On a sunny Tuesday afternoon, January 12, 2010, just before five o’clock, a 7.0 magnitude earthquake hit Port-au-Prince, Haiti. The quake destroyed more than 280,000 buildings in the metropolitan area, killed perhaps 230,000 people within the first 72 hours, injured more than 300,000, and displaced more than 1 million. Some smaller towns to the southwest, particularly the coastal town of Leogane, were estimated to have lost 90 percent of the built structures within the town limits in the first quake and another 5 percent in subsequent aftershocks.
Many members of the Haitian government died in the quake, as did important members of sectors as diverse as media, diplomacy, aid organizations, construction, education, air traffic control, law enforcement, transportation, communications, food provisioning, port services, medical care, water purification, and everything else. Haiti had found its Ground Zero.
The international community’s response to such a disaster has been finely tuned over the years. Within 15 minutes of the quake, the Global Disaster Alert and Coordination System (GDACS) and the U.S. Geological Survey each sent Red earthquake alerts to their subscribers’ cell phones. An hour later the United Nations’ Virtual On-Site Operations and Coordination Center (V-OSOCC) in Geneva began receiving urban search and rescue deployment and coordination reports, as emergency operations centers in international hubs like New York, Rome, Copenhagen, Washington DC, Singapore, and Dubai revved up for the response. By international charter and common sense, the United Nations, if invited to help by an affected nation, is in charge of an international disaster response. Among the first on the ground—bringing power generators, medical aid, shelter, communications, food, and water—will be competent and experienced teams of doctors and experts in engineering, logistics, and communications from UN agencies like UNICEF, the World Health Organization, and the World Food Program as well as UN Disaster Assessment and Coordination teams from the UN Office for the Coordination of Humanitarian Affairs. As a result of this system, more than 200 members of various international search and rescue (SAR) teams were on the Port-au-Prince airfield within 72 hours.
The next responders on the scene are often teams from other nations and, for Haiti, the Icelandic SAR team was the first outside national team on the ground. Once the scale of destruction was clear, the Haitian government quickly approved the arrival of others, including national field hospitals from Germany, Belgium, Russia, and Israel. The Israelis established the first field hospital, with patients in surgery by January 16. The U.S. response was also rapid and, by the end of the first week, the United States had a significant presence: 17 ships, 48 helicopters, 12 fixed-wing aircraft, and more than 10,000 sailors and Marines.1
A third tier of response comes from the nongovernmental organizations (NGOs), a steadily growing asset in disaster assistance. Few of these groups are a part of the United Nations’ official response plan, but the services they provide are proving vital. The humanitarian NGO that I lead, InSTEDD, was founded by Google to facilitate critical information flow in crises, and we usually work on emerging infection risks within the developing world. We also assist others in disaster response, and we were immediately asked to help in Haiti.
A few months before the Haiti earthquake, we had developed, with a branch of the Thomson Reuters Foundation (of the Reuters news agency), a mobile communication platform called the Emergency Information Service, or EIS, to supply both aid agencies and those affected by a disaster with accurate and current information using cell phones. Within 24 hours of the Haiti earthquake, the Thomson Reuters Foundation had asked us to join their team and we agreed to deploy together. By Thursday night, two days after the quake, I was on my way to Port-au-Prince along with our technology platform director, Nicolas di Tada, to set up EIS with staff from Thomson Reuters’ AlertNet division, and we arrived the following day.
We deployed to Haiti intending to try, very cautiously, a new approach to disaster management. We wanted to harness recent technological advances to help us connect with affected victims, providing them with targeted information so they could recover faster throughout the crisis. We also wanted to engage directly with those crowdsourced volunteers who were trying to help from thousands of miles away. Through previous real-world disaster responses to Hurricane Katrina and to the tsunami in Banda Aceh, and through civilian-military partnerships like Strong Angel,2 InSTEDD already had a solid list of potential partners.
