Saturday, October 1, 2005

Personal history of the Katrina PeopleFinder Project PART I

A personal history of the Katrina PeopleFinder Project PART I

The term “social source” is something I have used for half a decade to describe collision of nonprofit technology and the open source movement. It tries to capture the idea that technology can be harnessed for a social mission by employing community development, online community, and web 2.0 strategies.

The PeopleFinder project started for me with a fairly simple social mission. Some folks wanted to put up a website that included CiviCRM, the open source nonprofit constituent relationship management system being developed by a bunch of us at the Social Source Foundation.

Being a good cause and a good test of our technology, we agreed to help install CiviCRM in Its use would be to power a “peoplefinder” feature, like so many others on the web, to help connect evacuees with one another.

And then I thought about it for a second.
(1) Why build yet another small scale solution to a large-scale problem.
(2) We built CiviCRM to solve major, large-scale nonprofit effectiveness issues related to constituent relationship management.
(3) Open and distributed systems can scale to provide real solutions to national problems.

As this stuff was formulating in my head, I drew up the initial fields for the peoplefinder thinking that we could aggregate all the evacuee sites on the web. Never once did I think there could be a single, “master” database of evacuees. Instead I thought about ways all the evacuee sites could “talk” to one another.

So we needed a data standard with the right fields so that all these bulletin boards and online databases could interoperate. I’m not sure at what point I decided this was going to be a national solution to the problem rather than a small community based web site, but on September 1st, I observed in an email, “Seems like we could bang something usable out in a couple days, get volunteers to do data entry from discussion boards, etc. and have a pretty useful refugee matching solution.”

So then I went out to people I knew and started enlisting help. Andy Carvin, Marty Kearns and Deborah Elizabeth Finn got the first email. Kieran Lal and Zack Rosen from CivicSpace Labs were already involved and they brought in Steve Wright from the Salesforce Foundation.

And this kind of became the ethos of the PeopleFinder project. Send an email out about what needs to get done. People respond to that email and take charge of getting things done. Magically, a solution appears and you’re not quite sure what exactly happened, but you’re trilled that there is now a solution. And you move on to the next thing.

By the 2nd we had a comprehensive list of missing persons sites tagged in If we wanted to aggregate the bulletin boards and databases, we would need a dynamic, living and scalable list. did the trick. Note the first use of an open technology…didn’t have to buy it, could just use it to do good. This is an important theme, the technology has to be pre-positioned, accessible, and you can’t need to “ask permission” or even involve the folks that “own”/maintain the technology to use it for your purposes.

About this time, Foundation committed to providing the back end database and search engine. My motives in engaging Salesforce were twofold. First, they are good guys, committed to open standards…if only they were open source ☺ Second, I felt it important to get a big corporate player involved in the hopes that they could move resources latter on in the process, though their technology is pretty cool too. (Gotta remember, I’m part of the team building the open source nonprofit CRM—I think the nonprofit sector needs a class solution that meets their needs and is open source :).

The Foundation and the folks in the company have good hearts, do good work, and we were blessed to have them on the team.

Also on the 2nd we put up a mailing list, katrinadev, because it’s the Internet and you can’t do a project without a mailing list, and recruited folks and did countless other tasks.

On the 2nd we also made a critical technology decision… use a distributed technology like RSS to solve the problem of 20 different evacuee databases. Rather than force everyone to go to a central database, lets make EVERY database central by syndicating evacuee data. At this point the Godfather of the PeopleFinder Interchange Format (PFIF), Ka-Ping Yee, rode in on his white horse. I’m pretty sure Zack Rosen roped him into this, but I don’t actually know for sure how Ping got involved.

Actually, I thought the spec took 24 hours longer to write than it should have, but I have no technology skills, and, as it turned out, should have just trusted Ping and Jon Plax to do a good job… cause they did a stellar one.

I’m not even sure what we actually did the 2nd… Andy Carvin was great at helping flesh out the idea and introducing us to bloggers and others that could spread the call for volunteers. Other folks “spread the meme” and people kept popping out of the woodwork to do stuff.

I think it was the 2nd that Ethan Zuckerman and Jon Lebkowsky were introduced to PeopleFinder—they become critical to the story latter on.

On the 3rd, the team outlined a project plan and lined up internal resources. Kellan Elliott-McCrea connected us to some guys from Craigslist who were facing the problem of being a repository of missing persons anLinkd saw the benefits of a central database (they eventually coded part of the system we used to parse Craigslist into bite sized chunks for data entry volunteers).

We needed a website, a place for a community to self organize. I’ve been a member of the Omidyar Network ( it started, and have always thought they could be much more than they are. They exist so that more and more people discover their own power to make good things happen. Seemed like a good fit at the time.

On the 3rd Jon Lebkowsky came into the mix from Omidyar (I think). I’ve chatted with Jon a couple times, knew he was a good guy, and basically got out of his way. At some point Jon and Ethan Zuckerman from the Berkman center at Harvard became the point people on data entry. I literally have no idea how the code got written to enable volunteers to do data entry, how the training materials for volunteers got developed, or how that whole side of things happened. I just know there are lots of amazing people that came together and made it work.

Took us about 3 hours to outgrow Omidyar’s interface and move over to, a wikipedia site much better suited to the type of self organizing we were doing. Looking back at my email, I think it was Jon Lebkowsky that introduced me to Rudi Cilibrasi, the guy “in charge” of katrinahelp. I just remember trying to connect with Rudi on Skype and having the technology just not work. I ended up calling him (he lives in Europe) and having a 15 minute conversation that just lead us to trust one another… our goals and values were in alignment.

Again Rudi was providing open technology—a wikipedia site. We didn’t need his “permission” to start using the technology (though of course we got it first because that was polite). Even though we didn’t technically need his help to use the technology, he was an amazing resource because he understood deeply how his technology worked and could help others in the community use it to solve problems.

The relationship between PeopleFinder coordinating organizations (Social Source Foundation, CivicSpace Labs, and the Foundation) and is part of what I call Web 2.0 Collaboration. People, technology and organizations whose default position is trust…whose first question is “how can others leverage what I’m doing” rather than “how can I protect myself from other leveraging what I am doing.”

This whole experience underscores the absurdity of building insular communities that “discover their own power to make good things happen.” Communities exist all over the world and in cyberspace and just need a little infrastructure to catapult them into highly effective entities. That infrastructure of communication and simple directories of what is available needs to be distributed rather than centralized.

Around 3pm on the 3rd we started data entry and started distributing a plea:

At 11:30 pm on the 3rd, I figured it might be good to actually write down an overview of what we had been doing for the past couple days.

By the 4th, it was pretty clear that we had expanded past the point of being coordinated. So we tried to get some folks to “officially” lead sections of the effort. That effort fell flat on its face mostly because their were people already leading the effort… they were was to busy doing things to have time to list themselves as a leader.

Around this time my role became “human router” I would look at the email stream which was getting absurdly large, and simply connect people with one another. Hey person A, talk to Person B before you do thing C.

The fourth was about details… getting the HTML data entry form from good enough to good, getting the PFIF documentation to a place that it was really useful for developers. Lots and lots of details…

By 3AM on the 5th, we had 10,000 records entered into the database and the volunteer effort was snowballing.

More to come..

No comments: