(A Storify of the #hacks4d hashtag is available here: http://storify.com/azavea/hacks-for-democracy-round-2)

Hacks for Democracy was the first hackathon that Azavea has organized, and we felt it was a smashing success. For that, we owe a huge debt of gratitude to the participants and the sponsors.  We took a gamble on organizing Hacks for Democracy as a three week long “hackmonth” rather than a typical weekend hackathon. As far as we know, Hacks4D is one of the first civic hacking efforts to combine so many events for such a long time period: a pre-hackathon brainstorming session, a weekend hackathon, and three weeks of follow-up events and a second round of judging and prizes. At the outset, nothing was guaranteed. Maybe nobody would show up to the follow-up hacknights, we thought. It was possible we’d be left with boxes and boxes of leftover pizza and beer. Maybe hackathons and the apps created from them really are just fun weekend projects, and people wouldn’t have the time to stick together in teams and continue improving on them once their real lives came back into their schedules. But we wanted to try something different because historically, civic hackathons have too often had a sustainability problem – while great for prototypes, there’s not enough time in a weekend to build, test, and deploy a high-quality piece of civic software. Excitement dwindles after the weekend, and the community that came together disperses again. While there have been exceptions — locally, Sheltr, Cost of Freedom and Councilmatic are all examples of successful follow-through — for the most part, the critics are right, and we wanted to try to improve on this.

We were wrong to worry that no one would show up (but more on that in “Reflections” below). Newbies and veterans in the Philadelphia civic hacking community came to every single one of our follow-up events (and ate almost all of our pizza). The applications were greatly improved upon since their initial stages at the end of the first weekend. My team, State-Gov-Tracker, even added a new team member in the followup weeks. For those who like numbers:

• 50 attendees at our first Friday night brainstorming and presentation event
• 30+ attendees over the Saturday and Sunday hackathon
• 10-15 attendees at every followup Monday hacknight
• 25+ attendees at the final Friday awards ceremony
• 6 teams/apps total
• 4 teams/apps improved and carried through all events
• at least 1 new team member added after the first hackathon weekend

The Final Four

The final judging event on October 5 featured four teams who returned with significantly improved apps.

The first team to demo was “The Undecideds,” composed of Azavea developers Adam Hinz and Bennet Huber. At the end of the first weekend, Hinz and Huber had a working heatmap of 800,000 Philadelphia voter registrations using the open-source GeoTrellis. Over the followup weeks, the duo worked with City Commissioner Stephanie Singer to geocode 40 million addresses from Philadelphia County voter registrations from historical elections. That’s nearly 39 CDs worth of data – data that previously only existed on physical media at the Commissioner’s office before their project. Undecideds also added an improved user interface to change parameters for very specific geographic areas of Philadelphia and model possible election outcomes. This comprehensive use of geospatial data and analysis was enough for the judges to award the team the Esri-sponsored “Best Geospatial App” prize.

Next came a large (9 hackers!) team, State-Gov-Tracker.  This was the team I worked with (that’s me in the red shirt above). Chris Brown, Chris Nies, Jason Blanchard, Josh Borkowski, Lauren Gilchrist, Charlie Milner, Josh Darr, and I really only had command line data-gathering Python scripts at the end of the hackathon weekend and we were badly in need of a designer. Luckily, Chris Nies’ friend and “CSS guru” Nick Weingartner was interested in joining in. Our team were all regulars at the Monday evening hacknights and kept in touch regularly via a 100+ post email chain. Our app mashed up APIs like Cicero, Facebook, Twitter, CartoDB and OpenStates, along with 700+ MBs of screen-scraped data from the Pennsylvania legislature’s website. We came out of the process with a very quality piece of software that will make it easy for regular non-politicos to check a dashboard and learn about their Pennsylvania state senators and representatives and keep tabs on their ongoing activities. We hope to go public soon and keep it relevant with blog posts of political analysis (from Chris Brown, who’s a Political Science PhD candidate at U-Penn). While we were gunning for the “Most Improved” prize, the judges actually saw fit to award us with the “Greatest Potential Impact” award – something that we’re all proud of (and maybe even a bit shocked about!).