Crisis Mappers Net3 had launched just three months before the earthquake and was an informal linkage of experts in geographic information systems (GIS) capable of providing constantly updated maps of the human and geographic terrain of a disaster zone using open-source technology. OpenStreetMap4 provided another set of GIS experts, and they worked closely with hundreds of volunteers to help create a critically useful map of the devastated area. When map development began on January 13, the best available street map of Port-au-Prince (an area of more than 2 million people spread over more than 230 square miles) had just a few hundred map “objects”—street intersections, parks, major industries, schools, and government buildings.5 OpenStreetMap quickly expanded this by interpreting satellite imagery from damaged Port-au-Prince neighborhoods with the help of both Haitian nationals in the country and of the diaspora in the United States.
A third group of remote volunteers was associated with Ushahidi, an NGO based in Nairobi, Kenya. Ushahidi was originally formed to give cell-phone users the ability to text message about a location and an event and have that message show up on a Web-based map. Ushahidi had used the technology to good effect during the Kenyan civil violence in early 2008, and they were ready to do the same in Haiti.
Technology as an Aid in a Failed State
Haiti is, in most ways, a failed state, and the country is profoundly dependent upon outside aid. Haiti has been desperately poor for decades, with more than 60 percent of its population living as subsistence farmers on less than $2 a day. For most of the past 40 years, Haiti has additionally suffered under oppressive dictatorships and ineffective leaders, complicated further by the reported ubiquity of drug money and violent mobs. Haiti is one of the few countries in the world to have slid backward on the delivery of social services between 1990 and 2000. Access to clean drinking water actually fell from 53 to 46 percent. Rural access to improved sanitation dropped from 19 to 16 percent. These numbers are roughly on par with Togo, Yemen, and Laos, and they are the lowest in the Western Hemisphere.6
And yet, those who have lived in Haiti (this author included, 1984–85) have admired the people and the community they have created for the strength of their social relationships, the unifying force of an eclectic and ubiquitous religion, and a cultural integrity that transcends the diaspora, allowing Haitians around the world to gather to help through an intact language and cultural understanding no matter how long they have been away. It is worth noting that, despite the international rescue effort, the citizens of Haiti themselves conducted the greatest number of rescues of people trapped under the earthquake rubble. They provided urgent first aid, and they took care of each other in makeshift camps, establishing shelters for children, pooling clean-water supplies, carrying victims to medical care on improvised stretchers, and organizing teams for digging when live entrapments were discovered.
Upon InSTEDD’s arrival in Haiti, we established a self-contained working and living site within the search and rescue base on the east runway of the airfield in Port-au-Prince. The United Nations’ SAR dispatch center was almost adjacent to the InSTEDD tent site, and relationships were established quickly. In the initial hours, InSTEDD settled bedrock issues of personal safety, reliable power, clean water, and satellite communications and began assessing first steps for our integration into the UN response.
On that first weekend there were very few resources available beyond the SAR teams. The severely damaged Haitian cell-phone service was just reappearing, initially with SMS text messaging and then with intermittent, spotty, and poor-quality voice calls. Most of the local radio stations were still down and unable to communicate with the affected population, and electricity for radios was still sparse. The formal field-assessment teams struggled in those early days to locate the hundreds of sites where Haitians were gathering in open spaces, in need of medical care, water, and food. As an example of the resource allocation problems, despite the enormous number of injured, there were medical facilities that had underused capacity and no method to redirect patients away from overloaded centers. Further, relief organizations had no way of informing local citizens where or when resources such as water or food would appear.
SMS messages presented us with the first critical challenge. Although Haiti has a very low literacy rate and only about half of all Haitians can read, there were still hundreds of SMS messages sent each day about people alive and trapped in the rubble. Those messages were making it to the SAR dispatch tent through many routes, but (a) no one in that SAR dispatch tent could read Creole and (b) the message often identified an entrapment site by local landmarks that were now destroyed (e.g., “under the cinema beside Sacre Coeur, off Delmas 33”). The international rescue teams had no reliable method for translating these ad hoc descriptions to actionable dispatch destinations within a devastated area of more than 1,500 square miles of mostly urban terrain, containing almost 2 million people.
