John has a long history of traditional journalism, including time as a police reporter at the Wisconsin State Journal. But these days, he’s breaking more news for his programming than his writing. He’s also the author of Family Projects for Smart Objects: Tabletop Projects That Respond to Your World(Opens in a new window). Today we’re going to talk about data journalism, the Internet of Things, and how Keefe monitored voter behavior in real time across the country. Watch the full interview in the video or read the transcript below.

Dan Costa: Let’s start with the most basic thing. WNYC is a public radio station here in New York City. Why do they need a data reporter?

John Keefe: That is a really good question. The thing about the public radio audience is that they are information consumers, lifelong learners, always wanting to know about the latest thing, not just in politics, but also about the world around. A few years ago, we decided that one of the ways we can do that is by making really smart interactives online that are driven by data. At the same time, we’re in the newsroom. So we help our reporters do investigative stories and other stories when they get spreadsheets and databases and other things.

There’s sort of a joke in journalism that I went into journalism because I can’t do math. That’s not true.

Dan Costa: That is why I went into journalism.

John Keefe: It happens a lot, but we are sort of the math and data folks, including also code and design, plopped right inside the newsroom. So we help both with visuals on the site, with the reporters’ investigations and anything else that comes up that might be- … Require a little skunkworks that we operate.

Dan Costa: Yeah, because people think of journalism being all about the writing, all about the editing. Then publishing, and then the information’s out there. But the packaging and understanding the information is increasingly important.

John Keefe: Yeah, and it will often come up with projects that require a new way to talk to people. Whether it’s SMS text, or other things that we haven’t done before. So we’re the crew that sort of jumps in and gives those a shot.

Dan Costa: Talk to me a little bit about Electionland(Opens in a new window). What were its goals and did you accomplish them?

John Keefe: The goal of Electionland was to, in real time, that’s key, keep track of problems at the polls. So the laws around elections have changed, the voting rights, the Supreme Court changed the Voting Rights Act. Struck down parts of that, and so this was the first presidential election since that change, which led to some more voter ID laws and other limits, and other things around voting. We wanted to see, in real time, whether or not we could spot any problems at the polls.

Dan Costa: In the old days, you’d have reporters at a couple polling stations, three or four, and they would be there with a camera and they might be able to go live and show what was going on. They get five anecdotal cases across thousands of polling stations.

John Keefe: Right, fine, five would be a lot, actually, and what would happen is you wouldn’t hear about it until the end of the day. Like maybe for the evening news, you would hear about it, but we wanted to know, early on, in real time, across the country. So we had 11,000 journalists and journalism students, that’s over 1,000; 125 of them were based here in New York City. We set up this pop-up newsroom at the CUNY Graduate School of Journalism, and we monitored Twitter and Facebook, and Google Trends. And we also had access to data that included the folks when you call 1-866-OURVOTE, which is a phone number that people use to get help voting or reach lawyers. We had access to that data. Then we invited people-

Dan Costa: How did you get access to that data?

John Keefe: Well, we asked them if we could. We worked out an agreement with them to make sure to protect people’s identities, and everybody involved, but we wanted to know ‘Hey, if there’s a problem at this polling place, could we find out about it?’ Then we invited people to join a texting system, actually. That’s something that I helped to run which was before the election and even during early voting. We invited people to sign up for this texting service. And what we did is we checked in with you to see if you had voted, and how long it took you to vote in minutes, and anything you may have seen. Also, where you were.

And that way, we could add that to this pool of data, and we were able to monitor the entire country, and keep an eye out for voting problems.

Dan Costa: You set that up, a lot of these are public channels that anybody would have access to? SMS is an interesting one, because people don’t think of that as a channel that can be monitored, but that’s happening in real time. So what did you discover, having done this?

John Keefe: Well, a couple things. This was a collaboration of 400 or so journalistic organizations, including ProPublica, WNYC, the New York Times, and then a whole bunch of local papers and public radio stations and TV stations. Then it was also a combination of a lot of tech companies. We actually had help and funding from Google, but also, Dataminr, and a company called Midan, and a whole bunch of other tech folks.

Everybody came together. I think there may have been one written agreement. This was just people coming together saying ‘Yeah, uh, we’d like to monitor the vote, this is a, a good purpose.’ So we learned about a massive collaboration, what that takes. Then we really learned about how to monitor news, breaking news of a sort, in real time, in public channels and some private ones across the country. And it took a lot of work and a lot of coordination but we were able to pull it together.

