Tag Archives: crowdsourcing

Crowd-sourcing in ICT4D

Dr. Stephen Ward’s lecture to our class on the importance of crowd-sourcing really opened my eyes to ways the technology could be used in ICT4D. While in the western world crowd-sourcing has been used to improve websites but also provide controversial marketing information to advertising agencies and corporations, the concept as a whole was repurposed by “DigitalGlobe”, which uses GIS and satellite imaging technologies, to launch their crowd-sourcing platform “Tomnod” in an effort to assist locating the lost Malaysian plane in early March. Developments in crowd sourcing and GIS satellite technologies opened up the information source to thousands of users who were able to provide up to date information on the whereabouts of the plane. Applying crowd-sourcing to other endeavors in ICT4D to provide up to date and accessible information to those in the developing world on their surroundings could be endlessly useful in the coming years not only in disaster prevention but also in instituting projects who’s success depends on the nature of the landscape. DigitalGlobe’s five high-resolution imaging satellites were able to capture more than 3 million square kilometers of earth imagery each day, providing an incredible volume of imagery that would have been overwhelming were it not for the “Tomnod” crowdsourcing mechanism. The efforts of millions of online volunteers around the world helped DigitalGlobe rule of some of the major oceanic areas in order to hone in on more likely locations, leading to a more efficient search process.

The Value of Crowd-Sourcing and Private Sector Data Analysis in Disaster Response

Today, Senior Geospatial scientist Steven Ward presented to the class the ways in which his company ‘DigitalGlobe‘ combines ICT, geospatial data, satellite imagery for use in a number of industries, including development. DigitalGlobe operates a number of satellites that take images of the earth’s surface and disseminates them to a number of clients, including the US government, Google, the UN, and various NGOs, among many others. An even more critical aspect of the company is the data analysis it provides, which is largely supplemented by crowdsourcing techniques. For example, scientists like Steven Ward will publicize certain images of a disaster area, such as satellite photographs taken of a mountain range in which climbers have gone missing. DigitalGlobe employees will then look at trends of information tagged on these pictures by the public, an analysis that is augmented by a number of algorithms that help to determine the degree of validity of the information they are receiving. They can then analyze the aggregate data to try and find precisely where the missing climbers set up their base camp, climbed, and eventually fell (find the story here). Though this specific case is tragic, it reveals a host of ways in which vital information can be amassed through ICT techniques such as crowdsourcing, as well as how tech-based firms can contribute their innovations and analysis in times of need.  The company is an important example of the private sector’s role in aiding humanitarian crises as well as its contributions in developing key information systems that can make or break disaster response.

Another important take-way from Ward’s lecture was simply the logic surrounding open-source data analysis, which is an ICT in itself. Ward pointed out that “more hands make light work”, which is a critical notion in time sensitive situations such as Guinea’s recent Ebola outbreak, where health care experts need as much data as possible to determine the pathways of an extremely lethal disease in a population dense area. Though some might worry that information coming from the masses is more likely to be incorrect, this is actually a misconception; Wikipedia, which is a compilation made by thousands of ‘amateurs’ has a credibility ranking of 8/10, while Encyclopedia Britannica, which is a collaboration of fewer ‘experts’, has a score of 8.8/10. The fact that these sources have such similar scores demonstrates a key point of value for crowdsourcing techniques: the more people that contribute to and review the data, the more accurate it is likely to be. Therefore crowdsourcing in itself is many times one of the most valuable approaches to mapping disaster and crises, as well as other, less time sensitive development sectors such as poverty, agribusiness land-grabbing, vulnerable agricultural lands, and thousands of other factors that may be critical to the interventions of stakeholders within the field.


Is Crowdsourcing the Next Solution?

Before this semester, I was familiar with crowdsourcing only in the context of consumer behaviour, using it to search for the best restaurants, hotels, etc. It was not until Dr. Stephen Ward spoke to our class that I realized the endless broad and diverse applications of crowdsourcing using available GIS and satellite imagery of the Earth. Dr. Ward discussed how DigitalGlobe launched their crowdsourcing platform Tomnod on March 11th in order to increase efforts to find the missing Malaysian plane. Using Tomnod, over 25,000 people have been able to scan satellite imagery and tag highly important areas, which are then run through algorithms to sift out all irrelevant information. Within a couple days, Tomnod uploaded over 1,235 square miles of high-resolution satellite imagery of the Gulf of Thailand, making me question how, even with crowdsourcing, we would be able to efficiently sort through the massive amounts of data to find the important details. Although computers use complex algorithms to determine what is noise and what is most likely relevant, I cannot help question the reliability and efficiency of this process.

