York Times 538 column while further upping the fascination with applied statistics in new arenas.
Leonard Mlodinow, a trained physicist himself, in his sympathetic review of Silver’s new book, shares his frustration with statistical shysters. “The Signal and the Noise,” Silver shares “… studies show[ing] that from the stock pickers on Wall Street to the political pundits on our news channels, predictions offered with great certainty and voluminous justification prove, when evaluated later, to have had no predictive power at all.”
Andrew Hacker’s review of Silver in The New York Review of Books caught my attention when he questions James Weatherall’s intention as author of The Physics of Wall Street and exposing a different expectation.
“…the assumption that the quality of our thought can be enhanced by new methodologies.”
Certainly, Hacker’s impressive eloquence helps; but invoking quality in reference to thoughts struck a visceral chord. Variety and range implied by differences in quality intrigue us. They make the world more interesting. At the most basic level, variety compels trade and incites desire for around diversity. Frequently, recombining ideas defines innovation but does either necessarily signify progress, reflect higher power thinking, or even spread benefits more widely?
Variety in objects or tangible goods naturally reach their limits and so too does our tolerance for diverse ideas. In products, declining sales makes the limit recognizable in hindsight. In ideas, their displacement provides some evidence of their limited appeal as in the transition to capitalism in the communist bloc or the return of Islāmic fundamentalism in the middle east.
Does a valued quality suggest our preference associates with a higher ranking of an object or an idea? Naturally, higher ranking or rating indicates higher preference, especially when done consistently. For example, measuring liquid in litres vs. quarts does not enhance or detract from the quality of the liquid, the measure and the liquid’s qualities are independent of one another. In the US, quarts are the culturally preferred volume measure and it persists for numerous reasons, some irrational, but few suggest higher power thinking.
Of late, I am reading Scott E. Page’s book entitled The Difference. He provides a series of examples to show the added value produced when multiple perspectives and varying rule based approaches test a situation. Page’s training draws on the work of social scientists in multiple disciplines and his examples, by design demand minimal mental arithmetic and can easily be scaled. His fundamental premise challenges higher order thinking as the ultimate value varying diversity, flexibility and adaptability as ultimately more useful.
Then again, utility or use as an idea in spite of its competitors continues to prove itself resilient over time and earthly situations. I’m OK with some mystery, the unknowns that both Nate Silver finds challenging and James Weatherall believes his approach can resolve. Big data regardless of the measurement methods, analysis models and their possible recombination, I’m betting that diverse human preferences for truth will continue to prove self-limiting. That’s what ultimately makes life and all its diversity interesting!