a weblog by Schuyler D. Erle

Wed, 13 Oct 2010

[16:56] On the Community and Practice of Crisis Mapping

Last week, I attended the 2nd International Conference on Crisis Mapping (aka ICCM 2010) in Cambridge, MA. The conference was a rare opportunity to talk with some of the people doing the most innovative work these days in information technology for humanitarian crisis response and recovery. This year's ICCM was particularly enlivened by the involvement of projects like Ushahidi and OpenStreetMap in humanitarian relief operations over the last year. These are some reflections on what I see as being the salient issues facing the Crisis Mappers community, as represented at the conference.

If this essay's length tries your patience, read Kim Stephens's much, much more concise notes on the four general themes of ICCM instead. Or read this essay first, and then read those notes.

1. Crisis Mapping: Neither "crisis", nor "mapping". Discuss.

Quinn Norton pointed out to me that few (none?) of the IT projects represented at ICCM were specifically intended for mapping humanitarian crises, as such.

Ushahidi, the best represented of those present, was designed to crowdsource operational awareness in emergency situations, to which mapping is only an auxiliary function. Sahana is designed for similar ends, with mapping as but one of many plugins to the main system.

On the other side, OpenStreetMap, for example, is a project to produce free map data, but which only recently became useful in crisis response -- although Mikel Maron and Jesse Robbins accurately foresaw this use as early as 2007. MapAction was perhaps the only "crisis mapping" org represented at the conference, and only DevelopmentSeed's MapBox project comes close to an IT system intended specifically for crisis mapping.

In fact, in spite of the breathless attention paid to crowdsourcing as a practice, the conference itself arguably was not even really about "crowdsourcing for humanitarian relief", as the overwhelming majority of the participants in the discussion sessions on Saturday and Sunday were from large agencies, NGOs, universities, and even the US military.

You might say the conference topic was "IT for humanitarian relief", but that's a rather too broad appellation, and covers all kinds of things that weren't represented there.

So how exactly do we define the community and practice of "Crisis Mapping"? Despite the lack of an adequate definition, I'm going to make repeated reference in the following to "Crisis Mapping", all for the sake of argument. Caveat emptor!

2. Everything in its right place.

The past few months have yielded heated debate about the value of the practice of "crowdsourcing", i.e. the use of loosely coordinated volunteer participants in a knowledge creation or collection task, via the 'Net, SMS, et cetera, to disaster relief proper. The intensity of this debate has been remarkable, and arguably unproductive, but wasn't much aired during the conference, probably due to the nature of the event's organization.

To reiterate briefly: Crowdsourcing technologies (e.g. Ushahidi, e.g. OSM) have the potential to dramatically improve operational awareness for humanitarian crisis responders by distributing the costs and efforts of data collection and curation across a large group of volunteers, sometimes via the Internet. This has been shown to work in practice (Mission 4636, OSM) in Haiti, and is amazing, unprecedented.

The counterargument is that crowdsourcing risksproducing huge volumes of inaccurate data, and is vulnerable to manipulation and abuse. Unlike the Emergency 911 system in the US, which carries legal penalties for false alarms, crowdsourcing techniques lack any effective way to discourage misuse or abuse. Therefore, encouraging humanitarian relief actors to depend on unreliable sources of information irresponsibly chances diverting their attention from hazardous situations where people's lives may be at stake.

In fact, both sides are right. Mission 4636 and OpenStreetMap both very clearly, and somewhat miraculously, provided information that directly enabled crisis responders in Haiti to save the lives of people in danger. The value of crowdsourcing in disaster relief has been undeniably demonstrated.

The key question is how much and what kind of value. Crowdsourced data is absolutely not without potential flaws, and needs to be taken with a grain of salt. It won't always work perfectly, but we needn't throw the proverbial baby out with the bath water. Crowdsourcing itself can never be a replacement for traditional intel in humanitarian relief operations, but it can -- and should -- be used responsibly by relief actors to build an accurate picture of the situation on the ground.

As a chap from the US NGA (really) said to me at an information management cluster meeting at the UN Logbase in Port-au-Prince in February, "Unverified information is still better than no information at all!"

3. How to make miracles happen on time.

The elephant in the room is that the considerable utility of crowdsourcing to the international relief operation in Haiti came about with almost no planning or forethought.

Both Mission 4636 and the humanitarian OpenStreetMap response were organized, to one degree or another, on an ad hoc basis in the first few days after the earthquake, relying almost exclusively on people doing their best to employ the resources they had available to the greatest effect they could manage. The OpenStreetMap effort, particularly, had effectively no leadership to speak of, and relied on a wiki page to coordinate the efforts of over 600 volunteers in the first two weeks after the disaster. In fact, haiticrisismap.org, which housed most of the imagery used by OSM editors, was mostly the work of two or three people, particularly Christopher Schmidt at MetaCarta. Nevertheless, Mission 4636 reports were used in relief efforts by the US military's Southern Command, and OpenStreetMap data likewise by at least a dozen international agencies, all by the end of those two weeks.

This, to be blunt, was a bit of a miracle. In some ways, no one is more astounded than we that our efforts bore such fruit. The content of nearly all the discussions at ICCM '10 centered ultimately on the following questions: When the next disaster hits, how can we cause the same miracles to happen on time? When relief agencies start depending on the work of the crowd, how do we ensure that the crowd shows up?

These aren't questions with simple answers. For example, volunteer interest hinges, largely, on media attention. The tragic flooding in Pakistan this summer has adversely impacted ten times as many people as the quake in Haiti, but media attention has been far less strident, with the result that many fewer OSM volunteers have stepped up to contribute. Access to the commercial satellite imagery, which makes collaborative remote editing of OSM possible, has also been much harder to obtain, for similar reasons.

