The damage caused by tsunamis, hurricanes and earthquakes can be devastating, crippling cities, ruining homes and businesses and tearing families apart. Traditionally, much of the damage has been caused by insufficient warning before a disaster and a lack of efficient communication after it has passed, issues that continue to be major factors in coordinating relief efforts today.
Recently, things have begun to change though. The proliferation of smartphones across the world has made it possible to start engineering a new approach to disaster relief, led by the companies that people already trust and understand. Whereas just a few years ago FM radio was the best place to learn of news, now it is online on a smartphone. Not only can up-to-the-minute information be obtained at any time but the user can also call or text friends and family to let them know they are safe. Emergency services can be told the exact location to rescue missing people from, regardless of whether phone and power lines have been toppled by wind. It’s a powerful capability and one that social networks like Facebook have begun to realise.
Facebook Safety Check
Last year, the company announced the launch of Safety Check, a feature that is designed to make it easier for people to see how many of their friends are in an area affected by a disaster. People in the area can mark themselves as safe once their position has been secured, curbing the worry of friends and family.
Facebook said at Safety Check’s launch:
“In times of disaster or crisis, people turn to Facebook to check on loved ones and get updates. It is in these moments that communication is most critical both for people in the affected areas and for their friends and families anxious for news. We want to provide a helpful tool that people can use when major disasters strike, so we’ve created Safety Check – a simple and easy way to say you’re safe and check on others.”
Safety Check evolved from an earlier project known as Disaster Message Board. It was created by Facebook’s team in Japan in the wake of the devastating 2011 tsunami that led to the evacuation of over 400,000 people. The tool was designed to make it easier to communicate with others in the ensuing chaos and was so successful that its creators were “personally inspired” to continue working on the concept and roll it out in a refined form to every Facebook user.
The feature has been activated several times in the past year. During the Nepal earthquakes of April and May, 8.5 million people used Safety Check to inform friends and family they were safe, leading to 150 million users getting reassuring notifications.
Last month, 3.2 million people used the feature during the Chilean earthquake and 100 million got notifications. 164 million people were notified that a friend was safe during the recent 7.5-magnitude earthquake in Afghanistan. This week, Facebook again activated Safety Check in the aftermath of the shootings and suicide bombings in Paris, the first time that the tool has been used during an act of terrorism.
Mark Zuckerberg explains why Facebook activated its Safety Check feature for Paris but not Beirut @VentureBeat
— Patrali Bose Paul (@tvsiliconvalley) November 16, 2015
Facebook describes Safety Check as a “simple and easy” way for people in disaster regions to say they are safe and check on others. The company said to Digital Journal:
“The importance of connecting is heightened during times of crisis. After many natural disasters we saw many people using Facebook to tell friends they were OK, so we built Safety Check as a simple and easy way to say you’re safe and check on others. We activated Safety Check several times in the last year, for people affected by the recent earthquakes in Afghanistan and Chile, the two 7+ magnitude Nepal quakes, Tropical Cyclone Pam in the South Pacific and Typhoon Ruby in the Philippines. This is another example of the importance of global connectivity.”
The work of Silicon Valley
It’s not just Facebook that’s actively trying to improve disaster relief. In fact, most of the Silicon Valley giants are having a go, although some have been looking into how they can help for longer than others. Microsoft has provided support in the aftermath of over 300 natural disasters. In 2013, it helped with the recovery from Typhoon Haiyan in the Philippines by deploying special radios that provide Internet access. It also enabled communications between 5,000 humanitarian aid workers by providing connectivity kits powered by a special version of Skype. The total value of the technology was over $1.1 million and it proved invaluable in coordinating the movements of aid workers.
Earlier this year, the company provided over $13 million worth of cash, technology and in-kind support to help communities affected by the two major earthquakes in Nepal. To let people communicate with each other, Microsoft offered free Skype calls in and out of the country. At the company’s Innovation Center in Kathmandu, partners were trained on how to use the software to make calls and connect with family and aid workers. The students were then sent back out into the communities where they used Windows Phones to help people establish contact with relief providers.
Microsoft says this work was made possible by the vast amounts of real-time data now available online. The company cites sources including social media, impact and resource reports and drones as being the key providers of this important information that shapes how disaster relief is provided.
Harmony Mabrey, senior operations manager for Microsoft Disaster Response, said to Digital Journal:
“Technology has been changing how organizations, communities, and governments respond to emergencies, allowing them to more easily coordinate and deliver aid where it’s needed most. At Microsoft, we see that one of the biggest changes and opportunities made possible through technology in recent years is the availability of more real-time information and data. With this influx of data, machine learning, cloud computing, and data modelling will be critical to making information consumable and usable for making agile and informed operational decisions.”
Google – finding people with drones
Google’s work dates back to 2005 when its Google Maps and Google Earth teams helped provide vital data used to coordinate aid for gulf coast communities destroyed by Hurricane Katrina. Since then, the company has responded to over 30 more disasters including Hurricane Sandy, the 2012 U.S. wildfires, the 2011 earthquake in eastern Turkey, Hurricane Irene and BP’s Deepwater Horizon Oil Spill in July 2010. In the same year, the company worked with the U.S. Department of State to launch Google Person Finder, a tool that aims to help people “reconnect with friends and loved ones in the aftermath of natural and humanitarian disasters” in a similar manner to Facebook Safety Check.
In the future, Google intends to use drones to fly relief packages into disaster zones. The unmanned, camera-equipped aerial systems have already found a place in the arsenal of emergency responders, helping to map the damage to infrastructure if roads or other transport links are destroyed. After the 7.8-magnitude earthquake in Nepal this year, it was footage from drones that first revealed the extent of the damage to the country, showing in high-definition the crumbling buildings and cracked ground. Drones will soon be able to map locations in 3D as well as 2D, sending home data that operators could use to create models of damaged buildings and assess their risk of collapse.