The solution was to produce a translation service that would give the SAR teams the information they needed. SMS messages received by search teams and the media were sent to our team at Port-au-Prince, who passed them on to U.S.-based translation teams drawn from the Haitian diaspora. From several locations, first centered within the Tufts University Library, a community grew that supported Haiti staff 24 hours a day for the first week. The messages were then sent to mapping teams with OpenStreetMap and Crisis Mappers Net in Boston and San Diego, who added the data to the maps being built of Port-au-Prince. We then calculated the best-effort GPS coordinates to five decimal places, linked them to the message, and forwarded the coordinates and the translated message back to the InSTEDD team in the field so that the SAR dispatch center could send a team with precise instructions and the knowledge that there was someone trapped to pursue. The entire cycle could take as little as 30 minutes. That sequence resulted in confirmed lives saved and continued through all of the aftershocks.
As the volunteer community grew, it began to coalesce around technical response capabilities and to draw on models for rapid prototyping well known within the software development community. Weekend events in the United States and Kenya began to take place, called Crisis Camps, and there were thousands of participants internationally. Free and open-source software was created at these camps and released immediately to the disaster response community.7
Together with Ushahidi in Nairobi and another technology NGO partner, FrontlineSMS, we established an SMS shortcode, 4636, so that people in need anywhere in Haiti could text an SMS message to that number, free of charge, and the message could then be flagged and routed to the most appropriate response capability. Water needs, injuries, entrapments, dead bodies, lost family members, and found orphans are examples of tags we generated that led to a streamlining of the information flow from the affected population to the responders. Haiti’s two cell-phone networks agreed to the system (one after intervention from the White House with the parent company in Bellevue, Washington).
We then set up a series of feeds from one system to another so that Ushahidi, EIS, and other systems could take advantage of both the content of the messages and the metadata added by the cell-system carriers. Samasource, an organization focused on creating job opportunities for marginalized peoples, brought local Haitians into their system to contribute to the translation, precise geocoding, and classification of incoming messages using CrowdFlower’s distributed-task application, so that InSTEDD, EIS, Ushahidi, and the other systems could provide this information in an actionable format for the other response agencies.
Meanwhile, within a week after the earthquake, InSTEDD and AlertNet staff in Port-au-Prince engaged local Haitians to help us broadcast in Creole, on local radio, details about the new 4636 reporting system. Although the value of radio broadcasts was limited by lack of power, we received more than 6,000 text messages and more than 200 voice calls over the following week. In the subsequent three months, we received more than 90,000 text messages from the Haitian population.
The AlertNet Emergency Information Service was also up and running within 48 hours of our arrival, and the AlertNet journalism staff soon began collecting information from UN, NGO, military, and governmental response agencies for broadcast using cell-phone text messaging to the people of Haiti. Since anyone who wrote a message to any part of our system was automatically subscribed to EIS, those Haitians submitting problems to 4636 soon began receiving our free EIS text-message broadcasts. AlertNet used EIS to send out messages concerning medical facilities, water-truck routes, food drops, vaccination locations, lost children, and any other topic that main response agencies needed to broadcast. Through EIS we sent nearly 1 million text messages to our (free) subscribers. An independent evaluation done in May 2010 sampled more than 400 of the subscribers and found that more than 85 percent said they had received value from our messages. AlertNet staff remained in Haiti and continued broadcasting in Port-au-Prince until the end of urgent relief operations in May, four months later.
Disasters and Technology
While no single event like the earthquake in Haiti can be reliably given a clear causality, the scientific evidence is compelling that climate change is in progress and that even an earthquake might be related. Studies are evaluating the hypothesis that earthquakes may have a climate component, as increasing sea levels place hundreds of millions of tons of seawater higher on continental shelves around geologically unstable regions of the globe,8 and so coastal populations, in particular, may be at risk.