The third demo of the evening was Electory, by Chris Alfano, Brett Goldman, Tim Wisniewski, Mjumbe Poe, Jim Bazis, Donovan Preddy, and others. This app was complete enough after the first hackathon weekend to win 3rd place in Round 1. Electory aims to be a crowdsourced database of neighborhood-level election leaders, volunteers, and party committee-people, making the process of contacting the most local and most common elected representatives much easier. Electory added crucial functionality to their app in the followup nights, including electronic authentication of users who wish to add or edit information against the city voter file, using basic questions like date of birth, house number, and real name. Electory is effectively the confluence of multiple proposals brought to Hacks for Democracy for a dataset of committee-people, a real need in the City Commissioner’s office, and Tim Wisniewski’s idea for an improvement on electronic voter authentication systems. The app demonstrates that matchmaking and sharing ideas at the beginning of civic hackathons is crucial, despite the potentially competitive structure of these events. Had each of these teammates worked on separate projects, the resulting software would not have been as comprehensive.

Our final demo of the evening was Mike Zaleski’s SEPTA Routes n Reps app, which also had some help from City GIS folks Sarah Cordivano and David Walk. Zaleski transformed the UI of his original app by using Twitter Bootstrap to create a quality mashup of SEPTA bus and train route data, OpenStates, and Cicero legislative districts, displayed through CartoDB. The aim is to enable SEPTA commuters to more easily contact their elected representatives regarding service and schedule changes. The judges liked Routes N Reps’ new design enough to award it the “Most Improved” prize.

Zaleski, being a lead developer at SEPTA itself, was even able to host his app on the SEPTA website and show it off to SEPTA’s Government Affairs department. Zaleski said that SEPTA’s government team is potentially interested in the app, and gave him a list of possible improvements. (He might need a break first!) We’re all certainly hopeful that this hackathon project continues its life as a functional piece of the Philadelphia area’s e-government infrastructure!

They All Won

The judges got into the app sustainability groove as well.  The three Round 2 prizes were supposed to be $500 each. Upon seeing the quality of these apps, our judges – secondmuse‘s Mike Brennan, PPIIN‘s Neil Budde, and City Controller candidate Brett Mandel – decided that they wanted to encourage everyone to move forward. So, they knocked the main prizes down to $300 and gave each of the 4 apps that made it through Round 2 $150 to go towards hosting and development costs. Furthermore, Neil Budde said that the Philadelphia Public Interest Information Network is interested in partnering with some of the apps to get them more widely used. I know that’s something our State-Gov-Tracker team is considering, and it wouldn’t surprise me if other teams would be happy to collaborate with PPIIN as well.

Reflections

I really enjoyed this hackathon. The follow up events were awesome and a great opportunity to focus on making a finished product. I learned a lot and met some great people! Will definitely participate again in the future.

This was my first hackathon and I can’t imagine it being done any other way.

A little more time committment than I would have liked.

The above are all comments I heard from various participants. I certainly think the positives outweigh the negatives for this style of extended-hackathon. Most people seemed to enjoy it and had a lot of fun, and the apps are WAY better after so much seasoning time. With the judges’ decision and PPIIN, SEPTA, and the Commissioner’s office all interested in improving these apps, we have a bright future ahead for the Hacks for Democracy community. But as I hinted at way up in the first paragraph, and as that final quote highlights, it is important to reflect on the choices we made and challenges we had with this type of event. Like anything, it wasn’t perfect. Is this type of extended model for hackathons something the civic hacking community should adopt in the future? Let’s take a look at some of our organizational choices.

