After lunch, the Sunlight/Berkman conference moved to a format that I usually find pretty painful – the Speed Geek session. In these sessions, people give quick demos of their products at stations scattered throughout a room, and other groups move from station to station, hearing a pitch for a few minutes at a time. It’s kinda like all the bad aspects of speed dating combined with all the bad aspects of a venture capital conference.
But the presentations were extremely strong, the projects were interesting, and my fellow travellers asked good questions of the ten groups we met with. And so 90 minutes of 7 minute sessions went surprisingly fast, and I now know about 10 cool projects I knew very little about previously. Here’s a paragraph on each… which will probably take me longer to write than they took me to listen to:
– Ed Bender with the National Institute on Money in State Politics is working on putting information about contributions to state legislative races online. He tracks money gien to party committees and ballot measures – this is an enormous amount of data: 90,000 reports issued by 18,000 committees annually, totalling $2.4 billion dollars. The data has been heavily audited and peer reviewed, and includes coding of contributors into categories (labor unions, environmental groups, etc.) There are very impressive tools that allow people to create custom SQL searches or build widgets around APIs. But the most impressive part of the presentation is a set of scatter plots Ed shows, showing the relationship between money and election results in Arizona. Some years ago, you could see a very clear relationship between money and results, with the candidates who spent the most consistently winning races – after three rounds of electoral reform, the results from well and poorly funded candidates are very similar.
– Kerry Mitchell from Sunlight is pinch-hitting for Fedspending.org. It’s a project put together from OMBWatch which lets citizens investigate how federal money is spent, both in terms of federal contracts and grants. Useful data includes a table of top 100 contractors, with the ability to drill down and see whether these contracts are no-bid or transparent.
– Michael Wood-Lewis had a cool idea a few years back – he decided he wanted to know his neighbors in Burlington, VT. He started a neighborhood email list, which eventually included 90% of his neighbors. The content of the lists probably wouldn’t be interesting to anyone outside the neighborhood – a lost cat, worries about vandalism, etc. – but went a long way towards improving communication between neighbors. Given the success in his corner of Burlington, Wood-Lewis is trying to expand to the city as a whole, launching a project called Front Porch Forum. He’s divided Burlington into 130 neighborhoods, roughly 350 households in each. Some are thriving, with hundreds of people signing up – across the board, 5-10% of people across the city have joined so far. The next step involves generating sustainability revenue from microadvertising, reaching small communities with local businesses.
– Conor Kenny is one of the organizers of Congresspedia, a wiki-based project of SourceWatch which encourages readers to help develop constituent information sites for each member of Congress. Unlike Wikipedia, Congresspedia is professionally edited, is aggressive about requiring references for assertions made, and isn’t scared of banning information from certain sources (notably the LaRouche publicaion “Executive Intelligence Review”.) All users are email authenticated, and when folks get out of line, Conor says that a single email is often sufficient to keep the peace. The content created is released under a GNU documentation license, which means that Congresspedia can help repopulate Wikipedia.
– David Moore is standing next to Conor and pitching Open Congress, a project that seems to have a great deal in common with Congresspedia. (The site’s not live yet, though an associated blog is.) The project collects information, voting records, contribution records and similar types of information on individual congresspeople. If I understand the distinction between the projects, it sounds like OpenCongress is designed to have “more metadata” – information on what bills are being searched for most often and most discussed (the current iteration looks a little like Digg with popularity metrics for House and Senate Bills.) Open Congress also wants to host wiki-based discussions on legislation, which makes it somewhat similar to…
– MorePerfect is a wiki-based tool for creating policy and law. It’s the result of Tim Killian’s experiences as an organizer, trying to pass state initiatives in favor of medical marijuana. After passing a successful initiative in Arizona, he and his colleagues filed almost an identical resolution in Washington… and failed. As he and his team assessed the failure, they began reaching out to people who’d opposed the initiative and trying to incorporate their feedback into a new resolution. When they put forth a resolution the following year, the groups who’d previously opposed it didn’t actively fight it and it passed. Based on this lesson, he and Chad Maglaque decided to build a space where groups could try to interact with both supporters and opponents to create legislation that can actually pass. We had a lot of questions about how this will work in practice – how can MorePerfect avoid some of the problems that have faced political discussions on Wikipedia? MorePerfect has, understandably, abandoned NPOV (neutral point of view) in favor of a general rule: Be constructive. As more people use the tool for realworld campaigns, we’ll get a test of how that rule works in the real world.
