How does crowdsourcing coincide with humanitarian efforts and how can we step up as ‘Digital Jedis’ or digital first responders? Patrick Meier, an influential thought-leader on humanitarian tech and innovation, weighs in
Typhoon Yolanda was one of the most powerful Typhoons in recorded human history. Known internationally as Haiyan, the Typhoon gained superstorm status as it raced westward on a direct collision course with the Philippines.
The impact was devastating. And like other major disasters these days, Haiyan left an unmistakable mark on social media activity. Millions of tweets and Facebook updates were generated, along with tens of thousands of Instagram pictures and YouTube videos. Making sense of this flashflood of information, “Big Data”, is proving an impossible challenge for traditional humanitarian organisations, which is precisely why they’re turning to Digital Humanitarians.
Who exactly are these Digital Humanitarians? They’re you, me, all of us.
Digital Humanitarians are volunteers and professionals from the world over and from all walks of life. What do they share in common? The desire to make a difference, and they do by rapidly mobilising online in collaboration with international humanitarian organisations. In virtually real-time, they make sense of vast volumes of social media, text messages and imagery captured from satellites and UAVs to support relief efforts worldwide.
How? They craft and leverage ingenious crowdsourcing solutions with trail-blazing insights from artificial intelligence. Digital Humanitarians are the veritable Jedis of disaster response.
Take Typhoon Yolanda, for example. The United Nations activated the Digital Humanitarian Network (DHN) just hours before the superstorm made landfall. The DHN serves as the official interface between established humanitarian organizations and tens of thousands of Digital Jedis. The latter provide relief workers with the “surge capacity” they need to make sense of all the information generated during major disasters.
The digital response to Yolanda was no different. At the request of the United Nations Office for the Coordination of Humanitarian Affairs (OCHA), Digital Jedis deployed online with MicroMappers, a crowdsourcing platform designed to quickly analyse tweets and pictures posted to Twitter.
OCHA needed to know about any tweets relaying urgent messages, along with tweets referring to infrastructure damage and population displacement. They also needed to identify which pictures posted on Twitter showed disaster damage, and where in the Philippines these pictures had been taken. Note that some two million tweets were posted in the immediate aftermath of Typhoon Yolanda.
While MicroMappers was still being developed by QCRI at the time, Digital Jedis were able to use the platform to generate up-to-date crisis maps that provided the UN with the information they had requested to augment their situational awareness. In the meantime, other Digital Jedis use the Humanitarian OpenStreetMap crowdsourcing platform to rapidly analyse both satellite and aerial imagery of disaster affected areas.
Digital Humanitarians learned some important lessons during their response to Yolanda.
First, Digital Jedis from local communities are indispensable to digital relief efforts. These Jedis have the local knowledge, agency and contacts that others from across the planet simply don’t. Second, crowdsourcing alone does not scale. Third, the data filtered by Digital Jedis are too disconnected and thus need to be integrated.
I’d like to expand on all three lessons below.
First responders are not the official relief organisations or the government for that matter. First responders are by definition, the local communities. This has always been the case and will always be the case. Humanitarian organisations cannot be everywhere at the same time, but the crowd is always there; thus the crowd is increasingly the source of “Big Data” during disasters. To this end, local Digital Jedis will increasingly serve as first digital responders.
This digitally enabled self-organisation stands to increase the resilience of local communities across Asia and the world, which is a positive development for disaster risk reduction and rapid response. This new trend is potentially important for democracy as well. We have already witnessed that this type of collective-action makes regimes in China, Iran and elsewhere nervous. Indeed, there is reason to believe that crowdsourced disaster response improves civil disobedience.
Second, crowdsourcing alone is no match for “Big Data”. Millions more people, not fewer, are using mobile phones every day for the first time. So what happens when the next disaster generates 20 million tweets in just a matter of days? Alas, this already happened more than two years ago (Hurricane Sandy). How does one crowdsource the analysis of 20 million tweets to find those few tweets humanitarians need to improve their relief efforts?
The challenge is akin to looking for the proverbial needle in giant, growing haystack of information. It would take one person about 60,000 hours (or six years) to read each tweet posted during Hurricane Sandy. Of course, if you had 60,000 volunteers (which is certainly plausible as witnessed during the digital search and rescue efforts for Malaysia Flight 370), then it would only take an hour to search through those 20 million tweets. But is 60,000 man hours really the best use of human time if computers can do most of this filtering?
This is where QCRI’s AIDR platform comes in, which stands for Artificial Intelligence for Disaster Response. AIDR learns from crowdsourcing. For example, as Digital Jedis used MicroMappers to tag (classify) tweets in response to Typhoon Ruby in the Philippines last year, the AIDR platform was able to learn in real-time from the crowdsourced tagging and subsequently tag future tweets automatically. This hybrid crowdsourcing-AI approach is bound to become more prevalent in the future, and not just for tweets and text messages but Instagram pictures, satellite imagery and aerial imagery as well.
Data integration was the third lesson learned by Digital Jedis following Typhoon Yolanda. This simply means that the results of crowdsourcing the analysis of tweets, pictures, satellite imagery, etc., need to be integrated within one map. Each layer serves to enrich the others. Failing to combine these filtered data streams prevents the contextualization of information during disaster response. This explains why MicroMappers is being extended to also crowdsource the analysis of mainstream media as well as satellite and aerial imagery.
Digital Jedis are already playing an integral role in relief efforts across Asia, but they can’t do this alone. Next generation humanitarian technologies like MicroMappers and AIDR may allow them to win future battles with Big Data. That being said, the underlying scientific methodologies that power these prototypes are ultimately more important than their fancy brand names.
Brands will come and go, but human and machine computing will be around for far longer. So it is time for Digital Jedis in Asia to leverage advanced computing and craft even more sophisticated solutions to Big Data to better support response efforts locally, regionally and internationally.
Patrick Meier is one among scores of top thoughts leaders and practitioners who will help you understand and navigate through the many applications of crowdsourcing whether in business, smart city governance, finance or social innovation. Join us for Patrick’s talk at Crowdsourcing Week Global 2015 coming to Singapore April 20-24.
The views expressed here are of the author, and e27 may not necessarily subscribe to them. e27 invites members from Asia’s tech industry and startup community to share their honest opinions and expert knowledge with our readers. If you are interested to share your point of view, please send us an email to writers[at]e27[dot]co