Uncertainty is unpleasant and we work hard to talk it away

As a statistician, I make use of the mathematics of probability and variation every day. Outside of my work, though, I am very uncomfortable with uncertainty, whether in the personal or political realms. And I think most people feel the same way. Even in areas such as scientific research or economic and political forecasting where uncertainty is inherent (if you knew what would happen in science, you wouldn’t have to do research; if there was no future uncertainty, there’s be no need to make a forecast), a lot of effort gets put effort into avoiding or denying uncertainty, a “premature collapsing of the wave function,” to use an analogy from quantum physics. When a “statistically significant” result in an experiment is reported as a discovery, or when a non-statistically-significant difference is reported as a null result, this is a denial of uncertainty.

Collapsing of uncertainty reduces mental tension: it’s work to hold two conflicting ideas in your head at once, and a relief to be able to choose just one—especially if you are persuaded that this choice is justified by science. Hence the appeal of making strong statements, which also can get you some attention and respect if stated with enough of air of authority.

But then if your predictions get tested against reality, things can go wrong. Embarrassments arose when prominent economists predicted a recession in 2023 which then never happened, and in the 2016 election campaign when political journalists gave Hillary Clinton a near-certain chance of winning in the Electoral College, based on her narrow but consistent polling lead, not accounting for the possibility of systematic correlated survey errors. Discomfort with uncertainty is all too human, and it also comes with social costs.

The replication crisis in science is a product of systematic discomfort with uncertainty, with speculative results presented as settled fact, leading to distress when these findings do not hold up in later experiments: a scientific and emotional boom-and-bust cycle. Indeed, classical statistical methods seem almost designed to create this boom-and-bust behavior, when non-statistically-significant results are treated as if they were zero and statistically-significant results are taken at face value. I prefer a Bayesian approach in which estimates are partially pooled toward zero. There are also non-Bayesian statistical methods that do this sort of regularization. Whatever your statistical methods and philosophy, I recommend studying the process, not just the particular dataset. This is called the reference set in frequentist statistics or averaging over the prior in Bayesian statistics. In poker terms, the principle is to evaluate the play, not the strategy.

Those unusual people who thrive on uncertainty

In his new book, “On the Edge: The Art of Risking Everything,” Nate Silver explores the world of some people—notably, poker players and tech investors—who do not just accept uncertainty but thrive on it. Silver and these others thrive on uncertainty, directly because they can separate their wishes and fears from their assessments of risks and returns, and indirectly because everyone else’s irrationality leaves a lot of potential gains on the table. The messier and more complicated the game, the bigger advantage it is to have a cool head. I’m reminded of a passage from Frank Wallace’s classic “Poker: A guaranteed income for life by using the advanced concepts of poker” (which is pretty much devoid of any intentional literary merit but on its own terms is a kind of outsider-art masterpiece), where he advices the would-be sharpie to encourage games with lots of wild cards and unusual twists, because the more uncertainty, the more the suckers will get confused. Unlike Wallace, Silver is writing his book with altruistic goals: rather than offering readers tips on how to fleece the rubes, he wants to help them understand the world.

Silver characterizes as living in a conceptual country he calls “the river,” by analogy to riverboat gamblers, and, along with many interesting stories, he shares “Thirteen habits of highly successful risk takers.” The concept of “risk taker” is subtle, though: as recommended by the Kelly criterion of gambling, the optimal level to risk in a bet is proportional to your total financial resources—a point which is not invalidated by legendary gamblers who’ve gone to zero and picked themselves back up, as what is relevant here is lifetime resources, not just the stack you’re sitting behind at the poker table. The point is that being comfortable with risk is not the same as betting wildly. For that reason, I think the book’s title is misleading. A successful gambling strategy is not to live “on the edge” or to “risk everything” but rather to be aware of where the edges and risks are and act accordingly.

