Analysis

Nate Silver unloads on
The New York Times

May 11, 2016
Logo: FiveThirtyEight; Treatment by CJR

FiveThirtyEight‘s Nate Silver ripped into The New York Times in general–and the paper’s new media columnist, Jim Rutenberg, in particular–on the FiveThirtyEight election podcast on Monday. The minutes-long rant included loaded words like “dishonest” and “unethical.”

Silver and his operation had an alliance with the Times during the time of the 2012 election. The attack showed that there’s little love lost over the split, at least from Silver’s perspective. It also proved that Donald Trump, the apparent victor in a race that reporters across the spectrum called spectacularly wrong up until the very end, is still roiling the media world.

The catalyst for Silver’s unleashing was a column from Rutenberg, who stepped into the vacant David Carr job at the beginning of the year. The piece ruminated on the myriad errors made by the media over the course of the utter mayhem that has been the 2016 presidential race. The column wasn’t entirely focused on Silver; it mentioned failures in Times prognostications as well. But Rutenberg did seem to go out of its way to bring up FiveThirtyEight, especially in noting a bad call for the Indiana Democrat primary, in which FiveThirtyEight had favored Hillary Clinton to win but Bernie Sanders ended up taking in a romp.

There was subtext there, too. Several times in the piece, Rutenberg advocated for “shoe-leather reporting”—talking to “actual humans,” as he put it—and concluded:

That’s all the more reason in the coming months to be as sharply focused on the data we don’t have as we are on the data we do have (and maybe watching out for making any big predictions about the fall based on the polling of today). But a good place to start would be to get a good night’s sleep, and then talk to some voters.

To those with their ears attuned to fissures in the media world related to data journalism, the use of the word “data” was pointed. That, plainly, was what Silver responded to. The site’s election podcasts generally feature Silver and several other of the site’s election team discussing the race, with particular attention paid to polls. At about the halfway point (33:15 in, to be exact) in the conversation posted Monday, Silver unloaded:

Sign up for CJR's daily email

Jim Rutenberg and I were colleagues in 2012 when FiveThirtyEight was part of The New York Times. They were incredibly hostile and incredibly unhelpful to FiveThirtyEight, particularly when FiveThirtyEight tried to do things that blended reporting with kinda more classic techniques of data journalism. When we went to New Hampshire, for example, to go to The New York Times filing center … the political desk is literally giving us the cold shoulder like it’s some high school lunchroom.

“This happened, right? When we filed the story pointing out before anyone else at the time … that Rick Santorum had probably won the Iowa caucus and that was a story that involved a combination of data work and reporting … they were apoplectic, because their Romney sources were upset and their Iowa GOP sources were upset, so a story that, no. 1, was a perfect blend of reporting, which is what Rutenberg claims we need, with data, and, no. 2, got things totally right, pissed them off because they were mad they didn’t get the scoop and it went against what  their sources wanted. I mean, this guy was extremely unhelpful.

Rutenberg, in a telephone conversation, said he had been trying to interview Silver since February and had been talking to publicity people at ESPN, now FiveThirtyEight’s parent company.

“He declined an interview request the day before,” Rutenberg said. “Those interview requests stand. So I’ll talk about this a little bit, but to now litigate the story I want to write through other press just feels weird. He chose to talk after the fact, and that’s his right; he’s obviously upset. I tried to have a constructive conversation with him about data journalism and to be educated and he chose not to. The whole thing is a little baffling to me.”

I asked him about the intra-office coldness Silver described.

“I’d say there was definitely growing pains, yeah,” he replied. “Some of what he said I’d take issue with, but generally, sure there was. That was a very new thing. There was a group that, generationally, younger reporters had come up under older reporters. There was a way of doing journalism. And here’s this new thing.

“Since 2012 we’ve all learned a lot. My thinking about data journalism has evolved. There are things I want him to explain to me I still don’t get. For instance, the value of saying someone has a two percent chance, way out before an election.”

I laughed. Rutenberg’s reference was to an assertion from Silver very early on in the race that Trump’s chances of getting the nomination were minuscule.

“No, no,” he protested. “I’m not saying that disrespectfully. I’m really not. I’m sure there is a value, maybe, I don’t know what he would say, because he wouldn’t talk to me, but maybe it’s, ‘You have to start building your model.’ I want to have that conversation.”

In the podcast, Silver is plainly exercised. He stops to take a breath at one point, to the laughter of his colleagues. Then he ratchets up the stakes:

And this is someone who, by the way, doesn’t talk about that we were colleagues together at The New York Times, a person who cherrypicks the facts he’s looking at. So he mentions that in our Indiana prediction for the Democrat election, the underdog won [i.e., Sanders over Clinton], but in fact the favorite has won 51 of 56 times in our polls-only forecast. To me, that’s dishonest and unethical, frankly. And he doesn’t really take the time to truly understand what’s going on.

Asked to respond, Rutenberg said, “What do you say to that? I wish him the best. I’m not going to get into name-calling. I approached him with an open hand and open mind. I’m not going to get into name-calling. I don’t think he’s unethical.”

It’s probably fair to say that the Timesman’s contrast of FiveThirtyEight methodology and traditional reporting was somewhat cartoonish. The site publishes reams of analysis of polling data in a fairly nuanced way. For example, for races it publishes two columns of data, for the “polls-only forecast” and the “polls-plus forecast.” The latter includes factors like endorsements.

And as for “shoe-leather” reporting, Rutenberg’s take on “actual humans” was highly romanticized. You can spend all day talking to folks in Austin, Texas, and still be surprised when the Texas voting results come out. A conscientious operation will do more than that, of course; it might send reporters out to different places, and factor for things like race and whether the real people they are talking to are registered voters or even likely voters.

But of course, that’s basically data journalism, and a poll is a highly evolved form of just that.

Rutenberg: “It’s all good, it should all work together, it’s not an either/or. Some people try to force this column into a box, to say that it said that’s an either/or choice when it explicitly didn’t.”

Bill Wyman is the former arts editor of NPR and Salon.com. Follow him @hitsville.