Like Clay said, I’m the new guy. Well, not entirely new — I’ve been at Sunlight since late 2008. But I’m the one who’s going to be trying to fill the enormous gap that Clay is leaving. I thought I’d start to explain how I want to do that by talking about how I arrived at Sunlight.
I first became aware of the Sunlight Foundation while working as a programmer at a consultancy here in DC, building sites for large nonprofits and dabbling with using and writing about various technologies on the side. When I heard about Sunlight Labs, I thought it was pretty much the coolest thing in the world. Technologists using their skills to directly improve society. For people like me (and probably you) — people who have acquired a technical skillset that’s powerful, in a sense, but not always obviously useful — it’s an incredibly compelling prospect.
At the time I didn’t think I’d ever be able to hang with the folks in the labs, who seemed to be cranking out one brilliant project after another. But I did manage to convince my boss to let me make an entry for the first Apps for America on behalf of the company.
I got pretty excited about it. I’d found a physics library I could use with Processing that let me take a graph and turn its edges into springs with physical characteristics based on the edges’ weights. I scraped OpenCongress for the complete voting history of every legislator, then turned them into vectors and created similarity matrices, from which I built my graphs. I wrote perl scripts that took days to run, and gussied up the Processing applet with legislator pictures, and composed a huge exegesis on what I’d done.
It didn’t win, thank goodness. In retrospect it was awful. Don’t get me wrong: it was cool. It looked neat, it was polished, and some of the results were even defensible (big surprise: senators from Maine vote similarly). From a technical perspective I’d done a pretty nice job.
But what I didn’t do was bother to learn about the question I was trying to answer. I didn’t talk to anyone who worked on the hill; if I had, I would’ve known that I should filter out the many meaningless procedural votes in each legislator’s history. I didn’t talk to any political scientists; if I had, I could’ve benefited from the decades of work that have been done on spatial modeling of legislatures. I didn’t even talk to any computer scientists, who could’ve pointed me toward simulated annealing or other algorithmic approaches to constructing my graph that were more appropriate than the actually-quite-dumb technique I used. Instead I simply used some neural networks stuff I half-remembered from undergrad — that was the mental hammer I had on hand that made every problem look like a nail — and I started pounding out code.
Now, I’ve learned better than that since then, and my brilliant colleagues at Sunlight have been a big part of that education. But this is a bad habit that I can still fall into from time to time, and I suspect I’m not the only one. It’s very easy to become so focused on our tools that the things we’re building with them suffer.
Clay’s done a a truly amazing job of building this community, and my first priority is to keep it growing. So don’t worry: we’ll still have app contests and hackathons and blog posts geeking out over tech that we think is cool (in fact, I hope to have more of those). But if there’s one new thing I’d like to bring to the Labs community, it’s the sense that there’s no substitute for really digging into a dataset, whether that means poring through PDFs or emailing experts or even (ugh) picking up the phone. There are communities of experts who could benefit from our community’s expertise — and vice versa. I’m excited to work with all of you to make that happen.