Vibes

One fun aspect of the book is the stories about some interesting people I never otherwise would’ve heard about. I was bored stiff with the retelling of the overexposed Sam Bankman-Fried—I don’t really need to read an analysis of four different theories of that pretentious plutocrat—but that stuff is easy for a reader to skip. Only in Nate’s book would we get a fun mini-anecdote about “a hand played by the legendary poker player Tom Dwan on the Hustler livestream, where he’d correctly called a bluff against an opponent named Wesley in a record-breaking $3.1 million pot. Wesley’s behavior had been similar to Friedrich’s: a quick all-in followed by a turtle shell in an extremely high-pressure moment where it might be hard to conceal your emotional state.” And Nate buried that one in a footnote! I love this sort of thing, which gives me the feeling that he was pouring everything he had into the book, not holding back.

Here’s another throwaway: “There is almost never such a thing as a sure thing. Even when you’re literally cheating, it may not be a sure thing. In the infamous Boston College basketball point-shaving scandal of 1978-79, when two teammates were paid by the mob to throw games, only four of the nine games the mob bet on won money, with three losses and two pushes.” Boston College . . . the mob . . . hapless crooks . . . this is prime George V. Higgins territory!

The flip side of Nate not holding back is that sometimes he writes things that seem like clichés. Here he is talking about an artist who was “in the right place at the right time” and made a ton of money from non-fungible tokens: “When I spoke with Winkelmann—a.k.a. Beeple—I was expecting someone with the self-important air of being a serious artist or at least someone whose success had gone to his head. Instead Beeple was extremely down-to-earth, dropping f-bombs about once every fifteen seconds in a thick Wisconsin accent.”

Hey, wait a minute! The macho regular-guy artist . . . that’s standard operating practice in the art world. You’ve heard of Jackson Pollock, right? Who was from Cody, Wyoming—that’s even more earthy than being from Wisconsin. Silver’s quote reminds me uncomfortably of the story from Freakonomics of an unnamed “academic” who says something stupid, only to be shot down by regular-guy “Chuck Esposito, a genial, quick-witted and thoroughly sports-fixated man who runs the race and sports book at Caesars Palace in Las Vegas,” which in turn reminded of the punchline of that joke from grade school: “Hey, man, the smartest guy in the world just jumped out of the plane wearing my backpack.”

I’m not saying that Beeple isn’t a smart, down-to-earth guy; I’m just resisting Nate’s dressing him in rogue’s clothing. In some ways, this sort of thing is almost necessary: Nate’s giving us a tour of a world that he loves, he’s writing about to his friends, and so of course he wants to present them in a positive light. If Nate were to describe Beeple as, for example, “the typical self-important ‘serious artist’ who signals his regular-guy status by cultivating a thick Wisconsin accent and carefully dropping f-bombs into his conversation,” well, what would be the point of that? It’s kind of like sportswriting: with rare exceptions, we like the athletes to be presented in a positive light. Similarly, we are introduced to “Will MacAskill, a boyish-looking Oxford professor of philosophy with an endearing Scottish lilt.” I have a horrible feeling that if I’m not careful I’d be described as a “shifty-eyed academic who speaks in a nasal east-coast suburban accent,” rather than, say, a “charming gray-haired statistician with many of the attributes of the absent-minded professor.”

Attitudes toward risk

To be fair, though, the above sort of descriptions work with one of the themes of the book: Nate’s take on Silicon Valley is that “to make all of this work, you need a symbiosis between two often-clashing personality types . . . risk-tolerant VCs (investors) . . . and “risk-ignorant founders.” This is an interesting idea that goes well beyond business-book platitudes of risk-taking being good in itself—one of my pet peeves is when funders say they want to support “high risk, high return projects,” but then in the proposal they never want you to talk about the risky part, and I don’t think they really want high-risk, they just think it sounds good—instead distinguishing between different attitudes toward risk: there are normies who don’t want risk at all and will pay a high price to avoid it; there are savvy forecasters who try to accurately assess risks and then make optimal decisions; and there are the obsessives who try things nobody else would do. The savvies and the obsessives are similar in their willingness to jump into the unknown—and, according to Nate, they get along pretty well, considering each other as fellow citizens of the River—but they thrive on risk in different ways.