And we definitely saw problems, but they were the typical problems; broken machines, missing poll books, the things you have to sign, they’re in the wrong place, so they have to get moved around. There were some reports—before the election, you may remember, there was a lot of concern and even candidate Trump calling out for people to go monitor the polls. We didn’t actually see a lot of that, and we didn’t see any intimidation of any- … Actually, we saw some, but nothing organized.

In the end, I think what we learned was the election, the voting process went really, really smoothly, with some bumps here and there, that you would expect.

Dan Costa: So, one of the other technologies, we talked a little bit about SMS. We’ve talked in the past about chatbot, and how that’s an avenue that’s just coming into its own. It can be used for a lot of this real-time feedback. We’ve seen a couple of examples so far, Facebook’s got a platform. We know we can order a pizza using a chatbot. But ultimately, that’s a platform that we use for a lot more interesting things.

John Keefe: In its early days, the New York Times has done some interesting stuff, Quartz has been doing some really good stuff with chatbots. But a lot of it is mostly information coming at you. Maybe you have some ways to navigate it within Facebook Messenger, or in an app, in a chat app. Most of it is still the information coming at you with a little bit of interaction that you get on it.

I think what’s going to be interesting is how we can actually, as journalists and makers, get more input from the audience, and also how we integrate into our normal lives. Alexa and Siri and Google Home ($99.00 at Target)(Opens in a new window) , all of these intelligent agents are going to be a part of our lives more and more. If that’s the case, then how do we put information and media into those channels that make sense beyond saying ‘Tell me more, go to the next story.’ Which is useful, but …

Dan Costa: Very basic and linear.

John Keefe: Yeah, very.

Dan Costa: And you look at the companies that are driving the sort of chatbot development. It seems to be how can we replace humans, call centers, waiting times with chatbots to just answer people’s questions.

John Keefe: Right.

Dan Costa: And that’s useful and it certainly adds efficiency, but I’m waiting for that wave of chatbots where people are building their own and it’s serving their purposes and they’re really in control. It seems to me we’re a little ways away from that.

John Keefe: Yeah, maybe a little ways away, but not that far, I think. To the extent news organizations also provide useful information, even things about like where to vote and other civic information. Maybe an organization, especially local newsrooms have also been a source for that kind of information. Our team started out sort of helping people navigate the hurricane evacuation map for Hurricane Irene. Is that something you can look at and just view online? Sure. Is there some interactive way? Like if that were something where you say, you know, ‘Intelligent Agent, what’s my evacuation zone?’ That might be something that a news organization provides at some point.

I think there is a lot to come when it comes to chats and bots.

Dan Costa: One of the things that I think is most interesting about your work is that you say data journalist, you think about working with spreadsheets and you certainly do enough of that. You’re also out in the physical world, measuring things and accounting for things. So it’s not an abstract thing you do at your desk. You’ve got robots around the city of New York informing, giving you information about what’s going on. Give us some examples of that.

John Keefe: I’ve been really interested for the last couple of years, just personally, with hardware hacking and computers. Not hacking in the ‘hacker’ sense. Basically playing, tinkering, mainly, is what I mean. And sensing our environment and playing with little Arduinos and particles and things that would report back. I’ve just been having fun playing with those.

Then, a story came up earlier this year where we wanted to work…an organization approached us and said that they wanted to measure how hot apartments in Harlem got during the summer. We live on a heat island in Manhattan, and Harlem is a hot area. It’s literally hot, it also has a socioeconomic group that is sometimes at risk. So there are parts of the city where you might have lower-income folks or people who can’t afford air conditioners. So what is their life like, temperature-wise?

So we came up with these little…they’re little bots, little Adafruit-based Arduino things that would take-

Dan Costa: That you built yourself?

John Keefe: Yeah.

Dan Costa: And gave to them, and they took them home?

John Keefe: Yes, we made 50 of them, mostly I made them.

Dan Costa: Go get some interns.

John Keefe: Exactly. Lots of soldering. We worked with a community group called React, and they helped, and they knew people in Harlem. They worked with people to get these devices into their apartments, and the devices did two things. It took the humidity and the temperature every 15 minutes, and then recorded it to an SD card. Then, the volunteers would go around and get those SD cards and send the data back to me. Then I would provide that to researchers who were working with us.

In the end, we were able to track the temperature in Harlem apartments, and a lot of Harlem apartments without air conditioning, all summer. And we found out that they are hot and they don’t get cool at night. Even when you could get relief from outdoors because of the way the apartments are, the way the buildings hold heat, for so many reasons. These apartments stayed really hot at night, like above 80, sometimes 85 at night. And that we did many stories about.

