In 2016, Democrats were fairly confident of a resounding win over Donald Trump, a lying, racist sexual predator nominated for the presidency by the GOP. Instead, Trump eked out a victory by slim margins in a few key states. In 2020, Democrats were more wary of predictions — even after Trump’s impeachment and disastrous handling of a deadly pandemic — but still allowed themselves to dream of a landslide bigger than any they imagined for Hillary Clinton: Maybe Joe Biden would have it wrapped up as soon as the polls closed, romping through North Carolina, Georgia, Florida and even Texas before Trump could fire up the misinformation machine.
The fantasy of a sharp, decisive rejection of Trump died early in the night, just as the blue Rust Belt wall did in 2016. What gave anyone the audacity to hope for it, even despite Trump’s historic strength with his base and Republicans generally, the people who put him in the White House to begin with?
Polls, of course. And not just the largely symbolic national polls, which gave Biden a consistent lead that simply doesn’t translate to the Electoral College, but reams of state polls, and aggregations and close studies of those polls. As the possibility of Trump’s reelection lingered like a heinous fart, the man most associated with those polls began to trend.
After accurately forecasting Barack Obama’s 2008 and 2012 electoral maps, Silver and his analysis blog FiveThirtyEight became synonymous with the art — he’d probably call it a science — of extrapolating likely voting outcomes from polling numbers. Much was made of his “missed” call in 2016, even though he gave Trump better odds of winning than fellow prognosticators; in 2020, his calculations were fuel for the sentiment that Biden could thoroughly embarrass Trump instead of just barely defeating him, but his final pre-election column warned of the “fine line” between a blowout and a nail-biter, and he still gave Trump a 10 percent chance to tip the scales. That is, both then and now, you can’t exactly say that Silver was “wrong,” since he’s not in the business of reporting the future. He projects thousands of different scenarios (40,000 simulations, in this case), and if any of them match reality, then his model can be said to “work.”
Silver’s usual complaint is that his data is interpreted by political observers in ways he doesn’t intend — that polls are imperfect, he’s describing likelihoods (not calling results), and ultimately, we cannot know the ways in which an election will break. Even his book, The Signal and the Noise, carries the subtitle “Why So Many Predictions Fail — But Some Don’t,” a baked-in concession to the fickle nature of human events. But given the many, many caveats that litter his work, we are left to ask what information it actually conveys. Or, more to the point, what effect it has on our understanding of the presidential race. Because you can only blame bad-faith conclusions drawn from your material for so long before it starts to seem as if you, and the way you present this stuff, are the problem. That air of expertise and technical precision.
It always comes back to confidence, the root of the term con man — as in, the trust created to perpetrate a scam. FiveThirtyEight is meant to provide insight into the mechanics of how we pick a president, though in practice it’s a website where armchair wonks can alternately soothe and worry themselves with the false certainty of pure numbers, where they can justify the expectations they probably already had. It incorporates the drastic failures of the polls themselves, contributes to pundit class treatment of democracy as a sport (Silver got his start in baseball stats) and gives ordinary people a counterproductive sense of certainty where no solid ground exists.
That’s what’s so frustrating about his publication: For all that it cautions on margins of error and outliers and the limits of its methods, even a reasonable, skeptical reader gets sucked in and ends up developing an intuitive “truth” of what is destined to happen. At a time of great distrust in media, it looks appealingly sound, rooted in a rare style of objectivity.
And while Silver and the rest of the horse-race coverage keep us fixated on (or hypnotized by) the potential “whats” of the presidential contest, the “whys” and “hows” of the political realignments they try to anticipate are background static. This doesn’t aid anyone’s comprehension, and when the actual vote tally surprises people, they feel all the more confused and betrayed for having wasted hours on the shallowest exploration of a complex history in the making. The only solution is not to be fooled again. Which means accepting that we don’t know what we don’t know.
Okay, meet back here in two years to see if there’s any improvement.