– Campaigns Wikia is a project launched by Jimmy Wales on July 4th, 2006. It’s a project to get grassroots information on political campaigns in any place and in any language around the world. Anyone can add an election to the electoral calendar, and versions of the site have been put forth in 11 languages. Unlike Wikipedia, Campaigns Wikia advances the idea of APOV (Allow Points of View) to allow people on different sides of a campaign to make their arguments in separate, tabbed pages. Chad Lupkes, a lead volunteer on the project, pointed out a distinction between issues and campaigns – issues (“Should the US permit gay marriage?”) may be shunted to an “issues wikia” leaving campaigns as a space for political campaigns (“Should there be a constitutional amendment permitting gay marriage in the state of Massachusetts?”).
– I missed the pitch for MetaVid, but Ruby filled me in (as well as getting a good shot of me), and I love the idea. MetaVid grabs video from CSPAN and CSPAN2, encodes it with open algorithms and makes it available on the web under open licenses. Critically, they use closed captioning information and OCR to determine who’s speaking at which time, making it possible to search for clips that feature your congressperson or an issue you care about. Very, very cool.
– Capitol News Connection is a fantastically cool service from Melinda Wittstock and her team which provides coverage of Congress for local NPR news stations. The CNC correspondents talk to congresspeople about local issues, avoiding national issues that might be covered by conventional NPR reporters – as a result, congresspeople often seek out CNC reporters to talk directly to constituents about local concerns. NPR stations buy a set of “points” from CNC for $9,000 to $35,000 a year – one point buys you a sound byte, three a news spot with an actuality, up to forty points, which buys a day of a CNC reporter following a congressperson all day. (Melinda has a funny story about following Representative Henry Waxman for a day and watching him crash a buffet hosted by Minnesota farmers.) It’s expensive to provide this sort of “shoe leather” journalism, and CNC is supported not just by revenue but by grants. It’s very cool, and I hope to lobby my local NPR station to consider becoming a client.
– Quintus Jett leads the Gentilly project, an effort to show the “middleground” of what’s happening in New Orleans. He points out that New Orleans tends to be covered one of two ways in mainstream media: either as a city that’s recovered from Katrina and is on the rise, or as a destroyed wasteland. The truth, of course, is somewhere in between, and it’s this truth that the Gentilly project is documenting. Jett has been assigning pairs of volunteers parts of the Gentilly neighborhood. The pair walk the streets and document whether each house has been gutted, has been fully restored, or is still uninhabitable. They produce maps of neighborhoods which show the progress of bringing Gentilly back to life, giving a more realistic and nuanced picture than what’s generally portrayed in the media. Many of the lessons learned in the project have to do with how Open Source models do and don’t apply to volunteer effortsn – Jett observes that the myth of the leaderless project disappears pretty quickly when someone has to put 50 people out in the field on an afternoon. The project is going so well that ambitions are expanding beyond the Gentilly neighborhood, and expanding to cover the whole city.
Looking over these projects, a couple of generalizations come to my mind:
– We’re at a very early stage of development in this field. Give developers a couple more years and some of these projects will die, and some will be swallowed by other projects. Meetings like this are useful so that folks working on very similar efforts can meet each other and decide whether or not they want to have separate efforts or to join their forces.
– There’s an amazing wealthy of political data available in the US… in contrast to some of the other countries I work in. In many ways, the focus of these efforts is to get people to pay attention to and care about this data, not to collect data that isn’t available elsewhere. It will be interesting to see which of these tools might be valuable in less data-rich environments and which don’t make sense in very different circumstances.