Book Review: Nate Silver’s “On the Edge: The Art of Risking Everything”

Just in time for the coming US presidential election, FSG’s Patrick Marren warns about prediction in situations of strategic consequence in his review of Nate Silver’s On the Edge: The Art of Risking Everything.

Nate Silver’s last book, The Signal and the Noise: Why So Many Predictions Fail – But Some Don’t, contained the following statement in its Introduction to the 2020 edition:

“We need to stop, and admit it: we have a prediction problem. We love to predict things—and we aren’t very good at it. … If prediction is the central problem… it is also its solution.”

This is, of course, gigantically, dead, wrong. About which more below.

Nate Silver may have been reacting to the intensely negative response to his having said Hillary Clinton had a 71% chance to win the 2016 election. The many Democrats who had bought his book in 2012, around the time he had correctly predicted the winners of every state in the presidential election (he had picked 49 out of 50 in 2008), were dismayed by their seeming oracle’s “failure” in 2016. They had come to see him as a security blanket in times of uncertainty, and two straight Democratic presidential victories caused them wrongly to believe that Silver saying a candidate was more than 50% likely to win was a guarantee. 

Silver did not like the harsh reaction to his “failure,” and he seems, in his new book, On the Edge: The Art of Risking Everything, to have decided that the real problem is that elites in the world are divided into two types of people: whining innumerate liberal elitist woke types (“The Village”); and daring bold libertarian-leaning gamblers (“The River”). 

Nate Silver proudly claims membership in the club of “Riverians.” “People in the River trust me to tell their stories because—let’s be honest—I’m one of them. Their way of thinking, for the most part, is my way of thinking.”

That way of thinking consists of the following attributes:

• “Analytical” – Silver says this is “to resolve something complex into simpler elements.”

• “Abstract” – “trying to derive general rules or principles from the things you observe in the world.”

• “Decoupling” – “the ability to block out context…the opposite of holistic thinking. It’s the ability to separate, to view things in the abstract, to play devil’s advocate.’”

• “Competitive” – “[P]eople in the River are often intensely competitive.”

• “Critical,” “Independent-minded” – “[B]eing critical of consensus thinking, often to the point of being contrarian.”

• “Risk-tolerant” – “[B]eing willing to break from the herd… openness to experience and low levels of neuroticism.”

Nate Silver says, “People in the River are—for better or worse—my kind of people. Conversely, I’ve never quite taken to the Village, and I’ve often felt like media coverage of me and FiveThirtyEight was misinformed, particularly after the 2016 election.”

The reader might be forgiven for thinking that “misinformed” reaction was the impetus for this entire book.

Silver’s “Village”:

“…consists of people who work in government, in much of the media, and in parts of academia (although perhaps excluding some of the more quantitative academic fields such as economics). It has distinctly left-of-center politics associated with the Democratic Party.”

Nate Silver says the River is winning. Wait, what?

In Silver’s judgment, “The River is winning.” And I think he is right. 

But this fact encapsulates the problem with Silver’s analysis. Because “the Village” itself (government, media, and academia) is unquestionably being swamped by “the River.” Quantitative approaches to future uncertainty have run rampant in all these “Village” realms. “The Village” is not only losing; it’s being taken over and transformed by “the River.”

But really, we are all losing, because Silver’s dictum from earlier, that prediction is the solution, is dead wrong. 

Predictions do not fail us because they are somehow inaccurate. Silver’s about as good as it gets in the prediction business. His 29% chance of Trump winning the 2016 election was closer to the outcome than almost anyone else’s. 

Prediction fails us because, as Silver found out after November 9, 2016, it cannot help us with the most important problems we face. Prediction is not what we need there. IMAGINATION is what we need.

Every prediction makes gigantic assumptions. It pre-defines the possible outcomes. Will Hillary Clinton beat Jeb Bush in 2016? Will Joe Biden beat Donald Trump in 2024? 