Cities themselves are also going to be getting smarter. This week, Los Angeles announced it will be the first city in the world to upgrade its streetlamps to Philips SmartPoles. It’s a lot more interesting than it sounds as the next-generation LED lamps also include cell towers capable of providing 4G connections to smartphones. 100 SmartPoles will be installed over the next year and 500 more added by 2020.
In the event of an earthquake or other natural disaster, mobile towers are easily overwhelmed by hundreds of people all rushing to call friends at once, an issue avoided by using multiple smaller towers. If a city’s cell phone coverage is provided by just a few centralized towers then large swathes of the population will lose signal if one is destroyed. When the towers are in every street light, it’s a lot harder to disrupt mobile coverage. The citizens will be able to enjoy better mobile signal every day but will also benefit from more reliant coverage should disaster strike.
The use of technology in relief operations really began to gain traction after the earthquake in Haiti in 2010. The country’s communications were destroyed and the residents left reliant on mobile phones and services like Twitter for news as international aid arrived. At the peak of the Japanese earthquake, 300,000 tweets referencing the disaster were posted every minute. It was a similar story during Hurricane Sandy which ultimately fuelled over 20 million individual tweets. Each one provides an insight into the public’s reaction and can inform the authorities of up-to-the-minute on-the-ground conditions when photos or video are attached.
As far back as 2004, people were using mobile phones to capture and share footage of disasters. After the Christmas tsunami in Sri Lanka on December 26 2004, news stations worldwide opened their reports with shaky and blurred amateur video of the wave hitting the beaches, filmed on camera-phones by the people who had their homes destroyed.
There are issues though. With the general public uploading short segments of localised news to a multitude of different platforms, it’s easy to get confused by conflicting information. When the victims of disaster are likely to be highly distressed and panicky, the accuracy of details posted online can’t be guaranteed and anything incorrect will spread as quickly as the truth. This happened during the Boston Marathon bombings when over a quarter of Americans turned to social media for information. The flood of tweets and Facebook posts made it difficult for responders to work out what was going on as self-styled witnesses-turned-news-reporters began to incorrectly identify potential suspects.
Projects such as Tweedr seek to change this. The tool mines Twitter to build a pool of actually useful information in the wake of a disaster and has been in use since 2006. It uses machine learning to extract meaningful information from the vast reserves of noisy social media data created in the hours after a disaster, although there is still room for improvement. A paper about the technology explains that the low frequency of relevant tweets means it can take a while for accurate results to be generated, especially when it needs to resolve conflicting tweets such as one saying a person is missing and another saying they are deceased.
There are other projects seeking to find order in the chaos on Twitter that follows chaos in the world. In 2013, the Crisis Computing Team at QCRI unveiled the Twitter Dashboard for Disaster Response. The aim is to automatically identify and classify tweets based on a number of different factors, enabling responders to see at-a-glance where their assistance is needed most. Informative content by eyewitnesses who are reporting casualties or missing people is prioritised above personal messages with little value to relief workers, for example.
The sorted tweets are displayed in one highly-visual dashboard created by Patrick Meier, director of social innovation at the Qatar Computing Research Institute. The system has since evolved into AIDR, the Artificial Intelligence for Disaster Response. It lets any Twitter user build collections of tweets based on keywords and geographic location, giving everybody their own Twitter Dashboard for Disaster Response.
Tweets can be classified into different topics in the same way as a relief worker would operate. For example, a user can teach AIDR to automatically filter tweets into categories such as “infrastructure damage”, “casualties” or “missing people.”
The public can also teach AIDR’s public face. Users are displayed a tweet and given a series of options for what it references. AIDR learns from the selected response, once it has been confirmed by at least three different volunteer users. The tool is flexible and designed to work with all kinds of disaster. In an April 2014 blog post, Meier demoed it being used for both the Chilean earthquake of that year and the disappearance of Malaysia Airlines’ flight MH370.
Other efforts to make sense of the vast data reserves that could be tapped from Twitter are based around pushing users towards adopting a standardized method of writing their tweets. The United Nations has supported the development of a set of agreed “hashtag standards” for use in emergency situations. The model is based around each disaster being assigned three tags, one that clearly indicates its name, another for use by the general public when reporting and a third one for use by emergency responders.
There are issues with this approach though. Firstly, it doesn’t come close to solving the problem of false information being posted and it could also lead to confusion if people forget the standard form or start using incorrect tags. When intelligent tools are already being built to make sense of the tweets users naturally write, forcing everyone into a defined method of writing seems like a step backwards.
Additionally, the public would have to be educated on which tags to use and when a standardized tag would be appropriate, adding complexity that technology is already working to mitigate. Twitter’s own “trending” and “popular tags” features allow people to agree on how to tag each incident anyway, negating the need for standardization.
The future of disaster relief
In the future, technology is likely to play an even greater role in the efficient delivery of disaster relief. Being able to see on-the-ground conditions via live reporting from the people trapped in their homes or on the streets gives response coordinators access to an all-new perspective when deciding what aid to send where. There’s undoubtedly still work to be done to make sense of the hundreds of thousands of tweets that are posted in the hours but tools are already being built to solve this issue. With more developers being devoted to disaster relief than ever before, we can expect responses to become increasingly efficient over the next years.
This shift will be led by machine learning, autonomous technologies and the natural human tendency to share information with others, blending human reporting with the immense power of computer-powered data analysis. Each new disaster brings new insight into how the victims of the next one could best be served. If efficiently harnessed, this insight can then be used to teach computers how to make sense of the millions of snippets of data created by the distressed pushed into a life-threatening situation.