Joshua Tucker (JT): You just launched campaign-finance.org, a new tool for analyzing campaign finance data. What is it and what does it do?
Meredith Broussard (MB): Campaign-finance.org is an artificial intelligence tool to help journalists quickly and efficiently uncover story ideas in campaign finance data. We take open data published by the Federal Election Commission and present it in an easier-to-understand format. It includes data for all of the 4,000+ federal candidates in 2016, as well as more than 17,000 political committees.
Nicknamed Bailiwick, the tool organizes the data into visualizations that are easy to understand. For example, you can use it to see which PACs and super PACs are supporting a candidate, and what these groups’ total spending has been. You can also see the groups that oppose the candidate, and how much they have spent.
JT: What motivated you to produce this tool? Why did you see a need for it?
MB: I started my career as a computer scientist, then quit to become a journalist. Today, I teach data journalism at NYU’s Arthur L. Carter Journalism Institute. My academic research focuses on artificial intelligence for investigative reporting. I am interested in building technology that helps to uncover, understand, explain, and (hopefully) solve social problems.
Funding for investigative journalism has been shrinking for decades. A few years ago, I became interested in finding new ways to use technology to automate discovery so that journalists can do what they are good at doing, but faster and better.
My last big project was developing an AI tool that helped me uncover textbook shortages and funding problems at Philadelphia public schools. I thought the same core technology could be applied to other public affairs topics, like campaign finance or transportation. I was fortunate enough to get a grant from the Tow Center for Digital Journalism at Columbia Journalism School to develop the campaign finance tool. My collaborators and I worked on the tech for a full year before we launched it.
JT: Has anyone used the tool for reporting?
MB: Yes! I gave a talk about the tool in October at the Computation + Journalism Symposium at Stanford. Jason Clampet, the founder of Skift.com, happened to be in the audience. He called up campaign-finance.org on his laptop, got a handful of story ideas, and immediately assigned one to a Skift writer, who turned out “Clinton vs. Trump: Where Presidential Candidates Spend Their Travel Dollars.” Here’s a quote:
With one month left before election day, Republican candidate Donald Trump has now spent more than $5.9 million on flights hosted on his own private jet service TAG Air, according to the latest operating expenditure reports from the Federal Election Commission (FEC) filed on September 22.
In my political reporting classes, 25 student journalists are using the tool to find stories.
JT: How do you hope journalists will use the tool for the remainder of the 2016 campaign? How about after the election?
MB: There’s not much time left for long-term investigative stories. Those will have to wait until the next election. In the few weeks remaining before the general election, I hope that journalists will turn to Bailiwick for ideas for quick-hit financial stories.
For example, when you look at Donald Trump’s campaign committee spending, you see that the campaign has spent $1.9 million on hats. That’s up from $1.5 million in August. I would love for someone to write about this.
Bailiwick also has a feature that alerts you when there is a filing related to a candidate or a race that you are following. It will be interesting to use this to track what happens to candidates’ war chests after Election Day. If a candidate has loaned money to the campaign, the campaign usually pays it back; what is the rate of interest? If the campaign has money left, does a politician save it for the next election or give it to allies? If a candidate received a lot of donations from PACs and individuals in a particular industry, does the person — if elected — serve on a committee that makes decisions about that industry?
After the election, we’ll look for funding for the next phase of the project. We’ve been approached about commercializing the tool; it could be useful for compliance in highly regulated industries like finance or pharma.
I’d also like to build out the automated story suggestion feature, which we started but didn’t finish. When I used to explain, “I built an AI tool to help journalists find story ideas,” people would say, “So it’s a machine that spits out story ideas?” I would explain that no, it’s more complicated than that. Eventually I started to wonder: Could I build a machine that spits out story ideas? We built a proof of concept. I’d like to finish building it so it can be used before the 2018 election.
JT: More generally, what do you see as the biggest challenges facing the emerging field of data journalism?
MB: I usually describe data journalism as the practice of finding stories in numbers, and using numbers to tell stories. This is a very exciting time for data journalism. People don’t look at it as an outlier anymore. At outlets like Vox or FiveThirtyEight or The Upshot or ProPublica, data journalism is now simply part of what they do as journalists.
Campaign finance journalism is now where some cutting-edge collaborative projects are underway. Bailiwick can be used together with tools from ProPublica and the Center for Responsive Politics to illuminate how dark money flows in politics.
I try to assume that most folks in politics are acting out of a sincere commitment to public service, and are good stewards of public funds. But that’s not always so. We need a free news media to watch out for the public interest. Technology can help.