Journalist Stephen Dubner and economist Steven Levitt are the duo behind the popular “Freakonomics” books and podcast, crunching numbers and telling stories that explore the “hidden side of everything,” as their tagline goes. They were in Seattle this week to keynote Microsoft’s Data Insights Summit, and they sat down with GeekWire afterward for a conversation about data and the state of the world.
Listen to the GeekWire podcast below, download the MP3 here, and continue reading for an edited transcript of our conversation with Dubner and Levitt.
It’s been said that we’re in a post-fact world. Can data save the world, if people pay attention to data in an unvarnished, objective way?
Stephen Dubner: No, but on the other hand, I don’t think it’s a question of saving. … The more history you read, the more you realize that this period, while colorful at the very least, is not that anomalous really in terms of political discord. I mean, even American politics. Obviously, we’re on the side of data and we’re on the side of factuality, but I will say this: We’re not activists. We’re not advocates. We really try to be non-partisan, no horse in any race, but no matter how loudly we say that we think something should work this way or it would be better to do this, nobody cares. I mean it’s all preaching to choir. I think there is a role for people like us and people who use data and so on, but it’s always going to be — I don’t want to say around the margins, I’m more hopeful of that, but political leverage is massive and that’s hard to change.
Steven Levitt: I think that data is almost never treated in an unvarnished way. Even within firms, data is manipulated. It’s brought to the front when it supports a position. It’s pushed aside when it doesn’t. Ultimately, the world is run by people. People use data for their own purposes, their own incentives. Like Dubner said, we believe in data. We live and die with data. But after, for me, 30 years of trying to convince people of things with data, what Dubner said was true. Stories are better than data even if you’re trying to be in the spirit of data.
Dubner: That said, I think an interesting question is, in what realm does it have more leverage? In other words, does it have less leverage in the political realm than, let’s say, the corporate realm? I would say yes, definitely, less in the political realm, because they don’t really need to obey it because politics really is a lot about emotion, and Q-Rating and leverage.
Levitt: But at the same time, the first Obamacare revision was killed by OMB (Office of Management and Budget). That was a rare case where really sensible data analysis actually intervened. What did they do the next time? They tried to rush it through so quickly that the numbers wouldn’t be out by the time they had passed it. Now, that was just the House, and with the Senate, who knows what will happen. But occasionally, there’s a role for data.
Levitt, you talked on stage about your data consulting work. You let the data speak for itself. You don’t take people to conclusions or interpret it. You talked about your work with King Games, maker of Candy Crush. Is there one tech company or other company out there that’s just crushing it with data — that really understands it? We’re here in the home of Amazon and Microsoft and the tech world.
Levitt: Many of the tech companies, the new companies, do an amazing job with data. I’ve never worked too much with Amazon but every indication is that they do. King Games was an amazing data-driven company. I honestly, though, have never seen an old firm, like a brick-and-mortar firm, that did much good with data, ever. I don’t know whether that’s a legacy of the way they think. It’s often a legacy of the systems they use. But I think the future belongs to firms that know how to use data. The power of data in profit maximization is just incredible. If you can get people and pride and the need for power and expertise out of the way, the data can be unbelievable.
What’s worked in the best consulting situations you’ve had when you’ve been inside companies? What about the culture of the company that succeeded made it work?
Levitt: The kinds of companies that we work well with are companies which have had some success and, more recently, have had less than success. Because they know what it’s like to do well but they also see that it’s fragile. We wrote in our book, “Think Like A Freak,” about being willing to say, “I don’t know.” That, I think, is the key thing for success with us. We don’t know anything about the businesses we work with. We go in and say, we don’t know anything about what you do. We just come in as outsiders. It’s only the firms where people are willing to say, “We don’t know the answers.” The only firms that are willing to say they don’t know the answers are the firms that are getting clobbered. But if you’re getting clobbered too badly then things are spiraling out of control and no one’s actually got any time to do real work anymore.
Are journalists doing enough with data?
Dubner: It’s certainly gotten better. I used to work at the New York Times as my day job. I was always somewhat more numerate than average just because I’d always liked math and I like economics even though those weren’t my concentrations. It was really surprising to me how much the media was really innumerate. When you don’t know something about something, as we all know, your typical response is often dismissiveness or fear/intimidation. It wasn’t like the belief that this is important to tell a certain kind of story, therefore let me try to figure it out or find advice. It’s more like, “Let me go the other direction.” The other direction is journalism by anecdote. I don’t like that so much. I like a hybrid. I like storytelling with data in it.
