Evan has a habit. He’s not ashamed of it, but he doesn’t want to reveal too much about himself, lest his colleagues learn how he’s spending so much of his time. Like so many others, the middle-age software developer can’t look away from the presidential election. But his fixation takes a particular form: with every browser refresh, he hopes math will reveal the future.

Evan is a poll obsessive, FiveThirtyEight strain—a subspecies I recognize because I’m one of them, too. When he wakes up in the morning, he doesn’t shower or eat breakfast before checking the Nate Silver-founded site’s presidential election forecast (sounds about right). He keeps a tab open to FiveThirtyEight’s latest poll list; a new poll means new odds in the forecast (yup). He get push alerts on his phone when the forecast changes (check). He follows the 538 Forecast Bot, a Twitter account that tweets every time the forecast changes (same). In all, Evan says he checks in hourly, at least while he’s awake (I plead the Fifth).

Evan’s been obsessed with elections since elementary school. He remembers coloring in electoral map states with markers when George Bush beat Michael Dukakis for the presidency. But tracking the FiveThirtyEight forecast isn’t just about politics, really; it’s about feelings.

“My emotional state runs partisan,” he says. “I want my side to win.” (He’s a liberal.) “At the same time, if my side loses, I want to get it over with.” But, like everyone else, he has to wait. Until November 9, the forecast is the next best thing to an answer. In a time of great political upheaval and anxiety, poll forecasts have become an emotional anchor. They strip away all the rhetoric and spin, all the drama of the news cycle, and leave you with the presidential race distilled to its purest form: a number. A number Evan and I and those like us feel compelled to check—a lot, despite understanding full well that we won’t be able to divine the country’s future in the hourly fluctuations of the prediction. Psychologists say that’s partially because sites like FiveThirtyEight are designed in a way that’s irresistible to some people. Some people like Evan. And me. And maybe you.

The Magic of Checking, and Checking, and Checking

A compulsion is a repetitive behavior “aimed at preventing or reducing anxiety or distress, or preventing some dreaded event or situation,” according to the clinical definition. In a country as rancorously divided as the US, dread of either a Donald Trump or Hillary Clinton presidency is one of the reigning emotions of this election. If the FiveThirtyEight forecast shows the race leaning toward the candidate you support—right now it’s predicting Clinton as the overwhelming favorite—then you likely feel less fear. But compulsions are also “not connected in a realistic way with what they are designed to neutralize or prevent, or are clearly excessive.” Your candidate doesn’t stand a better chance of getting elected the more frequently you check poll aggregators’ forecasts. As I check the site for the fifth time today, I wonder: Where is the line between staying informed and practicing a kind of political magical thinking?

On the Internet, it can be very hard to find. “The Internet is where people gravitate in times of tragedy,” says Elias Aboujaoude, a Stanford psychiatrist who specializes in treating patients who struggle with Internet use. “It’s where people commune. It’s where people find mutual support.”

Aboujaoude says sites like FiveThirtyEight take this a step further. “There is something to how this data is shown and the odds of winning—there’s a horse race aspect to it,” he says. When Aboujaoude says “horse race,” he doesn’t mean the media cliché for presidential race; he literally means the race track. FiveThirtyEight’s forecast is not a gaming or a betting site, but Aboujaoude says it has signifiers of both. The fluctuating odds recall gambling. And the site’s big, colorful map that constantly changes colors depending on which way a state is leaning? It’s like a board game. A board game designed to keep you playing.

The Magic of Being Right

If obsessively checking the FiveThirtyEight forecast is a kind of magical thinking, the forecast itself has its own magical quality. In 2012, when Silver was still writing FiveThirtyEight as a blog for the New York Times, he correctly predicted which way every state would vote in the presidential race. His fame was instant. The Washington Post called him “the new boyfriend of the chattering class,” adding “everyone wants a piece of him and his methodology.” Silver’s uncanny accuracy and ensuing celebrity attracted the attention of ESPN, which hired him to start an entire site devoted to divining the world through data.

Also no doubt attractive to Silver’s new bosses was the presidential forecast’s ability to draw one of the most valuable commodities in online publishing: the repeat visitor. At the peak of 2012 election fervor, one-fifth of all visitors to the Times‘ website were checking out FiveThirtyEight. This year, FiveThirtyEight says its election forecast was the most popular single piece of content across all of ESPN’s websites, and competitors from the TimesUpshot forecast to the wonky Princeton Election Consortium to the straight-up RealClear Politics polling average offer plenty of other places to click to see another number.

In 2016, FiveThirtyEight senior political writer and analyst Harry Enten has emerged as the new wunderkind (or “whiz kid,” as his colleagues call him), not because of the forecast’s accuracy (the election hasn’t happened yet) but for his skill at combining humor and obscure sports references with sharp political analysis and a clear, impassioned defense of scientific polling. Unlike those of us who rely on the forecast for an emotional fix, Enten’s job as a forecaster and a journalist is to approach the polling data with dispassion—to let the numbers as much as possible speak for themselves.

“I don’t care about people’s emotions. What we’re trying to do is provide people an understanding of where the race is,” Enten says. “I’m a big believer in ‘let the numbers lie where they lie.’”

As a keeper of the forecast, Enten checks it multiple times a day, and he gets internal notifications when it changes. But he has to get those notifications. “Don’t go bonkers,” Enten says of how often it makes sense to check the odds. But he adds that following the forecast closely can help you discern when movement in the numbers is and isn’t meaningful. “I think it takes some time to train yourself what is the signal what is the noise.”

The Bot Signal

Dave Guarino is one obsessive who took Enten up on that challenge. In July, he quietly released the 538 Forecast Bot out into the world, a bit of code he pulled together in an evening that scrapes the FiveThirtyEight website and tweets every time the forecast changes. For the poll-fixated, the bot is both balm and torture. But as the forecast has changed over the course of the election season, the bot has also become a magnet for strong feelings.

That’s Facebook co-founder Dustin Moskovitz venting his frustration at a late-September forecast that showed the presidential race as a virtual tie between Clinton and Trump. Moskovitz had recently donated $20 million to several groups backing Clinton. As the race tightened, folks like Moskovitz had no one person to take out their feelings on. So the bot became a stand-in.

“People were losing their shit at my bot,” Guarino says of the stretch where Clinton backers were engaging in what Democratic operatives call “bedwetting.” “It became the anthropomorphized version of the trend.”

Believe it or not, venting frustration at an automated Twitter feed actually makes psychological sense. “Whenever there are emotions as strong as people feel about this election, the Internet becomes a natural setting for these emotions to be released,” Aboujaoude said. The online world, he said, is “a space that feels like you can get away with releasing emotions without the kind of restrictions that you might feel offline.” Maybe like yelling at a chatbot.

The bot is lucky. It has no emotions over which a fluctuating forecast can exert excruciating power. Evan, Moskovitz, and myself are not so fortunate. In the current moment, we are all looking for just about anyone to tell us that everything’s going to be all right when this is over. Sadly, it probably won’t be. But a bot and a bit of data can make it seem like someone—or something—has at least some certainty about what the future holds. Maybe with the next click, the fate of the nation will be revealed. I mean, did you see when Arizona flipped from light pink to baby blue? That has to mean something, right? Right?

Oh jeez, even Silver can’t handle this anymore. Hold me.


I Just Want Nate Silver to Tell Me It’s All Going to Be Fine