Almost as soon as the anticipated “red wave” of Republican wins failed to materialise on US election night earlier this month, the online finger-pointing at the pundits who got it wrong began. But for certain members of a small corner of Twitter, the Democrats’ unexpected success was a vindication of their predictive skills.
“The simple fact of the matter is that an anthropology professor in Arkansas and a group of 20-year-olds have managed to outclass some of the most qualified TV and media figures of the 21st century in this election,” tweeted Lakshya Jain, three days after the vote.
Jain, the 25-year-old co-founder of election analysis blog Split Ticket, was referring to an eclectic group of amateur online election forecasters, commonly known as “#ElectionTwitter”.
In the past two decades, a number of media brands, including Nate Silver’s FiveThirtyEight and Charlie Cook’s Cook Political Report, have risen to prominence as perceived bastions of reliable election prediction, using a mix of polling data, historical trends and demographic patterns to accurately forecast results.
However after underestimating Republican support in the 2016 and 2020 elections, polling firms and established election pundits faced criticism, and a space was vacated for a clutch of amateur political forecasters, many of whom have amassed large followings online in recent years.
According to Jain, Election Twitter has gained even more traction during this midterm cycle because of how accurately some of its highest-profile figures predicted the results — something he attributes to their being untroubled by past forecasting missteps and thus more able to trust the data in front of them.
“It’s a community of people who studied a lot of elections historically, but haven’t really lived through a lot of them,” he said. “[In] this cycle, a lot of whether or not you got your predictions right . . . was based on how much you let historical priors affect your view of the environment.”
While some of the participants in Election Twitter are academics and professionals, most are hobbyists, students or politics superfans. Many of its most active members, who are typically white men, use pseudonyms with account names such as “Dukakis Dude” — a reference to 1988 Democratic presidential nominee Michael Dukakis.
Buried under the memes and in-jokes is a community with a deep, sometimes obsessive, appreciation of elections and polling data.
One popular pseudonymous user, who goes by @umichvoter, correctly predicted every Senate race this year and missed the final margin in the House by only a handful of seats. A medical student who spends most of his free time creating maps and spreadsheets from election data, he has amassed more than 40,000 Twitter followers.
“I like looking at data. I like following the news. It was something that kind of kept me going these past few months,” he said. “I don’t have a career in this. So it’s just purely off interest.”
Meanwhile Ethan Chen, a 21-year-old elections enthusiast with several thousand followers on Twitter, predicted margins in the House within one seat ahead of the election. He stressed, however, that his analysis was a product of numbers from polling and forecasters, not a rebuke of them.
“This year, polling was actually fairly accurate, but . . . [pundits] don’t want to be accused of being biased towards Democrats again, [so] they had a tendency to interpret them a bit less charitably,” he said.
The established prediction outlets did not actually get the midterm results as wrong as many complained in the aftermath of election night. FiveThirtyEight and Cook Political did both overestimate Republican gains in Congress,…
Read More: How #electiontwitter took over the political prediction game