Josh Cohen used to be a high-profile standards designer at Microsoft. Back in 1993, Cohen even helped craft some of the Internet's most important underlying technical features, including a little thing known as HTTP.
Cohen ran the technology behind the effort — a task that was about as easy as building a jigsaw puzzle using pieces that came from a dozen different boxes. Based on that experience, Cohen is now at the head of an open-source software movement that aims to eliminate that problem once and for all. It's called the Open Supporter Data Interface, and the goal is to get previously incompatible systems talking to one another in the same way that any PC and any word processor knows what to do with a .doc file. If he succeeds, it could usher in the next chapter in political technology.
The problem with Big Data
Every competent campaign needs a set of basic online organizing tools these days: A tool to send mass e-mails. A tool to set up rallies and other events. A tool to manage fundraising. A lot of companies exist to serve these needs, but no single firm is the best at meeting all of them. So Cohen wound up doing a lot of mixing and matching. The extra work would've been no big deal if the services weren't also information silos, making it harder to take data from one tool and plugging it into another.
Here's an example. One of the ways that Washington United for Marriage raised money was by encouraging everyday supporters to act as fundraisers. When reporting to the government how much money it had pulled in, Washington United for Marriage had to somehow aggregate that data from various vendors. The problem was that not every provider allowed Cohen to grab the data automatically — and even when he did get a hold of the information, the inconsistencies across spreadsheets made it a chore to merge anything.
Cohen said he was forced to hack together a rickety little program that did the job on an ad-hoc basis.
"It imitated a Web browser," said Cohen. "A virtual mouse clicked on a button that said 'download CSV file,' pretending to be a human. It'd then parse that CSV data and make them usable. It was very fragile and painstaking; whenever the vendor updated their page, it broke our script. We had to write 10 of them, and we had to keep them up to date."
It's easy to see how this would be a huge waste of time at any scale. Washington United for Marriage surely wasn't the only group facing this problem, Cohen thought. As he later discovered at a progressive conference in Washington, D.C., organizers were clamoring for something that made data generated by one vendor comprehensible to another.
Everyone wants to talk to each other
The demand for a solution convinced a handful of powerful organizations in Democratic politics to take on the project together. The Open Supporter Data Interface (OSDI) ropes in Catalist and NGP VAN, two of the heavy hitters in left-wing data and political organizing and with strong connections to the Democratic party. Other members include Salsa Labs — one of the most popular campaigning tools for smaller groups — and Blue State Digital, the political consulting firm founded by alums of Howard Dean's 2004 presidential bid.
Many of these companies provide one or two staff members to contribute a few hours a week to OSDI. Others are hashing out in negotiations what the final product ought to look like. The project is on Github, the online repository for open-source software development, and decisions are made collectively. And oh, yes: The common API that'll emerge out of this process will only be available to liberal groups. Building a standard is an exercise in quibbling over details; Cohen is currently deep into talks over the correct format everyone should be using to identify a voter.
"The way we talk is to agree that, 'Yes, "person" means "first name" and "last name," not "given name" and "surname," ' " said Cohen.
This is where Cohen's background in designing standards comes in handy. And it's why some are so optimistic about OSDI after having repeatedly tried and failed to implement something like it over the years.
"We've never had someone like Josh. He's like, in love with standards in a really bizarre way," joked Seth Bannon, the chief executive of Amicus and an OSDI partner. "I don't think we've had someone who's so purely a standards person. And he's seen as impartial, because he doesn't come from any of the providers in this space."
While OSDI won't be ready in full for the 2014 elections, Cohen told me the group expects to complete a basic version of the API in the second quarter that will offer some level of interoperability among service providers. Campaigns that test it out in 2014 will highlight problems to be fixed in the next release. By the time the 2016 cycle rolls around, Cohen hopes to have all the major kinks worked out.
Following what's by now become a pattern in politics on both sides of the aisle, OSDI will only be available to left-leaning companies and campaigns Update: While the final API will be made available to anyone, including conservatives, the people designing the product are by and large liberals. "While membership in some OSDI committees, and governance of the effort, is restricted to organizations sympathetic to the progressive cause," Cohen wrote in an e-mail, "the API itself is published publicly on Github and there aren't any restrictions on who implements it."
A potential industry-wide shift
Beyond making life a lot easier for liberal campaigns, OSDI promises to introduce second- and third-order effects that could prove even more significant. The left already runs a sophisticated operation that plucks promising new organizers out of campaigns, trains them in online boot camps, and sends them back into the field where the lessons of each successive political cycle get fed back into the system. But OSDI could reduce the amount of training needed to get new organizers up to speed if all the major vendors in the space speak the same technical language. Training itself could be standardized so that everyone coming out of the pipeline boasted the same skills. And because of that, the graduates would in turn start to resemble components in a car or computer — any one of them could land in any campaign and get to work right away, no matter how unfamiliar the issue or the candidate.
The ecosystem of Democratic politics would also start to shift. Mixing and matching service providers would no longer introduce barriers, so businesses would likely begin to specialize more selectively, according Charles Parsons, director of product management for Salsa Labs. That's certainly the plan for Salsa itself. Like other companies in political organizing, Salsa Labs was initially built as a one-stop shop for campaigns.
"The platforms are expected to be one platform for everything because the data isn't easily moved," said Parsons. "[OSDI] opens up possibilities, because not every vendor is going to align with what the client is trying to do. It allows the clients to go ahead and use Salsa for e-mail and another tool for membership, and just use OSDI to bring the data over."
Amid the reshuffle, smaller startups would begin to identify business opportunities not being filled by the larger vendors, introducing another layer of competition and new data-related services.
All of that is still a long way off; OSDI is barely beyond the prototype stage. Yet the demand for it is evident. It used to be that the more data a campaign had, the better off it was. Now, interoperability — or the ability to move data from one place to another, thereby making it useful — may become just as important.