According to The Stream Official Blog, some users, reported coordinates for interesting objects, such as an outline of what appeared to be a plane underwater, and oil slicks and metal/plastic debris. However, several people are skeptical about the practicality of using crowdsourcing to find the plane, as the plane probably will not resemble a plane any longer and the lack of visibility of debris due to the limited resolution of the satellite. What prevents people from tagging every rock or garbage they see? Also, how are we certain that the algorithms don’t discard any relevant information?

Over the past five years the developments in crowdsourcing has enabled it to be applied to several disciplines, such as science, international development, and security. It has been used to find missing people, determine future famines, highlight current conflict areas, and supply information that would otherwise go unknown. That being said, I fear we still lack the scientific capacity to rely as heavily as we have been on GIS and crowdsourcing. We cannot significantly reduce ground searches and ground operations until we successfully use GIS and crowdsourcing several more times. In the future, I think GIS and crowdsourcing will alter the development sector; however, we must continue to develop innovative ways to more efficiently and accurately deal with the influx of data before we rely on this method.

“Syria Tracker” Uses Crowdsourcing to Help Report Human Data

The current conflict in Syria is constantly changing and hard to track. Many Syrians have fled the country and are refugees in neighboring nations, where their injuries, illnesses, births, deaths, and other usually easily trackable statistics are getting lost in the refugee camps. According to Ushahidi (http://blog.ushahidi.com/2011/05/24/syria-tracker-crowdsourcing-crisis-information/), there’s a crowdsourcing effort and reporting crimes via a new application called Syria Tracker. Reporters can report crimes via email, twitter, or voicemail. The reports are then verified by task forces and humanitarian workers to cross-validate the information in an effort to obtain a more accurate sense of the crisis in Syria and around it. Syria Tracker has documented more than 1000 crimes since it’s inception in 2011. This use of crowdsourcing is helpful in humanitarian and legal efforts, now and in the future. 

Online Search Party: the Novice Hunt for Malaysia Airlines Flight MH370

On Thursday, our class had the distinct pleasure of hearing a guest lecture from Robert Banick, a GIS coordinator at the American Red Cross, by way of Skype. According to Robert’s twitter , his work entails “Making maps with stuff, responding to disasters and everything in between.” Needless to say, his perspective was rather interesting and presented us with a good idea of just how important Geographic Information Systems are, even though they are often overlooked.

While demand for mappers like Robert is often contingent on natural disasters, similar disciplines are being employed at this very moment in the hunt for Malaysia Airlines Flight 370. For those of you who don’t know much about the current situation, last Friday, a flight from Kuala Lumpur, Malaysia to Beijing, China vanished from radar communication somewhere over either the Indian or Pacific Oceans. It is unclear where/when/why the plane went down. However, experts are now saying that the flight most likely veered West from its usual flight-path and put the plane down in the Indian Ocean. Without the recovery of the plane’s transponder, they can’t narrow the search area by very much. Therefore, there is a lot of ocean to cover in the search and that is very time- and energy-consuming.

However, experts are encouraging civilians, with no prior experience necessary, to join in the search. This crowdsourcing approach makes use of the website Tomnod.com. Tomnod is a software run by commercial satellite company DigitalGlobe, which repositioned two of its satellites over the area when this issue came about. Tomnod users are provided with a randomly chosen map from the search area and are told to drop a pin if they see signs of wreckage, life rafts, oil slicks or anything else that looks “suspicious.” An algorithm then finds where there is overlap in tags from people who tagged the same location, and the most notable areas are shared with authorities. Many people are calling this program a sort of “online search party.” While the results are so far inconclusive, that is no reason to be discouraged, as authorities are doing no better in solving the mystery. Like Robert Banick, Tomnod played an important role in the efforts around Typhoon Haiyan as well.

According to a Tomnod spokesperson, as of Thursday, every pixel had been looked at by human eyes at least 30 times. Although nothing significant has turned up yet, this is incredible progress in the search and saves authorities from a lot of redundancy. In short, this is a very current and real-life example of just how valuable GIS is in these situations. Without satellite imagery and a collective, off-site effort, it would take search and rescue teams weeks or months to cover the area that is covered by the online community in just a few days.

