I’ll confess, while not particularly invested in the outcome of the presidential race, I shared the amusement of my Democratic friends watching staunch conservative pundits doing their best impression of the former Iraqi information minister in the weeks before the election. We saw a string of prominent conservatives confidently projecting landslide victories for Mitt Romney, and waving aside all the (as we now know, highly accurate) statistical models projecting a solid Obama win on the grounds that “all the vibrations are right” for Republicans. That said, given the role that partisan pundits play, I can’t really say they were wrong to do so. Indeed, the next time the best available models project a clear Republican win, their Democratic counterparts would probably be wise to do at least a little bit of the same thing, and gin up reasons (however spurious) to think the polls got it wrong this time.
Ideally, professional pollsters have no particular agenda beyond accurately forecasting the outcome of a race. But pundits are trying to influence outcomes, and forecasts don’t just predict outcomes, but at least partially help to determine them. There’s plenty of social psychology literature showing bandwagon effects in elections: Voters on the fence often pick the candidate they expect to triumph anyway, because it’s nice to be on the winning side. Campaign workers become demoralized if they think they’re laboring those long hours for a hopeless cause. A 20 percent chance of victory is still a chance, after all, and you don’t want people throwing in the towel prematurely. Here as in many areas of life, when the odds are heavily against you, being a perfectly accurate assessor of your chances can actually make the odds worse. If you are rational, you will want to have some irrational beliefs. So I don’t expect supporters of a candidate who’s unlikely to win on election eve to acknowledge this, any more than I expect the coach of an underdog team to deliver out an honest read of the stats as a pre-game pep talk. We don’t make fun of coaches for this, because we all understand they’re engaged in a bit of socially appropriate bullshitting.
Which is all well and good when it’s just the pep talk—especially on election eve, when it’s too late to alter a losing strategy. The danger, of course, comes when the coaches start believing their own pep-talks—or as Chris Hayes (channeling Biggie) puts it, when you’re actually so high on your own supply that you start rejecting negative information generally, even when adaptation is still an option. One contributor to this is surely the epistemic bubble created by an increasingly complex, interconnected, and self-sufficient conservative media ecosystem. But there are other factors that probably increase the tendency too.
Pundits are most often, at least initially, writers—which is to say, storytellers. So when polls were less frequent, less accurate, and less widely disseminated, it wasn’t much trouble to weave a story about how polling that looks bad for your team has failed to capture some ineffable factor (“the soul… is not so easily number-crunched“) or was conducted too early to account for some supposedly game-changing news story like Benghazi. A compelling storyteller might even feel free to ignore the polls altogether.
Now, though, we’ve got a larger array of polls, as well as increasingly sophisticated tools for aggregating and weighting them in statistical models that can also factor in external data about things like economic performance. Our predictive tools are better, and our media environment guarantees that the outcomes they project will get wide circulation well in advance of the election. A decade ago, it would have seemed bizarre for a statistician to be one of the most controversial and high-profile figures of a presidential campaign, partly because nobody would have found it plausible that race polling roughly even nationally could be called for one candidate at a high degree of confidence weeks in advance. Even if it could, a political media dominated by a few networks and major papers would have found it convenient to give those projections less attention and play up the sexier “dead heat” narrative. Players’ morale aside, nobody much wants to watch a game whose winner is a foregone conclusion either. Ignoring and slapdash storytelling aren’t really viable options anymore.
The thing about pep-talks, especially when you’re not hearing them live from a charismatic speaker in a small room full of comrades, is that they don’t work nearly as well if they’re obviously just pep talks. It is rational to have some irrational beliefs, but it’s more or less definitive of belief that you can’t consciously recognize it as irrational while you hold it. So in an age of modeling, a little offhand bullshit no longer cuts it: Now you need elaborate counter-models, a-la UnskewedPolls, or at least some more facially compelling argument for why all the models are systematically biased and untrustworthy. What was once the self-contained rational irrationality of a few election eve beliefs metastasizes and builds its own supporting superstructure, in much the same way as one lie can give rise to a whole elaborate network of lies designed to cover it. Or, for the logicians, the way the principle of explosion dictates that a set of propositions containing a single contradiction can spawn an infinity of falsehoods. At some point, the cost of this cognitive infection becomes too high, and a bit of rational irrationality becomes plain old irrational. The belief system you have to sustain to maintain your election-day optimism becomes an obstacle to strategic adaptation that might actually justify that optimism.
A resounding electoral defeat focuses the mind wonderfully, though, and while no doubt we’ll see the most thoroughly embubbled floating dark speculations of some massive conspiracy to rig the election, conservatives who don’t like losing are probably going to get a lot less dismissive of statistical modeling pretty quickly. Mocking the scientists is all fun and games when it’s abstract evolutionary theory, but most people don’t want to refuse lifesaving biotechnology. They may also become somewhat more skeptical of large swaths of the conservative media ecosystem whose primary function is not, in fact, to achieve electoral victories, but to attract eyeballs and extract revenue from conservative audiences. Those goals aren’t necessarily complementary: Ideological publications often do better when the other side is on the ascendant.
In the longer term—notwithstanding the massive and massively daunting security and anonymity challenges it poses—the real solution to this tension (and the ludicrous lines at polling places around the country) is an Internet voting system that lets every citizen with a modern phone—or just access to a public library—cast (and change) their ballot anytime between the conventions and “election day” as easily as they’d update a Facebook status or Instagram a kitten. The results probably become obvious a lot quicker—with a bit of volatility as groups voters decide they can individually afford a third-party protest vote, then realize that collectively they can’t—but the cost of action falls so low that morale concerns are far less significant. And on election night, my journalist friends can all begin drinking much earlier.