NEW YORK - Nate Silver is on the downtown 1 train. Possibly because he looks like a (modestly) hip math teacher, and hardly looks up from his phone, he goes unrecognized until he reaches the PlayStation Theater in Times Square.
There, his name is in lights, and people start to nudge one another and point him out. Hundreds of fans - many of them male, young and white - have lined up outside, waiting to watch the data journalist and his colleagues record a podcast. Those who hold the priciest tickets ($100) even get the chance to mingle with the stars of the website FiveThirtyEight and have their picture taken with top editor Silver afterwards.
"We're giant nerds," explained Priyanka Mitra-Hahn, a PhD student from Brooklyn, when asked why she, her wife and a friend came out to the sold-out event.
If a statistics guru can be a rock star, Silver is surely it. But even rock stars have bad days.
Silver, 38, had a run of them a few months ago, when it became obvious that his consistent early dismissals of Donald Trump's chances to be the Republican presidential nominee were flat-out wrong.
And this University of Chicago graduate wasn't used to being wrong - not in sports, where he made his name using statistics to compare individual players' performances over time. (FiveThirtyEight is now owned by ESPN.) And not in politics, where he famously called two presidential elections nearly to perfection: He got 49 states right in 2008 and all 50 in 2012.
But when it came to Trump, Silver and Co. missed the boat with a resounding splash. As the Republican primary elections approached, Silver pooh-poohed Trump - by mid-February, he was still putting the businessman's chances for the nomination at less than 50 percent.
Many took comfort, trusting that Silver could not be wrong.
"For those of us who didn't want to believe we lived in a country where Donald Trump could be president, Silver's steady, level-headed certainty felt just as soothing as his unwavering confidence in Barack Obama's triumph over Mitt Romney four years ago,"wrote Leon Neyfakh of Slate. After it became clear that Trump would win the nomination, Silver apologized in a long post. And he explained that "we made a big mistake."
The mistake was allowing his predictions to be based, at least partly, on the very thing that he has criticized all along: Educated guesses.
In other words, punditry.
He explained in his mea culpa the consequences of his decision not to build a statistical model.
"Without a model as a fortification, we found ourselves rambling around the countryside like all the other pundit-barbarians, randomly setting fire to things," he wrote.
In his Upper West Side newsroom (Disney and ESPN staff are housed nearby, and all the ID cards feature Mickey Mouse), Silver talked about learning that lesson the hard way - and about the next hurdle.
"There's certainly huge anxiety," he said. But now that he and his colleagues are back to relying more fully on data and statistical models, "there's not much we can do about it."
"We make the best forecast that we can," and wait for the truth to arrive to prove them right - or wrong. (As The Washington Post's late David Broder once told a colleague, "The only poll that matters is on Election Day.")
Silver is pushing hard to keep expectations reasonable: "We've learned that we have to be careful about how we convey uncertainty."
Many of those who addictively check the FiveThirtyEight site take solace, no doubt, in the topline election forecast - it's the most popular piece of content among all of ESPN's offerings on the Web.
Last week, the numbers continued to seem reassuring to the never-Trump forces.
Hillary Clinton's chances of winning the presidency were at about 85 percent; Trump's at about 15 percent. (They've tightened since then. On Sunday, Clinton was at 79 percent and Trump at 21.)
On stage that night, podcast host Jody Avirgan referred to the "foregone conclusion" that Clinton would win.
Silver, alarmed, nearly shouted his response: "No, no, no, no!"
After all, he reminded everyone, what if you had a 15 percent chance of losing at Russian roulette? You might not call your survival a foregone conclusion.
In the three years since Silver left the Times for ESPN, his FiveThirtyEight site has grown from himself and politics editor Micah Cohen to about 45 writers, editors and data-visualization and multimedia journalists. Last month, according to internal metrics, the site had 8.7 million unique visitors.
Looking back, David Firestone, also a former Timesman and now FiveThirtyEight's managing editor, echoes Silver on what happened with the Trump miss.
"We let our gut lead our head in the wrong direction," he said. "We hope we'll never let that happen again." (It must be noted that FiveThirtyEight was far from alone among errant prosnosticators; Trump was, by all measures, an upset.)
But Firestone and Silver agree that uncertainty is ever-present, and no amount of numbers, or analysis, can change that.
"We don't know who's going to show up to vote," Firestone cautioned. The site goes out of its way now to remind its users of all the X factors at play.
Silver said he's painfully aware that many people would love to see him fall on his face.
"We were sympathetic at the New York Times because we were attacked by a lot of people," he said. Silver left the Times partly because of a much-discussed culture clash with the paper's political reporting staff, which never warmed to his numbers-driven methods - but also because ESPN was able to offer bigger money and more resources.
"It's hard to go from being an underdog to an incumbent," he said. "We're not this cuddly story any more."
If he's wrong next week - nah, can't happen, right? - cuddles could be even harder to come by.