Dan Costa: The takeaway being the human body is not meant to live at those temperatures for those periods of time.

John Keefe: Absolutely, and in fact, there are advisories about not spending that much time in that. I mean, 85 degrees, if you’re working at an office that’s 85 degrees, you’re sweating it.

Dan Costa: It’s about 85 degrees on the set right now.

John Keefe: So you get the feeling.

Dan Costa: I get it.

John Keefe: But it’s then, all summer long, living, sleeping in those conditions, we talked to people who wouldn’t leave their homes to go to cooling centers because the hallways were even hotter. And so we did a lot of work, and this was, again, based on these devices that were just running off of batteries and recording the internal heat, interior heat in these apartments.

Dan Costa: The thing that makes that story so interesting is you created 50 of them, but we’re going to, in five years, that kind of sensor technology is going to cost less than a buck, and it’s going to be built into phones, it’s going to be built into people’s homes and it’s going to be widespread. And there’s going to be this type of data collection, at least, theoretically possible at much larger scale.

John Keefe: Yeah.

Dan Costa: I mean, that’s the thing that I think is really interesting, when it’s not just Harlem, but you’re doing an analysis of all of Manhattan, of all of New York State. It really starts to blow up. I think that’s very interesting.

John Keefe: There are, of course, devices that do this right now. Right? What we wanted to do is a couple things. One, make them really quickly, exactly the way we wanted them. So we used off-the-shelf parts from Adafruit to put them all together and solder them together. But, then also make it available so other people could do it, right? This worked out to be about 50 bucks, but you’re right, the price for all of this hardware is just coming down. And if we can build them, use them, and then post on GitHub or wherever, how we made them, then other people could do the same project.

We’re not in the business of cornering the market, things that will measure your temperature.

Dan Costa: Right, right.

John Keefe: And recording. We were more interested in trying it out, giving it a shot, and then letting people know how we did and what worked and what didn’t.

Dan Costa: So, we talk a little bit about this data. You’re obviously trying to use this data for good, to inform people, to understand our world better. I was in Portugal a few months ago and I was giving talks about the smart home and automation and sensors and how your house is going to know when you came home from work every day, and wasn’t that going to be great? And every time I gave the talk, the first question was ‘That’s horrible, that’s scary, uh, we should all, what about privacy? Does nobody care about it anymore?’

Seems like people in US care a little bit less than people in Europe about privacy, but how do you read it? You work with this data all the time, how worried should the average consumer be that their entire lives are being tracked now?

John Keefe: Yeah, I think that’s a great, great question. When it comes to something as simple as a heat sensor that’s using Arduino, that’s just registering the temperature. It’s hard to see how you could be worried about privacy there. But it could be, there could be an issue. I think this connected data where, yeah, we’re being tracked. This device right here is tracking us a lot.

Personally, I think it’s something people should be thinking about, and thinking about what data is being collected, if you can figure it out, and who is going to hold on to it. Even if somebody is holding on to it, who else could get a hold of it? I think about that a lot, when we’re doing projects, right? We are an organization who respects our audience and if we ask people for information, we try to hold on to that, and in a secure way.

You never know when that might be compromised or if somebody else could end up with that information. So we’re a little bit careful about the kinds of things that we ask about and track as an organization. We think about that a lot. I think people who are installing that kind of equipment in their homes, personally, I think it’s worth the thought.

Dan Costa: Are there any steps that you take personally in terms of maintaining privacy and protecting your data? Do you use an encrypted browser?

John Keefe: I make sure to use FileVault on my laptops, so that if I happen to leave my laptop in a coffee shop, it’s basically a brick to anybody who looks at it. So I try to do that. I’m a big fan of Signal, which is an encrypted end-to-end chat. As as journalist, there are situations where I might need to talk to people who don’t want to have it revealed that they’re talking to me and I don’t want to have to accidentally reveal that.

So that’s something I use. I do have PGP key so people can reach me that way, online. I will just say, very few do.

Dan Costa: Nobody ever tries to reach me by PGP. I created a key, and I haven’t used it, probably in 10 years.

John Keefe: Right, and PGP is hard, it’s complicated. Things like Signal are a little bit easier. It’s basically an app that works like a chat and you can communicate with each other. Yeah, taking some steps, I think it’s a challenge to secure everything that you do, and I think we do our best.