And then reality happens, because prediction MUST wear blinders. It cannot abide a broader vision. It MUST be “abstract:” it MUST eliminate most possibilities; it MUST be “analytical”: “resolv[ing] something complex into simpler elements.” It can’t embrace the entire universe of the plausible. It blinds us. 

That’s why in 2024, Silver had to completely re-rack his entire model once Joe Biden dropped out of the race. Imagine what he would have had to do if Trump had been one inch further to his right on that fateful day in Pennsylvania. Could his “Riverian” approach have been of any use in that case? No, it could not.

The context really matters.

And that brings me to my main objection to Silver’s book. 

“Riverian” approaches work great in certain well-defined contexts: poker tournaments; baseball seasons; astronomical observations; other “hard-science” applications. They are definitely necessary in lots of everyday applications: dealing with the immediate consequences of changes in the world, e.g., weather events,  sudden changes in interest rates, revelations of changes in reliability of machine parts. Quantitative approaches are actually absolutely necessary in those areas. 

But few of those things really qualify as strategic. If something is predictable via Bayesian statistical approaches, guess what? It tends to be trivial, or at best tactical. It is only the things that are fundamentally unpredictable – where you cannot even be sure of what variables or operands your predictive model should contain – that  are of strategic importance. (The economist Frank Knight called it “the higher uncertainty,” which he distinguished from “risk,” which is quantifiable.)

And that is why “The Village” persists, to annoy “Riverians” like Silver. It’s not that the Village simply refuses to understand mathematics out of pique. It’s that the biggest problems in the world are, and possibly always will be, completely unsusceptible to “Riverian” approaches. They cannot be predicted; they can only be anticipated via imagination. So the Village, with its damnably “coupled,” “politicized,” non-numerical, compromised, big-picture approaches, is the only community that can deal with the biggest life-and-death issues we have to face. 

And even when predictive models can be constructed for fundamentally uncertain issues – like Silver’s models of the 2024 presidential race – what are they actually doing for you? NOTHING.

Strategy is about decision-making. What decision could you make based on Silver saying Donald Trump, on September 17, 2024, has, say, a 54% chance of winning the election in November? (I just made that number up, though it’s probably not far off his estimate.) 

Nothing! No decision. For decision-making purposes, you’d better be planning for both plausible outcomes – or possible third outcomes. Which means that instead of applying Bayesian forecasting methods to this race, taking its temperature every day, and being completely paralyzed and fixated on that swiveling dial, we all should be thinking deeply about what one or the other candidate winning (or neither!) might plausibly mean for us.  

“The River” is terrible at this. They want to abstract away all complexity and make a bet, because “the River” is a bunch of what Silver calls “degens” – degenerate gamblers. So we saw the 2008 global financial crisis arise almost purely out of increasingly abstract arcane quantitative risk management approaches, crowding out imagination that might have anticipated disaster. 

And despite that failure and many more, quantitative predictive approaches are metastasizing throughout our economy and society. On-line gambling, for example, is mushrooming right now, and having a nontrivial negative impact on our economy. 

“The River” is winning, all right – like a cancer. A natural mechanism that has jumped out of its proper context, and is now replicating itself mindlessly throughout the patient’s body.

Rigorous imagination

If 2024 has taught us nothing else, we should now be well aware that we cannot predict. But it has also taught us that prediction is like a drug. Silver’s book, half of which is about gambling, will sell like hotcakes, even though it may actually do far more harm than good. 

And that is because prediction is addictive. “We love to predict things.” But prediction is not a solution to our real problems; in many cases, it IS the problem. 

Only rigorously applied imagination – as FSG practices in scenario planning ­– can help us to anticipate and plan for the fundamentally uncertain. 

But imagination, by its very nature, cannot be modeled. So it can’t be made into a drug. It’s spinach, not heroin. 

But unlike “the River,” it’s good for you. 

***

Patrick Marren is an FSG principal based in Auckland, New Zealand. He is the author of the forthcoming Fatal Certainty: How a Cult of Prediction Made the 21st Century an Era of Strategic Shock, and How Rigorous Imagination Could Bring Us Back.