Change Assembly Change Assembly solves complex human problems through data, technology and insight. Tue, 15 Jan 2013 19:41:12 +0000 en-US hourly 1 Crowdsourcing and Crisismapping Fri, 16 Nov 2012 20:43:29 +0000 Sara Farmer Hurricane Sandy Crisis Map

Google Crisis Response map for Hurricane Sandy

A crowdsourcing system parcels out small pieces of work to lots of people.  Crowdsourcing is usually performed online, and its tasks are usually connected to data or information gathering and processing. Crisismapping is similar to crowdsourcing in that it involves a team of people distributed around the world who are each doing part of a larger task, but it is usually specifically focused on disasters and human development. Crisismapping can include data gathering and mapping, and complex, specialist tasks like situation analysis and imagery interpretation. It usually occurs over short (1-2 week) periods of time as a focused, well-organised set of tasks called a “deployment”.

Many crisismappers are experts who are volunteering their time to help people caught up in disasters. But it’s not all specialist work; some tasks don’t take a lot of expertise (although the people doing them usually learn skills quickly) but can be very very powerful if done by a large number of people at once.

Typical Crisismapping Tasks

Crisismapping is over 10 years old, but for many ‘mappers’, crisismapping started with the 2010 earthquake in Haiti, where the government and NGO information needed to respond to the earthquake was lost in the earthquake. Mappers created online maps from satellite images, put SMS and media reports onto an Ushahidi map, added documents to the Sahana coordination tool, added missing people’s details into Google’s Person Finder tool, built websites and software to help, and did dozens of other small data jobs that made a difference to what was available online and in the country (mappers even sent data updates into the country on USB sticks). Since then, crisismappers have worked remotely and on the ground in places like Libya (refugees), Japan (tsunami), Chile (earthquake), Pakistan (floods), Somalia (refugees), Alabama (tornadoes), New York (hurricane) and been ready to help with dozens of other crises around the world.

Mappers are well known for their work turning social media feeds and messages into online maps and situation reports, typically for large complex emergencies where the volume of information would otherwise overwhelm the responders on the ground.  But their work includes non-emergency “prototype” deployments and simulations, exploring new technologies and protocols, e.g. recent SBTF prototype deployments include testing the new Uchaguzi platform (with iHub Nairobi for the recent Kenyan elections), automating crisis indicator gathering (with ACAPS), mapping health facilities in Libya (with WHO), and improving GIS references in aid spending data (with USAID).

Mappers also have a lot of experience in areas like satellite imagery analysis. For example OpenStreetMap creates digital maps using data provided under the Disaster Charter and marked Hurricane Sandy damage in aerial images of New York and New Jersey, and Standby Task Force (SBTF) volunteers tagged buildings in satellite images of Somalia’s Afgooye Corridor (to help UNHCR estimate how many people they needed to serve). There are also mappers with ground experience: MapAction creates maps in the field, Humanitarian OpenStreetMap teaches crisis-prone populations to map their own neighbourhoods, and mappers provided on-the-ground support to local organisations both during and after Hurricane Sandy. Supporting local response is an emerging pattern in crisismapping: it makes mapping more sustainable because it’s valuable to the people doing it, and local people already have local knowledge about and connections within an affected area.

Getting Started

If you’re considering crowdsourcing or crisismapping, especially crowdsourcing specialist data and information tasks, there are two mains ways to go: you can either create your own deployments or you can work with existing and experienced VTCs (Volunteer Technical Communities).  Whichever way you choose to go, there is a great deal to be learned from signing up to the Crisismappers Network email list, watching the presentations and conversations on it about existing crisismapping deployments and initiatives, and helping with a deployment if you can.

If you create your own deployment, you’ll need to understand the types and formats of requests for help that volunteers need, the types of tasks that they’re capable of, the things that motivate them to help, the tools available etc. Alternatively, you could work with an existing VTC.  VTCs are volunteer technical groups formed around specialist data tasks.  They include the Standby Task Force (crisismapping at the request of response agencies), GIScorps (graphical information systems), DataKind (big data and data visualization), Statistics without Borders (statistical analysis), MapAction (creating maps in the field), Humanitarian Open Street Map (creating maps with affected communities), CrisisCommons (crisismapping in countries with US-style emergency agencies) and Geeks Without Bounds (creating crisis-related software).  There are also university-based groups like the New Media Task Force and hackathon-based groups like Random Hacks of Kindness.