I think all of us have all three of these attitudes in different aspects in our lives. At a personal level, I’m a risk-fearing normie: I treasure my secure job and secure family life, and I don’t borrow money. When doing statistics, I’m a savvy forecaster; assessing uncertainty for inferences and decision making is central to my theoretical and applied work. In my career, I’m an obsessive: I wrote Bayesian Data Analysis back when people were telling me that it wasn’t a savvy move, and more generally I throw myself into what I want to work on, and I try to persuade others to join me. There are other obsessives whose careers have not gone so well, which fits Nate’s typology that risk-ignorant founders can create social good even if they’re not making personally optimal decisions. The other point here is that these three often-clashing personality types don’t just power Silicon Valley; in some mix or another they power all of us. As memorably dramatized in the Inside Out movies, each of us is a sort of committee, and Nate’s portrayal of different people’s attitudes toward risk can perhaps give us a better understanding of our own incoherent selves and how this incoherence can be a strength.

Bringing this closer to home, this all reminds me of statistical workflow, how we bounce back and forth between scientific hypothesizing, design of measurement, data gathering, statistical modeling, and comparison and checking of models. When forming hypotheses and constructing models, it helps to have a damn-the-torpedoes “founder” attitude: don’t think about the risks, just do it. When designing an experiment and assessing model fit, it’s good to be risk-aware—not risk-averse, but clear-eyed about uncertainty. Tons and tons of unreplicable science has occurred from thoughtlessly-optimistic designs and analyses. By analogy to Nate’s description of Silicon Valley, empirical research needs the obsessive attitude for modeling and the savvy attitude for design and analysis. I’m not sure where the normie, risk-averse attitude comes in, but given its omnipresence in human behavior, I expect it serves some useful psychological purpose in keeping us sane, if for no other reason than that denial of uncertainty frees up our brain cycles to think about other things.

Ethics

A theme running through the book is the ethical question. If you’re a River dweller and can thrive under uncertainty, what do you do with this ability? I would say that Nate employs it for social good: for over a decade he’s supplied trustworthy forecasts for elections and sports, he’s participated in public debates, and he’s published two books which allow ordinary readers inside some important communities. Nate’s also made some money along the way, which is fine: we live in a capitalist society, and someone has to pay the bills for all these public goods. Some of the people promoted in the book though, I’m not so sure about. For example, Nate interviews Gary Loveman, an executive at Caesars who is famous for implementing schemes to extract fortunes from gambling addicts. Now, you could argue that Loveman just plays a role in society—if he weren’t destroying these people’s lives, it would be someone else—but that would contradict the other part of Loveman’s story, which is how innovative he has been. If you want to give credit to Loveman for being such a trailblazer, then he deserves the blame for the consequences of his actions, no? I don’t think Nate disagrees with me in general terms—he writes that when playing slot machines in Vegas, “one thing is certain: in the long run you’re going to lose,” and he writes that “there’s something about slot machines where the transaction between casinos and their customers feels fundamentally unfair.” This just didn’t come up when he’s interviewing Loveman.

Again, I’m not pure myself—just look at the list of sponsors of our research and I expect that, whatever your political persuasion, you’ll find two or three that you’ll object to—so think of the above paragraph not as a moral objection but as an intellectual objection. The River includes scam-infested waters, from Vegas casinos to electricity-guzzling cryptocurrencies. Arguably, scams are as an important aspect of the River as corruption is to the Village, and to not fully explore this is a missed opportunity, in the same way that it would be a missed opportunity to write a book about applied social science without checking in on the various prominent examples of fraud and unreplicable research that have been enabled by the culture of the scientific establishment.

Politics

Nate’s book is mostly about people who thrive on risk, and, as discussed above, I found lots to think about. He’s got good stories, interesting ideas, a personal connection, and a sense of history and the interconnectedness of society. The ranking of poker hands on page 42 was familiar to me, but even there I learned something—apparently, poker-heads write a 10 as T, for example, illustrating a full house with TTT88. I’m guessing that this is standard notation now, and I’m wondering if it has something to do with modern poker thinking in which a 10 is just about as good as a face card? Or it may just have something to do with computerization: with so much of poker being online, it’s convenient to always use a single character to indicate the denomination.

