Look. I get it. And if you’re being frank, you probably do, too. We’ve all had one big success long ago that we’re proud of, that we can point to, and that we desperately latch onto to prove to the world outside we are not frauds and failures. For election data guru Nate Silver, it’s the 2008 election.
He’d invented a model to project the race—which weighted the relative accuracy of polls—that ran somewhat counter to the prevailing narrative. While pundits saw a tight race between Sen. Barack Obama (D-Ill.) and Sen. John McCain (R-Ariz.), Silver said Obama would cruise to an easy victory.
He was right, and the people who wanted to see it as a tremendous win for data and statistics (data nerds, wonks with rectangular glasses, the people ruining baseball) saw it as a tremendous win for data and statistics. Silver parlayed his blog’s success, known as 538, into a plum job at the New York Times, then ESPN, then ABC.
But in the Trump era, his model’s been, how do you say… shit? His primary analysis was completely off-base. He was raked through the coals online for his models showing Hillary Clinton would win, and doubled down when President Donald Trump pulled it off, complaining that it was everyone else’s fault they didn’t properly understand percentages and his models.
Which brings us to last night. Remember last night? It was just last night. Silver’s 538 model heavily tilted for Joe Biden, even more so than Clinton in 2016. In his final projection, simulations said Biden would win 90 times out of 100. Many of his potential scenarios showed Biden cruising easily, with Trump having to walk a narrow tightrope to electoral victory.
And because we are forced to relive 2016 forever and ever and ever, the results started rolling in, and a Biden landslide was taken off the table.
People had one thought. Why the hell are we still listening to Silver?
Silver, for his part, fought the good fight by… getting mad at people who got mad at him.
The thing is, Silver will have a defense for himself when this is all over. As things stand right now, it seems Biden will win. But the Electoral map is starting to look like it would have had there been a polling error similar to 2016 (which still favors Biden). Which Silver talked about a great deal in the run-up to the race. So bully on him.
It’s just that when you made your name with a super-sophisticated model that was specifically designed to discard bad polls, and the model of the polls keeps being bad, maybe you aren’t the data guru you claim to be.