Economics, Poker, Baseball, and Politics
An unmissable subplot to the 2012 presidential election season goes by the name Nate Silver. His public $2,000 bet to Joe Scarborough that Obama would win reelection was just a tiny sliver of what was really at stake for him, of course. He projected state-by-state outcomes with almost reckless specificity, all while keeping a poker face in front of millions of readers on the virtual pages of his New York Times blog, FiveThirtyEight. But he got it completely right.
For political junkies, it should feel like having watched Babe Ruth point to center field before homering in the World Series. A home run is a home run, but it’s something bigger and wilder when you call your shot. Silver’s gamble had a similar-but-nerdy braggadocio. Political projections can be interesting at best, usually. Calculated bravado made Silver’s riveting. Now he’s the other-other man of the hour, with accolades in long supply.
Chattering about what Silver’s game means to the worlds of politics and journalism has run in the background. Scarborough, as everyone’s heard by now, called Silver an “ideologue” and a “joke” for putting Obama’s chances of reelection at around 75% in late October. Silver retorted with his high-profile bet, which in turn earned him a rebuke from New York Times public editor Margaret Sullivan. David Brooks, also with the New York Times, has argued for moderation in the consumption of polling data, largely because “unquantifiable events change the trajectories of tight campaigns. You can’t tell what’s about to happen. You certainly can’t tell how 100 million people are going to process what’s about to happen. You can’t calculate odds that capture unknown reactions to unknown events.” The Christian Science Monitor has meanwhile wondered whether Silver has single-handedly destroyed punditry.
You can probably tell that I’m about to jump in.
Let’s start with what I think Silver isn’t: a scientist. By itself, his bet to Scarborough takes him out of that category. Silver effectively called the election for Obama with a 75% level of certainty. That might sound like a lot, but 75% buys you exactly nothing in scientific circles. Rock up to an academic conference with anything less than 95%–or even 99.5%–and you’d better make sure to clear an unobstructed path to the exits before taking the podium.
Your next stop should be a casino, though. Seriously. At a casino, 75% buys you the house. Professional blackjack players make their living on mere fractions of a point over 50%. Doing what it takes to play at 52% can even get you blacklisted. So anyone who eats thanks to routinized calculated risk–professional gamblers, insurance executives, hedge-fund managers, etc.–will tell you the same thing: Bet the farm on 75%. In the aggregate, 75% bets will make you very rich over a lifetime of risk-taking.
What I know about Nate Silver is only what everyone else knows, which includes the general arc of his career. After studying economics as an undergrad at the University of Chicago, he started his professional life as an economic consultant at KPMG, a job he eventually left to become a professional online poker player and a baseball sabermetrician. Then, in advance of the 2008 presidential election, he started blogging his political projections, got most everything right, and won a contract to do the same in 2012 for the New York Times. Economics, poker, baseball, then politics.
Economics, poker, baseball, and politics all share something fundamental in common. They are all bounded systems. By this I mean that each operates within a closed set of assumptions or rules, which makes each susceptible to certain forms of quantitative analysis. The discipline of economics begins with bedrock assumptions about rationality that channel the extrapolations of its practitioners. Poker, baseball, and politics likewise all proceed according to rules that are prescribed exogenously in advance, and that limit the number of potential strategies and outcomes. You can’t hit a home run in poker. There are no flushes in baseball. There are thankfully–or at least hopefully–no bats of any kind in politics. Because these systems are bounded, their components are knowable. With enough data, their behavior is to a great extent quantifiable and predictable. Not quite entirely, however.
For the sake of contrast, imagine a completely static baseball league. The players never age, never retire, and never change teams. No new players ever join. All of the games are played indoors, on uniform fields, and at uniform altitudes. After a while, there would be little point in watching. Teams would play each other under the same circumstances over and over again, until it would be possible to project the outcomes of every game and of entire seasons, on into infinity. Nothing would change, so there would never be any reason to expect different results. Even what inherent randomness there is in game-play would, over time, balance itself out.
This is where Brooks’ observation comes in: “You can’t calculate odds that capture unknown reactions to unknown events.” What makes baseball watchable for most fans over the course of a lifetime is that reality is not static. The mutability of factors that produce wins and losses is what makes the game exciting. Heavy favorites can lose and underdogs can win because of the unknowable effects of trades, aging, injuries, human psychology, performance enhancing drugs, weather, and any other phenomena that affect the game, but that operate entirely outside the bounded system of its rules.
What sabermetric–and similarly economic–analysis does is determine optimal strategies for playing within a bounded system. It’s about giving yourself the best chance to win in view of your limitations and what you know about conditions. But it can’t account for external shocks to the system–the unknowables that concern Brooks. A freak car accident can end a baseball player’s season. Unperceived risk in mortgage-backed securities can cause an economic collapse.
Brooks’ apprehension toward quantitative models like Silver’s is well founded on this basis, but is ultimately completely off base on another. His admonition that “[y]ou certainly can’t tell how 100 million people are going to process what’s about to happen” misunderstands what Silver does.
In Silver’s case, he assumes that what’s about to happen is going to look pretty much like what’s happening right now. It doesn’t matter “how 100 million people are going to process what’s about to happen”. What matters is how a representative sample of the 100 million processed what immediately just did happen. So long as there are data on the immediate past, prognosticators like Silver can refine and tailor their projections right up to the last moment.
Silver’s predictive model only works within the bounds of the assumption that people will vote at T-1 as they claim they will vote at T-0, however. If some unexpected event intervenes between T-0 and T-1–a natural disaster, a terrorist attack, an assassination, a revelation of criminal activity, etc.–then he’s out of luck. He has no data. There’s been an external shock to his model. With the right framing, I’m sure Silver would embrace this limitation to his methodology and concede this part of Brooks’ point.
Against this backdrop, Silver’s 75% bet should tell us much about his sure-to-expand role in national political discourse. Silver came to the table like a grizzled gambler: part odds-man, part showman. He lacked anything that could be called scientific certainty within his own model, to say nothing of the likelihood of an external shock. Yet he called it. He was already all-in. He earned his place through high-stakes and highly public specificity in 2008, and would lose it anyway if he couldn’t deliver in 2012. What would an extra $2,000 and a wrist-slapping on the back pages of the New York Times matter? The upside was becoming one of the most famous and influential voices in political journalism. With a 75% probability on your side, that’s a wager you always make.
Did Silver single-handedly destroy punditry? Not at all.
*By the way, if you want to know where SLB has been since May, it involves two continents, three countries, more than a dozen states, four universities, and four modes of transportation. We’re back, but on the other side of the world.