Maps: More than Point A to Point B


Maps are mind-blowing. Click here to see what I mean. However, since mapping allows creators of maps to present all sorts of information in a variety of ways it’s importance to be aware of any agendas that may be operating in the development of these maps. This is one of the reasons that OpenStreetMap is such a cool premise. Anyone can edit it, so theoretically there’s no cause for concern about one overarching agenda.

This week our class has joined the OpenStreetMap community and taken up the task of mapping Chitwan, Nepal. I’ve found that watching the lines and squares appear while tracing roads and buildings is both empowering and intimidating. Contributing to the world’s largest crowdsourcing and open license project certainly has implications far beyond the walls of our classroom, but it’s easy to feel disconnected from the on-the-ground impacts of the technology.

The benefits of mapping are fairly clear in the context of humanitarian responses, but how can maps be of use in a broader development sense? OnTrack and CPD Maps are two examples where the power of maps has been successfully harnessed to target resources most effectively. After all, one significant advantage of maps is their ability to get us from point A to point B most efficiently.

Unveiling OnTrack at the 2013 Esri User Conference.

Unveiling OnTrack at the 2013 Esri User Conference.

OnTrack is a citizen feedback platform developed to facilitate communication through citizens and governments. However, this communication becomes more challenging when a lack of data on local infrastructure hampers monitoring of the status of various projects. That’s where mapping comes in. A “Mapping Party” at the 2013 Esri User Conference used OpenStreetMap’s platform to map infrastructures and identify project sites and beneficiaries, creating upwards of 640 building footprints. This allows for more effective communication between project implementers and targeted communities, and facilitates monitoring of development initiatives.

CPD map.

CPD Maps is an application that allows donors to target funding to neighborhoods with the greatest need for assistance. It works by providing data and maps that help identify census tracts with particular conditions, such as funded projects, neighborhood rent, and economic need, which allows for an overlay of areas of poverty on the maps. Moreover, donors and the public can access CPD maps from the US Department of Housing and Urban Development’s website. This allows anyone to see where federal investments are being made, information which may empower individuals to suggest future development targets.

In this way, maps may increase the accountability of government agencies and development organizations to the communities in which they work. Furthermore, crowdsourcing projects like OpenStreetMap may decrease the chance of a specific agenda shaping the data that is shared. Looks like maps aren’t static after all.

Crowdsource Volunteers are HOT HOT HOT!

The earthquake that hit Port-au-Prince, Haiti on a Tuesday afternoon in January 2010, forever changed the way that emergency responders use crowdsource mapping to provide need-based aid.

HOT volunteers writing OSM manual in Kreyòl

HOT volunteers writing OSM manual in Kreyòl

According to a U.S. News Editorial about crowdsourcing in various disaster affected communities, volunteers from all over the world began collecting data information from several sources coming out of Haiti, including SMS, Twitter, and news websites. With enough specific geographic information, these sources were used by volunteers to annotate a live map on OpenStreetMaps (OSM) to aid emergency responders on the ground in Haiti. We have been using OSM in class this week, and the sheer pace that these volunteers traced roads for 24 hours a day remotely from the disaster point was nothing short of amazing. These annotated OSM maps were vital to the success of the U.S. State Department’s SMS relief program’s short code 4636. Texting 4636,“INFO,” meant that anyone within the Digicel mobile network  in Haiti could text “I need water” or “I need medical help” and their location, and these messages were routed to aid organizations and emergency responders like Red Cross on the ground for free. The maps that the volunteers filled in on OSM were essential to NGO emergency responder’s execution of relief aid to any area requested.

The success of this collaboration spurred the formation of the Humanitarian OpenStreetMap Team (HOT). HOT workers gather base data on disaster-prone regions remotely and on the ground from available satellite imagery to improve disaster preparedness in that region. Some HOTOSM (HOT + OSM = HOTOSM) project sites include Somalia, Cote D’Ivoire, Mongolia, and Indonesia. From my nerdy interest in plate tectonics, I know that Somalia and Indonesia are their own plate boundaries, which make them prone to earthquakes and volcanoes. But after researching their disaster statistics on PreventionWeb (a detailed disaster reduction resource), I learned that more deaths occur in Somalia from floods and epidemics than from earthquakes. I can now understand how the unique disaster-development challenges in each region motivates volunteers to negotiate access to imagery and trace roads for hours on end, like we are doing in Nepal and like HOT volunteers doing in Somalia. Just for our own motivation for the our HOTOSM project, I researched the disaster statistics in Nepal. The most common disasters that affect and kill people are storms and floods. But wildfires bare most of the economic burden to Nepalese development.