ReportHers

Q&A: Emily Withrow, Bot Editor at Quartz

EMILY WITHROW is editor of the Quartz Bot Studio, and is on leave from Northwestern University, where she holds an appointment as assistant professor of interactive narrative at the Knight Lab and Medill School of Journalism. At Quartz, she explores the ways people and bots interact, and how journalists can use bots and emerging technology to connect with new audiences. Previously, she worked as an editor at The A.V. Club. She is based in Chicago. Follow her work at Quartz at bots.qz.com.

–

ReportHers: What is a typical day like at the Quartz Bot Studio, and what is your role on the team?

Emily Withrow: I’ll admit that this question made me laugh—there’s really no typical day. Or at least, not yet. The Studio is still in its formative stages. Right now, full-time, it’s just me (editor) and John Keefe (developer). We also have an awesome product intern, Eva Scazzero, working with us. We each have different strengths, and we’re working on several projects at once. Our first year is funded by the Knight Foundation, so we have a mandate to experiment, and to document those experiments as we go. One of those experiments is likely to be for voice (Alexa, Google Home, HomePod), one is likely to be chat-based (like Slack or Facebook Messenger), and one is likely to try to tackle open responses from users (free text, as opposed to options). So we’re running after each of those three individually, and plan to roll out some tools for journalists along the way.

So my days are part exploration—understanding what these spaces are like, what works and what doesn’t—and part trying to build things for those spaces. We’re doing design research to fill in some of the blanks. None of these are truly mature platforms, so we have more questions than answers at this point.

RH: There are a lot of jobs in journalism—and in the world in general—that didn’t exist just a few years ago. How do you think about where you are in your career now, and where you want to be?

EW: My career has taken a number of twists and turns already. The through line of my work is really an attraction to big, messy problems. Easy answers and predictable days never did it for me… so lucky(?) for me, I entered the journalism world right around the time the housing market fell apart, and around the time mobile and social started to come into their own. No one was ready. Cutting and pasting stories from print to web wasn’t working.

I was at The A.V. Club around that time. By all traditional metrics, we were performing well, but we were also going through layoffs, having to cut costs. I wanted to be part of the solution, and for me that meant telling stories that reflected how and when they were being seen. I thought there was a danger in thinking of the website as this Mother Ship to which all traffic must go—especially when our content was being shared and consumed in so many other places. I wanted to be able to build for those new spaces. So I taught myself to code. I started paying more attention to emerging technologies and platforms, and tried to understand what people were doing, and what type of stories would work well there. I ended up teaching nontraditional narrative at Medill, and eventually, that led to bots, and to Quartz. I’m right where I want to be! In the future, I hope to follow the same path—to embrace promising technology that could change how we tell stories and deliver/consume information.

RH: What was the first bot you made?

EW: The first bot I made wasn’t a “real” bot in the purest sense, because it wasn’t really doing much intelligent on the backend. It was scripted, and was only a bot in the sense that it did asynchronous content delivery. (Though, I guess I shouldn’t say it wasn’t intelligent, because we did rely on Amazon for all its NLP smarts.) At the Knight Lab, I was playing around with the idea of a personal news anchor. I’d been using Alexa frequently at home, but was frustrated that all of the news was confined to the Flash Briefing section—this device I spent so much time in dialogue with, when I asked it for the news, just turned on NPR. It seemed like such a disconnect for me.

So I started to toy around with the idea that we could provide additional context for the news. I designed an experience around the idea of a water cooler conversation you might have with coworkers about the news. Alexa (or our “skill”) would float a headline: “An Uber driver is accused of killing six people in Michigan.” If you’d already heard that story, you could ask follow up questions or move on. If you hadn’t, you could ask background questions.

The bot never saw the light of day—really, I’m not sure people want this experience—but we learned a ton about how people talk with bots, how bots should talk, how to prototype and test for voice devices, and what people ask about news headlines. It was a really successful experiment from that perspective.

RH: What is the biggest risk you’ve taken, or boldest move you’ve made to get a job, interview or internship?

EW: Oh, wow. I’m so timid when it comes down to it! I’m getting nervous just thinking about having done bold things. But it’s something that I force myself to do, pounding chest and all. I’m lucky enough to have loved just about every journalism job I’ve had. (Not all pieces of every job, but on the whole.) So leaving has always felt like a huge risk. My career has included some really wild swings. From McKinsey & Company to The Onion, then to wildly stable academia, then to an experimental studio.

I guess one of the boldest things I’ve done has been to write about all of the mistakes I’ve made in that process. At one of those jobs, for example, I didn’t negotiate my salary. So I wrote a piece about all the shame I feel about that—and the lessons I learned for the next time. I believe writing about my failures makes it easy for other people to take risks. It’s certainly not easy for me.

RH: I had a similar experience, and the sting of having not negotiated—and for me this was almost 10 years ago—still motivates me to be super bold with money. At one job, I asked for a pretty big raise after doing some research that led me to believe I was paid under market value. I didn’t get the raise, but I am so glad that I tried, and it’s made me more comfortable talking about money in the future: I asked for a really big raise and no one died.

EW: That’s so incredible! You lived to tell the tale. There are so many things I grew up believing about the way women should behave—and even though I know all that now—it’s incredibly hard to shake off. Wanting to please people and not cause a fuss is my worst enemy #1. I’m working on it all the time! I’ve gotten so much better, and bolder, but it still makes my hands sweat every time I have to speak out. I’m still in the “faking it” part, but I’ve gotten very good at faking being a confident, take-no-prisoners kind of woman.

RH: Yea! I am not sure to what degree we fully leave the “faking it” part, but in my mind, as long as you push through it’ll be ok!

EW: Totally. I’m in my mid-30s, know a ton of incredibly accomplished women, and everyone says they feel the same. Onward!

RH: Speaking of doing something new, what advice do you have for small bot projects can you suggest for people who want to get into bot making? (I am working on this right now. I have a set of circuit boards on their way from Hong Kong as we talk!)

EW: That’s awesome!! One of my huge frustrations in the advice-giving sections of the Internet is that you should “just make something.” It’s the “just” that really bothers me. Making things is absolutely the best way to learn something, but sometimes it’s hard to come up with an idea for a useful project.

The world of bots is so huge—they can include physical sensors and gear, or be as one-sided as automating tweets, like @tinycarebot, which reminds you to take a deep breath or go on a walk. So my advice is to choose one corner to play in—whether that’s Slack, Twitter, etc., and start really small. Can you get just one Tweet to show up? How about five? How about five on a schedule? Where are you pulling the Tweet from? How do you level up from there? Could you scrape it from a site? Pull from a separate JSON file?

Another way to dabble is to fake it! There are tons of prototyping tools out there that allow you to design or build a bot without going deep into code. We’re working on running through many of them at the Studio, and plan to write up our recommendations as we go, as well as projects you can replicate.

RH: That’s great. Thank you so much for taking the time to chat!

EW: Thank you!