On the Red Couch with FiveThirtyEight Managing Editor David Firestone
Gabriella Schwarz / September 22, 2016
FiveThrityEight is Nate Silver‘s addictive presidential polling site, refreshed constantly on phones and desktops—a smart reminder of who’s up and who’s down in this tumultuous and unpredictable election cycle. But it’s more. The data journalism pioneers have extended their reach to economics, culture and science in an attempt to make sense of the world and the large quantities of data now available.
The team of journalists at the site, founded in 2008 and now owned by ESPN, take a numbers-driven look at the news and our lives. In addition to predicting the outcomes of elections and sports games, they’ve looked at gun violence across America, transportation patterns in a post-Uber world, and even how many curse words are uttered in Quentin Tarantino movies.
“We don’t do hot takes,” Managing Editor David Firestone told us. “We’re not interested in quick opinions, which is what so much of internet journalism is these days. And we think a lot of readers are getting tired of all the hot air.”
We spoke to Firestone, a 21-year veteran of The New York Times, about what this data shift means for journalism and technology and the site’s plans after its biggest driver, the polling predictor, comes to an end this November.
How much of a boom has this election cycle been for FiveThirtyEight?
Enormous. Our traffic has at first doubled and then tripled. We’re just kind of amazed every day at the number of people who are coming to us for forecast numbers, for analysis, for features. It’s kind of stunning. I think people who are really absorbed in this election, and there are quite a few of them out there, as you guys know, can’t get enough. And they don’t want to just read it from a single source. The day where you could somehow just read The New York Times or see it on CNN is over. Now there are so many specialized sites, whether they’re partisan or have other themes or gimmicks. They are all doing extremely well because the curiosity about this election is unending.
So, who will win in November?
At the moment [September 14] we’re predicting that Hillary has about a 60 percent, 62 percent, 63 percent chance of winning. But that gap has been narrowing for the last week or two. It started before all of the health crisis stuff came up, and I think there are some legitimate concerns for Democrats out there, who just two or three weeks ago felt pretty comfortable. We’ve noticed that there was a period where she was so far out ahead that people weren’t coming back as often because they felt it was an incredibly comfortable lead and it was essentially over. It was never really over; some of those polls were a little misleading. And now I think the true level of competitiveness in this race has started to become clear. They’re separated nationally by really only two or three points. There’s going to be a lot of dips and dives over the next few weeks. This is a very unsettled election.
You mentioned some issues with the polling. I feel like we’re in polling and numbers overload this cycle. Is it ever too much? Or do you feel a responsibility to make sense of them continuously?
We absolutely do. We don’t let the numbers speak for themselves because they’re very hard to interpret. So now almost every day we have what we call an Election Update, where we explain a little bit about the new polls that are driving these figures, which polls are doing things differently. Some of course are looking at likely voters, some are looking at registered voters, some have different estimates of turnout, some include different methodologies of how they poll, some use online systems, others use cell phones.
We’re one of the few sites that has really gone out there and tried to rate the pollsters so that we can try and figure out which ones do the best job and give those greater weight in our forecast. Most of the other sites that we’ve seen don’t really attempt that and so you basically get a lot of unaggregated data thrown into a formula. We’ve tried to distinguish ourselves by really looking closely at how pollsters do their job. And one of the things that we write about a lot on the site is how they do their job and how cell phones have changed the nature of polling. How online firms like SurveyMonkey and Morning Consult, for example, have changed the nature of polling. I think this is going to be a different landscape over the next four or five election cycles up until the point you may not recognize the polling situation as we know it now.
How do you think it will change?
I think that a lot of online polling, for example, which is not completely accepted by orthodox pollsters at this point…will be far more common and will become more refined and will allow pollsters to survey a much larger group. There’s a real concern that traditional pollsters—who use land lines—aren’t reaching young voters, many of whom don’t have landlines anymore. The nature of how you poll people with cell phones is still a little tricky and not everybody does it the same way. There aren’t really particularly clear standards for how it’s done. This is true I think of rating systems and this is one of our big themes. Rating systems and polling systems throughout all industries—whether you’re talking about music ratings, movie ratings, Nielsen TV ratings—all are changing and don’t really have the validity that they used to. A lot of companies have not really kept up with it, and they’re often surprised when it turns out they have no idea who their audience is. That’s a huge theme of ours on a very wide variety of subjects, and political polling is only one of them, though right now it’s the most prominent.
Given the pace of all of this, have you noticed a change in the reading habits of your readers?
Obviously, like a lot of sites, we’ve noticed a huge migration to reading on cell phones and tablets. And that, to some degree, changes a little bit of what we do. So much of our stuff, for example, involves complex interactive graphics. Initially we designed for the system we use here at work, which is the desktop. Then what you find is that’s not the most popular method accessing the site. So we had to make sure that all of our graphics work on a phone and that’s often difficult to do. We’ve put a lot of effort into that. We have discussed whether or not longform stories continue to work well on phone. Our experience so far has been that they do, that that has not changed reading habits. If they like a story they’ll continue to scroll all of the way down, whether they’re reading on something small or something large. But the one thing that is clear is that the phone allows you to come back to a site. So it’s a very good thing for our forecast right now. People can bookmark it on their phones and when they get a free minute they just check and see: is Hillary still ahead? Yes. And they put it back in their pocket. A lot of our traffic is based on that.
What you have to do is if you come for the forecast, please stay for the rest of the political content. And if you’re reading that you might also be interested in some of our economics content, which is often tied into politics and maybe our science and other topics that are still a huge part of what we do here. We want to make sure people are seeing that [content] before a period when we know the traffic will go down after Election Day.
So how do you keep up the momentum then, post-election?
I think one of the things is to show people that this is a concept of journalism that works, that doesn’t have an expiration date. We are not by any means just about predicting elections or predicting the outcomes of football or baseball games even, though we do that a lot. People appreciate it; it helps them understand politics or sports or the latest numbers on middle class incomes. But nonetheless, that’s only a part of what we see as a good data journalism diet. We want to show people that by looking at numbers and by taking an empirical approach to journalism that we can do stories in a way that will stay compelling long after Election Day, long after the World Series is over, for example.
We don’t do hot takes. We’re not interested in quick opinions, which is what so much of internet journalism is these days. And we think a lot of readers are getting tired of all the hot air. What we want is to get behind all that and explain a little bit about what the numbers are telling us about our world. It doesn’t mean you don’t interview people. You still go out there. We have reporters on the campaign trail, we’ve sent them to states around the country for politics. We did a gigantic project on gun violence in America, where we sent reporters all over the country to write about the causes of gun violence. In many ways, I think, the result surprised people who didn’t know the nature of gun violence. The way it’s reported in this country, you’d think it’s all done with automatic weapons, and it all takes place in giant massacres. That’s not what gun violence is really about in this country. We showed the reality behind it using both old fashioned interviewing techniques and data. I think people really got a great deal out of that, and we’re going to keep doing it.
Do you think data journalism is a reaction to technology, or is journalism helping shape technology?
Well, I think the technology that’s out there has definitely helped us make this kind of site possible. We’re using tools that are still very new and that not many people really understand—I add myself to that. I’m not an expert on this; I’m a traditional journalist who until fairly recently did not really work in this space. But I am unendingly impressed with the talents of the young journalists here who grew up with this as their journalistic vernacular. They understand it and they have used and developed these tools to make things possible. So there are database programs out there now that allow you to very quickly assemble enormous amounts of data and analyze it in ways that you could never do.
Just one example that your readers might appreciate: we got all of the pickup and dropoff information for Uber cars in New York City and then compared them to all the pickup information for Yellow Cabs, the traditional taxis in New York, to see whether Uber was displacing or changing the nature of transportation in in the city. You can’t even imagine how many millions of points of data that required. Every time a taxi stops and picks something up, it’s recorded. But no one has ever been able to analyze this data before because it’s too big. The database is gigantic and in the old days it would take weeks and weeks of time just to kind of figure out what you had. But our people got it through a Freedom of Information Request from the city, put it into a high-powered computer and within a day or so they were mapping every taxi pickup in New York. It was an amazing thing to watch, and I think the stories that resulted from that (and there were several) shed completely new light on how a new form—a new startup, really—of transportation is changing the nature of life in the city…That’s what we’re about. I think this is going to be a gateway to understanding our world in very very new and exciting ways.
Do you think those data skills are necessary for journalists now and is that a good shift for journalism?
Yes. They’re not 100 percent necessary; we still need good feature writers, we need great photographers etc., but there are just fewer outlets for people who can do that. So many small and medium-sized newspapers are shedding staff on a regular basis…It’s really tough to get hired by those places and find a career in journalism, but if you know how to do this, if you can figure this stuff out, you have a much better shot of getting hired somewhere because not enough people do it well and everybody’s hungry for it.
What areas are prime for the next FiveThirtyEight attention/what area most needs that type of coverage?
We are building up three sections that we have done work on in the past, and they’re going to become a much bigger part of our site. Those areas are economics, science and culture. It might seem obvious, for example, how we would do this for economics and science. Culture might seem like a little bit of a stranger fit, but in fact one of the things that we found is you can do amazing stories about issues like religion, like movies, music. We even analyzed the number of words in Broadway musicals to show how many more words Hamilton packs in than virtually any other musical in history, which is a light but really delicious piece…People love this stuff.
We uncovered a scandal in the crossword puzzle industry by using data to show that USA Today and one of the big syndicators were essentially just plagiarizing other puzzles. The New York Times puzzle was plagiarized by these guys for years. Well, how do you prove that? You can only prove it by analyzing through a computer system thousands and thousands of puzzles. It seems crazy to even think of doing this, but we did and it was a great story that got a huge amount of attention.
We know our traffic is not going to be quite the same after November 8, but we have a lot of confidence that these new areas and the new interest among readers, both young and not so young, are going to keep us going. It’s a really exciting set of possibilities.
Follow FiveThirtyEight’s unique reporting from sports to TV by following them on Flipboard.