Many VTCs are part of a larger organisation called the Digital Humanitarian Network (DHN). The DHN exists to connect traditional response organisations like field-based NGOs and the UN, with specialist digital volunteer groups.  The DHN keeps responders and volunteer groups from being overwhelmed with connections:  it accepts activation requests from response organisations, and matches these to the most appropriate VTCs. This gives responders access not only to mapping volunteers, but also to specialist skills like big data and technology prototyping.  It also gives access to experienced groups with strong verification systems and analysis teams, built up over a long time and lots of deployments.

Working with VTCs

Organisations that have worked with crisismappers usually report back positively: they gain valuable information, they have access to expertise that they wouldn’t normally have, their staff learn more and often improve their own workflows and toolsets.  But “crowdsourcing is free as in kittens are free” (Chris Fabian, UNICEF): VTCs, for example, will expect an activating organization to work with them to create a good plan for the deployment, and be available to answer specialist queries and help guide the work.  A VTC might also reject a request for help if it could be better fulfilled through another means, e.g. an existing project, a crowdsourcing site like Crowdmap, Crowdflower or Mechanical Turk, or by training communities on the ground.

Other considerations include safety, both of the volunteers and the people on the ground that they work with or gather information from. For a prototype deployment, design is important too: e.g. there’s rarely a purely-human or purely-technology solution, but sometimes there is a solution that requires less human effort, e.g. for Somalia, JRC Europe ran a set of algorithms that produced a similar set of results to the crisismappers, and could be applied to similar situations (finding buildings in satellite images).  Verification is important too: results should be triangulated against ground-sourced data if possible. And lastly, respect: volunteers have day jobs and lives, and are giving their own time to help, so they should always be thanked for this (where “thank you” ranges from an email saying thank you to a certificate that links them to the deployment).

]]> 0
Welcome to Change Assembly! Mon, 08 Oct 2012 14:58:47 +0000 Chris van der Walt Sara Farmer and I started Change Assembly in July 2012, fresh from working together in the United Nations at the Global Pulse project. While there, we established a good working rapport and realized that we shared a common vision about the challenges and opportunities in humanitarian response and international development in the Information Age. We also wanted to continue development of Hunchworks and had come to the realization, along with the Global Pulse team, that while the UN Secretariat does many things, agile software development is not one of them. We toyed with the idea of starting a non-profit, however we decided that a commercial entity would ultimately be more sustainable and would allow us the creative freedom that we longed for after working at the UN.

The last few months have been very busy. As anyone who has taken the fateful step of starting a company can tell you, the experience is at once exhilarating and exhausting. I feel we are finally coming out of the “what’s an E3/W4/I9?” phase of business development and are able to focus on getting down to the real work.

Not that we have been idle in the meantime. Here’s a taste of what we have been working on:


We kicked off the current round of Hunchworks development with a design review workshop at the beginning of July focusing on the prototype feature set and the awesome P.J. Onori flew out from San Francisco to participate. We want to broaden the applicability of Hunchworks, so we decided to start over with a clean new codebase and interface. Several interesting new features emerged from the session including a social data manager (more on this and other features in the coming weeks).

At this stage, we are about 12 weeks away from beta and we have been engaging with expert communities to help us iron out the kinks and suggest improvements. We would love to have more beta testers, so if you are eager to get your hands on Hunchworks, please sign up.

Humanitarian Data

Sara has been getting her hands dirty with humanitarian data over the past few months. Initially she focused on the challenge that many were having accessing in the absence of an API. Once she was able to work out the requisite steps to access the data, she also helped to clean and organize the data to make it easier for it to be aggregated and analyzed for broader use among humanitarian and development groups and organisations.

Sara then went to work analyzing the provenance of humanitarian crisis indicators. A lot of crisismapping comes down to persistence, capability and trust. We need to trust the people we’re working with, the systems we’re using (within sensible limits), and we need to (again within sensible limits) trust the data we’re using. Trust in the data that happens during a crisis has been talked about a lot recently. So Sara investigated the provenance of the data that goes into various humanitarian indicators published by international organizations. She found that much of the data is, well, dodgy. She then put the theory into practice joining an ACAPS deployment.

Sara has been writing about her findings on her blog and she will be publishing more here in the coming weeks.

Upcoming work

We also have a few other projects in the pipeline including a community threat mapping project in Africa, a design workshop with a UN agency, an African collaborative journalism initiative and a large-scale university deployment of Hunchworks. However all of these are still in development, so that’s about all I can say about them for the time being.

Stay tuned…


]]> 0