Is Nate Silver doing it all wrong?

Ryan Grim, The Huffington Post’s Washington bureau chief, lit up Twitter Saturday by criticizing the celebrated FiveThirtyEight forecaster’s predictions for the presidential race. Grim’s central objection is to Silver altering poll results for his election modeling, which results in Silver concluding that “Trump is about 3 points behind Clinton—and 3-point polling errors happen pretty often.” In Grim’s view, Clinton’s polling lead is actually much greater, and this “trend line adjustment” by Silver is “merely political punditry dressed up as sophisticated mathematical modeling.” He accused Silver of “making a mockery of the very forecasting industry that he popularized.”

That didn’t go over well.

Silver said the polls are adjusted because “that’s what works best empirically.” He added that people like Grim “don’t actually give a shit about evidence and proof,” tweeting at the reporter, “The article made clear you have **no fucking idea** what you’re talking about.”

In defending his model, Silver pointed to recent “high quality” polls that receive extra weight in the FiveThirtyEight model.

He also cited this evidence:

New York magazine’s Jonathan Chait raised the issue of early voting, which includes a surge in Latino turnout that almost certainly benefits Clinton and harms Trump.

“Your model may be doing a better job of synthesizing polls,” Chait tweeted at Silver, “but if those polls missed the Latino vote, people might be making an outcome-based judgment.”