• Two Venues – We were worried about our office being able to handle the bandwidth needs of lots of hackathon participants for an entire weekend, which was why we partnered with Venturef0rth to host the main weekend event. Meanwhile, we expected the followup events to have a somewhat lower turnout, so we kept hosting those at Azavea. This worked well, and Venturef0rth was a generous and cooperative host. I know at least one visitor to the hackathon came because Venturef0rth was close to his apartment and he wanted to see what we were up to. If we had only used one venue in one neighborhood, perhaps we would not have received the level of participation that we did. However, it did add logistical challenges: we had to get our own folding tables and couldn’t use our own office furniture, for instance.
• Swag and Sponsorships – We decided not to go with swag or t-shirts for Hacks4D – partly for cost, partly for the organizational difficulties it would have presented. Nobody seemed to miss it. Instead, Azavea gave all interested hackathon teams 10,000 Cicero credits for a year, and CartoDB offered 1 free year of its Magellan plan. The State-Gov-Tracker and Routes and Reps teams each took advantage of both of these offers – arguably benefiting the apps more than a t-shirt. Other sponsors, Knight-Mozilla OpenNews, Esri, secondmuse, Jarv.us, and ElectNext, provided the funds that made our prizes possible.
• Special Guests and Judges – We were glad to have City Commissioner Stephanie Singer, Jared Marcotte from the Voting Information Project, Anthea Watson from the Google Elections Center, Michael Strickland from the New York Times, and the Sunlight Foundation’s Bob Lannon at the events to demo their APIs and tools and provide help from their vast experiences. We also had some “celebrity” judges, including Philadelphia Chief Data Officer Mark Headd, the William Penn Foundation‘s Phoenix Wang, secondmuse’s (Random Hacks of Kindness) Mike Brennan, PPIIN‘s Neil Budde, and candidate Brett Mandel. I felt like the diversity of tools and help presented by these guests and the diversity of perspective provided during judging made the app submissions and eventual winners that much better. Plus, by attracting members of the City government, the potential for further community and partnership development (the Undecideds, Electory, and the Commissioner, for instance) is definitely there.
• Brainstorming and Matchmaking – In my opinion, the pre-hackathon presentation and brainstorming night was critical to making Hacks for Democracy what it was. Some people come to civic hackathons with ideas, other people come with skills and need an idea to latch onto, and many people are new to the whole thing and need a guide. We felt that our mission as organizers was to help get teams to coalesce and bond from the first night because hackathons should build community first and apps second. We organized it “town hall” style, where we presented our APIs and sponsors, then opened the floor for anyone to present an idea, and afterward let people meet each other without the pressure of writing code yet. There was no coding at all on the first evening! That low-impact focus on the first night also got a lot of non-developers in the door, and some stayed for later events.
• Charging for tickets – Unlike many hackathons, Hacks for Democracy was not free to attend. We set up a TicketLeap site ahead of time and were charging a nominal $5.00 fee, which included food and refreshments at all 7 events.  We had sponsorships to cover most of our costs, so fundraising or making money was not an issue. Rather, there’s a theory that charging for hackathon tickets results in a more effective count of who will attend the events. It’s a “commitment fee” – people have to put their money where their mouth is and promise to come. A commitment from everyone makes things like pizza orders and venue setup much easier to predict. In practice, I’m not sure charging was a good decision. Quite a few last minute walk-ins never bought a ticket, and while we were very glad to have them, that seems unfair at some level.
• Scheduling and the Follow-Up event Structure – When planning an extended-length event, the potential for scheduling conflicts rises quickly. In addition to PennApps conflicting with the main weekend, General Assembly held an event in Philadelphia the night of our final awards, which also probably impacted our turnout. In our planning process, we limited the weekends we had to pick from due to the voter registration deadline in Pennsylvania on October 9. Our follow-up hack nights were always on Monday, preventing anyone who had a recurring meeting then, for example, from coming. And then we have the quote from the attendee above about the follow-up events being too much of a time commitment. These are tradeoffs to an extended event, and the first especially highlights the need for local tech communities to communicate with each other about the events they’re organizing in advance. However, while I can’t speak for everyone, I heard many positive comments about the Monday evening hack nights and having three weeks to work on projects. The hack nights were very informal and very easy to plan (we essentially just kept our office open late and ordered pizza and beer), and around 10-15 participants a night from all different teams attended each of them. I consider our schedule one of our key successes.

So, did our gamble with Hacks for Democracy’s “hackmonth” format succeed? You can use one of the apps and be your own judge, but count me as at least one hacker who’s hoping for more well-planned extended hackathons in the future!