Nate situates his ideas about risk into a theory of politics: as he puts it, successful governmental policies have “encouraged calculated risk-taking.” This makes sense, and it’s something that policy-makers have thought a lot about, not always with much agreement, as can be seen from the continuing debates about international development during the past several decades. There are no easy answers here, and I respect Nate for bringing up political issues rather than just treating risk-taking as a matter of individual behavior.

There’s one part of Nate’s book that puzzles me, though. He focuses on the River, which he characterizes as “an ecosystem of people and ideas” who “speak one language with terms such as expected value, Nash equilibrium, and Bayesian priors.” I’m pretty sure that I live on the River! I enjoy poker (OK, I haven’t played it recently, but still), I “have close ties to the tech sector,” and I’m one of those people with an “appetite for involving themselves in all sorts of problems and controversies.” Indeed, I think my River-dwellingness makes me a good reviewer for this book.

Nate contrasts the River with the Village: “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.” Despite working in the news media himself, Nate writes, “I’ve never quite taken to the Village,” and I kind of know what he means, as I feel that way about academia, which can feel cringey in a way that isn’t quite what you see on the River.

OK, so here’s my problem. If the River is risk-takers who are skeptical about politics, with views ranging from the far left to the right and everywhere in between, and if the Village is normies who are center-left and who have some connection to political power or at least would like to wield it, then . . . where’s everybody else?

There are two kinds of “everybody else” I’m thinking about here.

First, there are all the people in the country and around the world who aren’t well connected. They’re not professional gamblers or hedge fund investors or tech developers or election-forecasting statistical analysts (hi!), nor are they politicians, journalists, Ivy League professors of cultural studies, or even bloggers. They’re just regular working stiffs. I’m not saying Nate should write a book about these people; it just seems important to recognize that the River people and the Village people he describes are two small tribes within a large society. To put it in crude terms, the Riverians want to make money from the masses and the Villagers want to boss the masses around. That’s a big area of disagreement (notwithstanding that many River dwellers want to boss the masses around and many Village residents would like to make some money); the key is that the battle between the River and the Village is taking place among the larger population and the larger economy.

The other group of missing people are . . . approximately half the people who run the country. Where’s Ted Cruz, for example? He doesn’t seem like a gambler, and he spends much of his time trying to tell people what to do. He works in that small village of Washington, D.C. But he doesn’t have “distinctly left-of-center politics”! And then there are the six Republican judges on the supreme court: not river-dwelling gamblers in any sense of the phrase, these judges have the most risk-averse career paths imaginable—but they’re not in Nate’s Village, as he has defined it. Now, you might say that individual Villager can lean right even as the Village as a whole has distinctly left-of-center politics, but that doesn’t really work given that the Republicans control the judiciary, half of congress, and a few years ago the executive branch of government as well. I think the only possible solutions here are: (a) redefine the Village as being divided between the two parties, or (b) define the Village as having left-of-center politics and then introduce a new entity—the Fort?—that has all the normie, risk-averse characteristics of the Village but on the political right. Ted Cruz and Samuel Alito are comfortably situated within the Fort.

I kind of like this River/Village/Fort thing. The Village is fighting against the Fort, and the River has to figure out what to do about it. Riverians are split: those on the left are in favor of growth in an environment of political equality, and they support higher taxes on the rich, government action on climate change, abortion rights, and other aspects of the Democratic platform; while those on the right are in favor of economic freedom and social stability, and they support lower taxes and spending, less business regulation, restrictions on abortion, and other Republican policies. People of all political persuasions can hate the River because they can find a group of people on the River who annoy them. If you’re on the right, you can hate smug successful Riverians who want to kill the capitalist golden goose that created the conditions for their success; if you’re on the left, you can hate greedy Riverians for whom all the world’s riches aren’t enough.

Who would be the Riverians’ ideal politician? There’s the Libertarian Party but they never get many votes, even in the seemingly ideal year for them of 2016. Considering major-party figures, I’d have to say the closest to a Riverian would be . . . Mitt Romney. He’s a risk-taker, he invested in businesses, he was a cross-party success as governor of Massachusetts, and his notorious “47%” statement (“There are 47 percent of the people who will vote for the president no matter what. . . . are dependent upon government . . . believe that they are victims . . . believe the government has a responsibility to care for them . . . these are people who pay no income tax”) is a hard-headed statement that sounds very River: it might not be how we want things to be, but it’s how things are.

