In Year of Political "Big Data," NationBuilder Makes Voter Data Free
BY Nick Judd | Thursday, September 13 2012
The team at NationBuilder announced Thursday that they were releasing API and limited bulk data access to a nationwide voter file with records on 170 million voters, for free.
As NationBuilder co-founder Jim Gilliam explained in a blog post, data on voters is becoming increasingly important for campaigns, advocacy organizations and researchers hoping to understand the American electorate. At the same time, U.S. voter records have two crucial and widely accepted problems.
First, the voter files, which are collected at the county level and kept at the state level, are often victims of poor management. Boards of elections frequently keep on the books voters who are dead or have moved, for example. This is not an issue for voters who have died or emigrated, but it is an issue for anyone using elections officials' tallies to make decisions about who to contact for a campaign or where to spend resources.
The second problem is that voter data is often hard to use. As public records, they are available, with restrictions and at a cost that varies from state to state. But even the best-kept voter file requires time and attention to transform into something usable for research, outreach, or planning. (Trying to find ways to weed out those dead or relocated voters, for one.) That requires some combination of technical expertise, money, or political connections.
It is in this context that NationBuilder is offering campaigns, researchers, and generally anyone whose use of the voter file would comply with applicable law, free access to that slice of the file that they need. Developers get API credentials and access to a sandbox where they can build and test applications using fake data. A campaign with data access can authorize developers to work on their behalf, allowing web applications to pull from real data on the fly as well as unlocking bulk downloads for the developer to work with elsewhere.
Open political innovation?
"I'm a developer, I've wanted to build off of this data for like a decade and it's just completely impractical because it would just cost a ton of money to bring this together, a ton of time, you wouldn't get access through all the parties, the tools are all partisan," Gilliam said Thursday. "There's no ecosystem around it. And that's really stunted the innovation in the political tech world."
NationBuilder's voter data release comes at a time when data is dominating both the study and practice of political campaigns. Most observers readily acknowledge that the 2012 presidential campaign will be decided by the outcome of a handful of battles in just a few key swing states — identified thanks to the reckoning of data scientists and pollsters. Beyond the presidential campaigns, ideological groups are looking to exert influence in state legislatures, town halls and even sewer control boards. They're doing this by matching their understanding of where they might have successful candidates against their understanding of where their foes are weak or absent, and groups like American Majority Action are deploying tools for their volunteers that provide additional intelligence about the voters they meet at every step of the way.
The kinds of tools that the Obama and Romney campaigns have developed and the predictive modeling put to use by all sides has shown the potential power this kind of campaign intelligence can unlock. It's just that in order to get to the point where you can build applications on top of useful data, right now, you probably need to have allegiance to a political party or a major outside group like a union or a PAC. And you need to be doing something they are specifically authorizing you to do.
"My feeling is that by opening this up, particularly to developers, that a whole new set of innovations, developers, and startups can emerge around it to start to solve all kinds of problems that we couldn't hack before," Gilliam explained.
In time for 2012?
It's not fully open, of course — NationBuilder is still a gatekeeper, although Gilliam explains that his company is basically trying to make sure people who use voter data are not unintentionally violating some piece of the patchwork of state voter laws in doing so. And the data has been cleaned only by checking it against a national change of address database and by picking the most recent record of each voter, meaning the voter file is nowhere near as refined as what is available through more expensive or partisan-affiliated means. Gilliam hopes that individual voters will "claim" their records through NationBuilder, verifying their registration and updating their data if it's incorrect.
The Help America Vote Act required each state to keep a state-level voter file, notes Drew Brighton, the CEO of voter data firm TargetSmart Communications and one of the companies that NationBuilder competes against. But HAVA did not offer too many specifics on what states must do to keep that file up to date and accurate, Brighton said.
"I think in spirit what they're doing is providing a service to make it more open and accessible," Brighton said, approvingly, of NationBuilder. "But having worked with data my entire career, I've always found that bad data can be a lot worse than no data at all."
He said that NationBuilder's data-cleaning methods were "a good start." But the additional work that data firms (like his, naturally) put in is necessary to bring a dataset up to modern standards, he said.
The release comes with less than 60 days to election day — not a lot of time for a campaign to adopt, learn and develop on a new platform. It does, however, come with ample time between now and early January, when the next Congress will start its session and the next presidential term begins. And voters are important as constituents in issue advocacy, too.
The product, which NationBuilder is calling their "election center," might not have arrived in time to make an impact in November. We'll see. But it is certainly fodder for the ongoing conversation about voters as data points, and data points as a kind of political commodity. The current machinery of politics is leaning heavily on the idea of information about voters being a kind of currency, like political fundraising dollars. If voter data becomes more like a commodity — available at lower cost and greater ubiquity — Gilliam's theory is that this will make democracy function better because more people will have equal access.
Many significant points about privacy aside — voter files include name, address, age, party affiliation and which elections each voter participated in — if voter data becomes a basic requirement for being an effective participant in politics, this being a democracy, shouldn't everyone have access to a clean, accurate voter file?
I posed this question to Brighton, who is in the business of delivering voter data, for a fee, to basically anyone who is not a direct opponent of the Democratic Party. (We've previously reported that TargetSmart is a licensed broker for access to the national voter file that state Democratic Parties collectively own.)
"Having spent a career selling names and addresses for a living, let's be blunt, I don't think so," he said.
Maybe some sort of census block-level aggregate thing he could get behind, he said.
"When you're actually talking about running campaigns and voter contact, influencing the future -- phone calls, direct mail, all the ways you reach out to voters, to do that right, and to actually maintain that as a long-term thing, there's actually got be some amount of money spent on that to ensure quality."
Our conversation continued. What if this became a responsibility of state boards of elections, to keep more accurate and up-to-date voter rolls?
In that case, he said, maybe. That would remove data quality as part of "the strategic advantage" that each party and ideological group now seeks to maintain, improving efficiency in communication for everyone. In the meantime, he said, would continue selling names and addresses.