Dan Costa: So one of the other things, we want to talk a little bit about Internet of Things. I want to talk about your book a little bit. Most people getting into the smart home, making their home smart involves going to Best Buy, picking up a Lyric ($279.99 at Amazon)(Opens in a new window) , picking up a Nest , setting it up in-home, and [connecting to] a smart home hub. Your approach is a little bit more hands on.

We’ll show the book here. It’s Family Projects for Smart Objects. I like the way that sounds. Tabletop projects that respond to your world. I think that’s the interesting thing, here. Knowing your work is that it’s not just stuff you build in the abstract. There are objects that actually work with the environment around you. Talk to me a little about the book and some of your favorite projects.

John Keefe: It’s funny that you say that. Our smart home has got little Arduino things all over that my daughters know how to push the buttons on and stuff, and that’s kind of how we start our day, on occasion. The book is really about making. It’s a way into this world of Arduinos and sensors. I try to make it as friendly and easy to use as possible, really walking you through. You do not have to know anything about how to do this to build a lot of projects.

An example of one that I like is just using tin foil as a capacitance sensor. So the technology that, basically, you’re using and the idea that you’re using when you touch your pad or the phone is sensing the capacitance in your body, and you can do a quick little demonstration, where you can touch some tin foil. It’s not a button, it’s just you’re touching it. You get really close to it, and you light up an LED. Another example is the ‘somebody moved my stuff’ alarm, which is that you create a little pressure sensor and you tie it to your Arduino, and then you can set a toy or somebody on Twitter put a beer on it. Like, if somebody takes it, it sets the alarm off.

Dan Costa: Which is a huge business. There are a bunch of products using Bluetooth to set proximity alerts.

John Keefe: Yeah.

Dan Costa: You can build those at home.

John Keefe: So the things that you could build here, you could certainly use, and we do use some of these things in our house. Mainly the point is to do it, to experiment with it, to play with it. Because the prices are coming down on all of this hardware. Some of these sensors just cost pennies. It’s not hard to do. Not hard to try, and I designed it to do with kids. So it’s not just a kids thing, it’s more like kid/parent experience, and that’s really the goal. To just really understand that the technology isn’t as mysterious as it seems.

Sure, we’re not going to be building iPhones or fully connected smart homes, but if you can make some parts of it, you understand it a little bit better.

And it’s fun, and that’s the big part.

Dan Costa: I think that’s one of the key takeaways, is that people feel this anxiety that things are getting out of control, they don’t understand how the world works, they don’t understand how the Internet works and how all their smart home appliances work. There are smaller pieces of it that you can break away and to give you a feeling of control and to help you understand how the digital world interfaces with the physical world. It can be child’s play.

John Keefe: I actually do teach this at the CUNY Graduate School of Journalism. Sometimes this notion of making things. One of the things I like to do is take apart an old iPhone and just show you all the pieces in there. There are analogs so that you can buy a little tiny camera, you can buy the little tiny microphones. There are other components that are in there. Sure, you’re not going to build the iPhone, but yeah, you’re right, the sensors that are involved, on some level, you can play with pretty easily, and that’s what’s fun.

Dan Costa: So you are obviously, pretty technologically sophisticated, working with all these advanced tools. What advice do you have for people who are uneasy about where we’re going technologically? What can they do to take control of this new world? Listen to more WNYC.

John Keefe: Listen to public radio.

Dan Costa: At least you didn’t ask anybody to pledge.

John Keefe: Yeah, no. I think getting out and trying your hand at some of this stuff, it’s easier than it ever used to be. There was a time where if you wanted to graph or plot on a map, a bunch of points, right? You needed really expensive software to just make a map and put a bunch of points on there. Now, you can do it in Google Maps, right? It’s super easy, and it’s free. That changed.

We’re talking, the software was like thousands of dollars, and now it’s free. It’s not exactly the same thing, but the concept has come down, right? In the same way a lot of things online, a lot of the hardware stuff is coming down in price in the same way that ‘Hey, it’s not actually that expensive to tinker.’ And if you have an idea, and you want to play with it, you should. That’s at least, I find that to be, both enlightening and just rewarding as a human, to just understand this stuff a little bit better, and then you understand your world a little bit better too, and you can ask good questions, you can think through some of the things like security and other things like that.

And if you’re a journalist, you can also ask better questions of your sources and of the people around you.

Dan Costa: Very good, and your book’s a great primer for that. Family Projects for Smart Objects.

John Keefe: Available on Amazon(Opens in a new window).

Dan Costa: How can people find you online and interface with you directly?

John Keefe: You can tweet at me, I’m @JKeefe(Opens in a new window) and my website and my blog is JohnKeefe.net(Opens in a new window).