I don’t have the impression that Romney had anything close to universal River support, however, and I suspect the reason is that in 2008 the River went all-in (as Nate would say) on Barack Obama, who was Romney’s opponent in 2012. Obama, as a liberal law professor and Democratic politician, is Village through and through, but he has various Riverian vibes: he’s a basketball fan, he’s a technocrat, and you get the impression that, if he doesn’t himself talk about “expected value, Nash equilibrium, and Bayesian priors,” that he’s friends with some people who do. Obama shares some characteristics with tech investors (who can also talk the talk of expected value and Bayes, even if they’re not actually crunching numbers themselves)—he’s even friends with Richard Branson!, and you can see how Riverian economics professor Steven Levitt and occasional climate-change denier wrote that in 2008 he thought Obama “would be the greatest president in history.” By 2012, much of that bloom was off the rose, but perhaps the Riverians of the center and right didn’t realize what they had in Romney.

Statistics and poker

One thing the book made me think about are the differences between what it takes to do statistics (or data science or “analytics”) and what it takes to play poker. Here’s my take: to be a good analyst, you need:

(a) some understanding of math and probability,

(b) the willingness to learn from data—to use data and analysis to come to new conclusions rather than just to confirm your existing beliefs—and

(c) the ability and willingness to apply subject-matter knowledge.

To be a good poker player, you need:

(a) and (b) as above,

(d) the ability to practice deception, and

(e) the ability to “read” people when they’re trying to deceive.

I’m a good statistician (or so think the people who pay me!) and I think I’m very strong on all of (a), (b), and (c) above. I’m just an OK poker player—I’m sure they’d clean me out at Vegas in short order—OK because I’ve got (a) and (b), but I’m weak on (d) and (e).

I don’t really know, but I think that Bill James is like me: good with (a), (b), (c), but not so much with (d) and (e). I’d want him as an analyst on my baseball team; I wouldn’t put him in charge of negotiations.

Nate is good at all of (a), (b), (c), (d), and (e), which has allowed him to become a world-class analyst and a world-class poker player. I do wonder if his strengths in all of these traits causes him to somewhat blur the boundaries between them.

In particular, I might argue that being bad at (d) and (e) might make me and Bill James better scientists than we would be, if we happened to be good at deception. I don’t say this because being good at deception means we would cheat—I think Nate’s good at deception in poker but I don’t think he cheats in any way in his analytical work—, but rather because, if you’re bad at deception, this kind of forces you to a more direct and transparent approach to science. There’s something brute-force about how I do science, and I think that brute force here is a good thing; I’m not in the habit of hiding things and so everything ends up naturally out in the open.

Or maybe this is just a rationalization. The relevant to Nate’s book is that there are different aspects of “Riverian” comfort with uncertainty. One aspect is the Riverian analysts’ traits of comfort with probability, willingness to learn from data, and integration of subject-matter understanding into analytics. The other aspect is the Riverian poker players’ traits of deception and ability to read through others’ deception. In mathematical terms, the (a) + (b) + (c) is decision theory and (a) + (b) + (d) + (e) is game theory. I don’t know how different these two sorts of River dwellers are; it’s just interesting to think that there are different ways in the River to be comfortable with uncertainty.

Summary

I enjoyed Nate’s new book. It had lots of great stories and an interesting take on risk attitudes, not just in themselves but how they interact. The book feels very honest; it seems like it’s written from the heart. I don’t think the political angle was fully thought through, but with a small extension (from River/Village to River/Village/Fort), it kinda works. The book is a kind of cultural tour of some people who thrive on uncertainty—Nate focuses not on people who take risks out of desperation but rather people like him (or me!) who would do just fine in the Village or Fort but choose to enter the River because it’s more exciting or just because it feels more real. How do these unusual people fit into the rest of society? It’s an interesting question that Nate takes some interesting steps to answer.