Looking for Haiti’s Lost, Online

How information technology can streamline Web searches

As the dust was settling over Haiti, journalists were boarding planes, a response center at the State Department was manning the phones, and humanitarian organizations were swinging into action. The new media community also sprung into action, notably around the issue of connecting people in Haiti with their families abroad, and also developing emergency dispatch via SMS.

Phone systems in Haiti were partially damaged and quickly swamped, but for Haitian Americans phones were ringing constantly with calls from friends and family looking for news about loved ones. The diaspora loves Facebook, and Haitian expatriates logged on to post notices of the lost and missing. Other social media sites, bulletin boards, and blogs were soon posting photos and names, much like the walls at NYU and the Armory in the days after 911. Marvin Cherry, a recent immigrant, launched a Creole-language blog named koneksyon.com, designed to allow visitors to post information about people lost and found. Koneksyon had twenty thousand unique views in its first day.

A blog or BBS (bulletin board system) is great for chronologically ordering stories or conversations, but the serial format leaves much to desire for exhaustive searches, and two blogs are more than twice as bad. If a cousin of “Jean Deaux” posts that she is looking for news of Jean on one site, and Jean’s friend posts that he is safe on a different site, the cousin might never see it. The greater the number of sites posting lost or found information, the less chance there is that the right people can connect. According to Reed’s Law, the value of a network grows exponentially with the number of its members. Silos, while great for grain, are terrible for information. What is called for is open, interchangeable data.

This lesson had already been learned in prior disasters; in the aftermath of Hurricane Katrina there were already initiatives to standardize on data about missing persons. For instance, a People Finder Interchange Format (PFIF) standard was created, allowing for a standard means of logging, organizing, and exchanging data about lost and found people.

Within twenty-four hours of the earthquake, artist and programmer Tim Schwartz launched a web application named haitianquake.com. His team developed a set of software scripts to “scrape” information from the many blogs and bulletin boards, including the American Red Cross Family Finder, and put the data in a single PFIF store, allowing for exponentially more successful connections. By January 13th he had thousands of names, and people were successfully connecting. By the 15th, Google
was working on a similar initiative, though it lacked many of the features of haitianquake, from photos to actual data. By the 16th, Schwartz decided that Google’s reach and brand recognition would lead to greater adoption, and since the project’s success was proportionate to the number of people submitting information in a common data store, Schwartz passed his code and data to Google, and hung their “widget” on his home page. Koneksyon.com and other sites soon followed.

A week after the earthquake, as phone and SMS services came back on board in Haiti, many diaspora families and their counterparts in Haiti were able to be connect directly. This somewhat reduced the need for a people finder. But in coming weeks, as the cities are cleared for reconstruction, refugees will once again be disconnected from their loved ones. It is urgent that the Google volunteers quicken their pace, adding more photo support, an RSS update feed, and callbacks via SMS and email to let users who posted a “lost” entry know immediately if it updates to “found.” Eventually, the site must be open sourced to allow contributions from the concerned community, and run in conjunction with organizations like the Red Cross and the United Nations.

Another significant initiative undertaken by the tech community was the 4636 project. Organized by Ushahidi, InSTEDD, and others, the 4636 shortcode is a number to which any Haitian can send an SMS free of charge. Each message is piped directly to a Web interface, where Creole-speaking volunteers in the diaspora translate and categorize it. Over the past week nearly 8,000 messages have been processed by a dispersed group of translators logging in from Miami to Montreal. Requests for urgent assistance (like gasoline for a clinic, or hospitals with capacity) have been translated (on average) in less than one minute, and dispatched to the appropriate responders on the ground in ten.

While neither the people finder nor the 4636 short code are perfect, they represent real contributions to the relief effort that can – and should — be extended and perfected. Strikingly, they were built almost entirely by volunteer information activists using distributed “open source” methods of software development. The success of these methods should be noted by aid and relief communities, as well as the traditional media.

Chris Csikszentmihályi directs the MIT Media Lab's Computing Culture group, which works to create unique media technologies for cultural applications. He also directs the MIT Center for Future Civic Media, which develops new technologies and techniques to strengthen geographic communities.