That said, in the last 10 years, you look at the Upshot at the New York Times, you look at FiveThirtyEight, I think there has been a huge improvement. That said, I feel like they basically built a better silo. The people who produce that and consume it, is still relatively small. CNBC financial reporting has some of the most inside-out backwards data proclamations that you can ever hope to see. Granted you’re dealing with the stock market primarily, which is one big weirdly misconstrued black box of people who don’t really know what the outcomes are, pretending that they do. I think that there are people fighting the good fight. I’m not saying we’re some kind of heroes. We just do it because we like it. But I think it’s still a little on the margins.
I’ll say one more thing about what Levitt said about in response to your question about firms or institutions changing. I think a big part of it is that, like you see in hospitals, for instance, or big insurance companies, the people who have the leverage to make decisions tend to be older or more entrenched. They’ve made it in their career and it’s scary to say, “You know what, I’m going to take everything I thought I knew about how to run this business and figure everything and embrace a different data approach and learn that.” I think that’s why you see the younger firms, or whatever you want to call them, digital natives or data natives, they have a totally different approach to it. And I get it. It’s incentives. I see why people are protectionist without it but I don’t like it.
Levitt: While it’s great to talk about data as the driving force in business or politics, the scarcest resource we have is people who can sensibly analyze data and then communicate that. I think there’s almost no way to learn how to do that. You certainly can’t learn it in school. We don’t teach it at the University of Chicago. I don’t think anywhere they really teach that. In the absence of talent, there are really limits. I draw a very stark contrast to computer programming. When there was a need for computer programming that started to arise in the ’70s, and has grown ever since, we figured out very effective ways of training computer programmers. Take somebody who is reasonably intelligent, and within four to six months, they’re a semi-decent programmer. And within a few years, they’re an excellent programmer, usually. But we haven’t figured out, the market has not yet figured out, how to train people to do data analysis. In some ways, it’s a much more amorphous task than programming. It’s more difficult to teach and there’s such a small set of people who have the experience and the training and the talent to do it.
Dubner: You should talk about your dream, your academy.
Levitt: I had a dream of doing a data science academy, but the reality was there was so much work. In the end, it just didn’t make sense.
There’s clearly a demand for it. This conference sells out.
Levitt: The demand is clear. If you look at the projections of the job of the data scientist, and the salaries of data scientists, it’s obvious that is the future. So what I say to my talented undergrads is, “Forget about getting an economics PhD. Forget about going to law school or med school.” The best jobs in terms of fun, interesting, always different, challenging, the sky is the limit, right now, it’s a data scientist.
We share something in common in that we’ve each spent a lot of time with Nathan Myhrvold. (His stratoshield idea was featured in SuperFreakonomics) Is it more needed than ever now, given (the U.S. withdrawal from) the Paris Climate Accord?
Dubner: I don’t know if it’s needed more than ever because of that. Honestly, I haven’t been following closely the temperature data. There was a period where Nathan’s group I don’t think made a lot of headway, but David Keith’s group and some others were making headway in trying some small-scale trials of geoengineering. Look, I do hope that, whether it’s that idea specifically or, I don’t know if you remember in SuperFreak, we also wrote about their idea for hurricane mitigation … which they have no idea if it would really, really work but that is the kind of idea — forget about data, per se — but it’s the kind of idea that really does change the world, that it would be nice to experiment with more.
You know, the other thing I learn about, the more history I read, is that almost every trailblazing, truly groundbreaking new idea — history of medicine, history of science, history of finance, you name it, agriculture — almost every one out of the gate is immediately ridiculed and treated as total garbage until tens or hundreds or thousands of years later, people appreciate it, because change is really scary. So, I take that to heart and know that change can often take a really long time. That said, I think the trends for this kind of thinking — data-informed thinking — the trend is definitely up and I find that really encouraging.
Tell me what work so well about your partnership. Because obviously it’s successful. You’ve been doing it for many years.
Dubner: I haven’t thought about that in the long time. It’s been so long since we’ve started together. People asked that in the beginning. I remember you described it as, we appreciated the complementaries. I appreciate that Levitt is world-class at what he does. And also what he does is rare. If you’re working with someone who does something that a lot of other people don’t do and he’s really, really good at it, I’m not going to try to do that. I’m going to try to do my thing. Levitt praises me as having higher-caliber qualities than I truly have. He thinks I’m better.
Levitt: I was going to give the exact opposite answer. Which is that, I think we’re both better at doing what the other one does than outsiders would expect. When we’re talking about research and how to interpret research, Dubner often has amazing ideas. In the storytelling, sometimes I can come up with the end of the story. What I would say is actually it’s a lack of ego. … On two separate occasions, you wrote entire chapters that took you months to write. And I said, “This is awful.” And you said, “Yeah you’re right. This is awful.” And literally just threw it in the garbage and started over from scratch. It’s almost unthinkable that anybody would be willing to do that.
Follow Freakonomics on the web and Twitter, and watch the keynote by Dubner and Levitt at the Microsoft Data Insights Summit below.