Such unexpected links have led to modeling and studies around best practices in disaster risk reduction. The UN International Strategy for Disaster Reduction (UNISDR) determines and implements such best practices, with lessons collected through interviews and retrospective analyses. Like the 2005 tsunami in Banda Aceh, the 2008 Cyclone Nargis in Myanmar, and the Sichuan earthquake two weeks after Nargis, the Haiti earthquake has given UNISDR and other disaster responders a glimpse of the risks inherent in an increasingly dense, increasingly connected, and increasingly coastal population as well as the mitigation strategies likely to be most effective.
2007 was a milestone year for urbanization. We humans, as a species, became more urban than rural, with more than half of the world’s people living in cities for the first time in human history.9 Unfortunately, many are recent immigrants from inland agrarian areas, migrating to coastal areas as potable water pressures, desertification, and other issues force them to make difficult choices. The migration of rural populations to coastal cities is most noticeable in China because of its sheer magnitude (144 million people in the country’s Low-Elevation Coastal Zone),10 but China’s coastal population is not the largest national percentage at risk from coastal water disasters. That distinction probably belongs to Vietnam, another country in the midst of a rural-to-urban migration, with 55 percent of its population at risk from coastal water disasters.10
The result is profound fragility, which induces a combination of dependence and complexity that means when structures fail, very few can predict the consequences. As a complex adaptive system, much of the critical infrastructure within the developed world has accreted over time and, in many cases, there is no longer any comprehensive understanding of the system or its interdependencies. When disaster strikes in wealthy countries, even beyond the obvious initial tragedy, the results can be disconcertingly unpredictable. Failure modes within more ad hoc, unmapped, nonstandard infrastructure in the developing world, in countries like Haiti, can be even more surprising.
As we saw in Haiti, we now have social-media software and collaboration tools that can facilitate community resilience, risk reduction, and response coordination in volunteer groups within affected populations anywhere in the world. Volunteers can be remote, or deep within the local effort, and still contribute effectively in ways that augment and extend the professional response. The result of successful volunteer contributions, over time, could be fewer lives lost, fewer permanent injuries, less property damage, and a more engaged donor community, able to see increases in efficiency resulting from its funding.
With a little effort, early engagement using low-bandwidth social software (like SMS messaging, InSTEDD’s GeoChat, Skype, EIS, and that employed by Ushahidi and others) for disaster risk reduction and improved response may also reduce barriers between communities and local authorities, diminishing the risk of social unrest after catastrophic events. Such tools may eventually find their greatest value when integrated beforehand into the formal response plans of the agencies that deploy regularly into disaster zones. That incorporation, however, requires a greater degree of performance reproducibility within the volunteer community than is currently found—a discussion, and a correction, already underway.
Models for that performance reproducibility exist, and we are all learning from systems in use elsewhere. For example, NASA’s Technology Readiness Level (TRL) scale,11 used for software and hardware intended for space, is a technical reliability scale, and in the United States we have a training system for the human certification process in the National Incident Management System (NIMS). Formal NIMS courses prepare graduates for work within the incident command centers of the U.S. National Response Plan and can provide a common language for use between response agencies and volunteers. There are also Community Emergency Response Team (CERT) courses, Neighborhood-Watch training, FEMA classes, Red Cross Disaster Services guides, and easy online ways for almost anyone to learn more about robust collaboration tools, social-media software, Amateur Radio licensing, and first aid. Those educational efforts can rapidly and inexpensively improve the collaboration between volunteers and professionals during a crisis and are worth the small effort they require. Such improvements in volunteer professionalism can help form a fusion between governments, international aid organizations, volunteers, and donors, and the entire community gains as a consequence.
That increased effectiveness, as we saw in Haiti, is a critical need. Disasters are going to be more frequent, more intense, and more widely distributed over the coming 20 years and beyond,12 and vulnerable populations, often far from help, will suffer the brunt of their impact. Global response agencies are, regrettably, too small and too poorly funded and staffed to meet the expected needs of populations at risk. Information rapidly shared in a disaster through social media, within affected populations and with the response community assisting them, should become one more method by which resources can be efficiently matched to needs in a crisis. We need that improved collaboration between responders and the populations they serve. It will save lives, reduce damage, and aid the formation of a calm and rapid recovery.