Uncertain Probabilities

Nate Silver became a prominent political forecaster with his wildly accurate prediction of Obama’s 2012 Presidential victory. His web site, FiveThirtyEight, gave Obama a 90% chance of winning the Electoral College. He also correctly predicted the results of the Presidential election in every state that year.

Silver’s political predictions are always framed with probabilistic language. Through most of 2016, FiveThirtyEight gave Clinton a reassuringly high chance of winning the Presidency. But even then Silver reminded his audience that a high probability is no a sure thing. The chances of surviving in Russian roulette may be quite good, but no reasonable person risks their life on good odds.

By election day, FiveThirtyEight gave Clinton somewhere around an 70% chance of being elected our next President. 70% means that seven times out of ten, in a similar situation, she would win the election. But seeing as there is only one election, a probabilistic prediction can be right and feel wrong at the same time. Technically, a probabilistic prediction can be considered correct in any outcome as long as the odds were between 1% and 99%.   The only way to judge the ‘accuracy’ of a prediction is over time and is of little consolation in this world of singularity.

In his thought-provoking book THE SIGNAL AND THE NOISE: WHY SO MANY PREDICTIONS FAIL—BUT SOME DON’T, Silver writes: “The amount of information was increasing much more rapidly than our understanding of what to do with it, or our ability to differentiate the useful information from the mistruths. Paradoxically, the result of having so much more shared knowledge was increasing isolation along national and religious line. The instinctual shortcut we take when we have ‘too much information’ is to engage with it selectively, picking out the parts we like and ignoring the remainder, making allies with those who have made the same choices and enemies of the rest.’ (page 3)  

Silver is referring not to the internet, but to the invention of the printing press, which he claims was one of the prime contributors to the next 200 years of wars between religions and nationalities. Whether historians would agree or not, his observation that more information can actually lead to a narrowing of perspective rather than a broadening, feels true to our time of false news cycles, tweets and information hacks.

In THE SIGNAL AND THE NOISE, Silver points out the poor track records of most predictions: many ‘expert’ predictions are ‘barely better than random chance’ and the clarity and specificity of a prediction may have a negative correlation to its accuracy. Silver engagingly educates us in the ways of uncertainty, risk, chaos and complexity. Reading the book, we can become better consumers of information and perhaps even a little more at home in this probabilistic and unpredictable world.