Moreover, every non-profit org, formal or ad-hoc, has to contend with volunteer fatigue. The more acute and immediate a disaster is, the easier it is to focus attention on it, which, again, has made crowdmapping Haiti far easier than Pakistan. The longer a situation wears on, the less urgency a volunteer will feel to act, and the more likely it will become that any given volunteer has other, more pressing things to do.

By the same token, casual volunteers can only be called into service so many times before they start to tune out the requests. Consequently, the "Crisis Mapping" community needs to steward its volunteer strength carefully. What criteria do we use to determine that a disaster is acute enough to warrant mounting a response?

Additionally, since our community's capacities naturally vary with time and with the nature of a given crisis event, how can we best set realistic expectations and maintain relationships of trust with the larger, more formal aid organizations that we seek to support?

4. A world without Web.

Question: What happens when a thousand relief actors deploy to the same disaster zone and all turn on their BGAN units at once?

Answer: Absolutely nothing.

Inmarsat's Broadband Global Area Network, aka BGAN, is a satellite broadband service that provides (ideally) a half a megabit downstream to a transceiver that fits in a backpack. Even though the service can cost as much as US$7.50 a megabyte (!), BGAN nevertheless has unimaginable utility to humanitarian aid workers, who have deployed to a part of the world where the (probably already meager) infrastructure has been totally smashed. BGAN transceivers are comparatively cheap -- thousands of US dollars, instead of, say, tens of thousands -- so nearly everyone has one.

Which means that, if everyone shows up in theater with a BGAN unit, the geosynchronous satellite serving that region becomes overloaded instantly, and nothing gets through. This was the case in Haiti for the first few days after the quake, until Inmarsat re-tasked some of the relevant satellite's antennas, at which point BGAN bandwidth went from absolutely nothing to a mere trickle.

It's tough for those of us blessed with a wealth of Internet access to imagine what this is like. While visiting the MapAction volunteers at Logbase in Port-au-Prince, I volunteered to download them the latest OSM updates over their satellite link.

"See, it's only 14 megabytes," I said.

"14 megabytes??" I was greeted with incredulous horror. "Dear Lord, don't download that!"

(Also, it would have cost something like $100, which I didn't know at the time.)

This means, point blank, that any IT solution intended for use by relief workers in a disaster zone needs to work independently of the Internet. I say this, without pointing fingers, because I keep seeing humanitarian aid tools being proposed and developed by well-intended individuals and organizations -- some of them *quite* large -- that depend on ample network access to be of any use. Seriously, guys. Knock it off.

That same technical ingenuity needs to be put into working out how information technology can be used to coordinate humanitarian aid volunteers, both in *and* out of country, on a minimum of bandwidth, as low tech as possible. Andy Smith of MapAction described, for example, an arrangement that pushes incremental updates of GIS data between an ArcGIS server in the UK and another deployed in a disaster response. Much as I dislike the tech they're using -- which is my problem, not theirs -- this is the right kind of thinking, and we need to see more of it.

5. A time to sow, and a time to reap.

"Crisis Mapping" is an inherently reactive practice. The disaster, whatever it is, has already happened by the time the mapping starts, which isalmost by definition, the worst point in time to do anything. The best time to map a disaster, naturally, is *before* the disaster occurs.

Knowing when a disaster will happen entails an unrealistic degree of prescience, but a lot of research goes into anticipating *where* disasters may occur. The disaster hazard and vulnerability for any given locale can be viably assessed by experts, and the places at greatest risk should be prospectively mapped. In fact, this is one of the goals that Stuart Gill and Galen Evans are pursuing with the GeoNode project, as part of their work in disaster risk management at the World Bank. Similarly, MapAction have started sending pre-emptive missions to vulnerable places, e.g. Nepal, to prevent repeats of the scenario that resulted in Haiti, where the national mapping agency's facilities were destroyed, and their data rendered inaccessible.

Meanwhile, back at the conference, Kate Chapman presented the ongoing work of the Humanitarian OpenStreetMap Team in conducting trainings in Cité Soleil, in Port-au-Prince, and Jamie Lundberg talked about the development of Map Kibera in Nairobi. I would like to see the Crisis Mappers community develop more projects like these. Building local technical capacity is slow, sometimes tedious, and far less glamorous than disaster response. On the other hand, working proactively with communities at risk to develop their resilience to possible future disasters is far, far better than having to help pick up the pieces afterwards.

6. Beware the white man's burden.

As a wealthy, educated white guy, I am probably not even entitled to speak up about this, but I fear that the idea of Crisis Mapping runs a perpetual risk of turning condescending, and even a bit imperialistic, towards those whose homes are being mapped. I believe that we who live far from the misery of poverty need to tread very carefully, and be exceedingly humble, when contemplating action ostensibly on behalf of people who are suffering. I observe, without meaning to criticize anyone, that though the name "Haiti" was in the explicit theme of ICCM '10, I nevertheless saw precisely one person from Haiti (Kurt Jean-Charles, from noula.ht) at the conference. To be a responsible practice, Crisis Mapping needs to be fully inclusive, and incorporate the voices and expertise of people who live in the places where the mapping takes place.

A great place to start would be to follow Emily Jacobi's suggestion of raising funds to enable members of the Crisis Mappers community from outside the US and Europe to attend ICCM 2011. The OpenStreetMap community has done this for the last two State of the Map conferences, with reasonable success. A word of warning to next year's organizers, though -- even this little can backfire: In 2010, 7 out of the 15 recipients of "scholarships" to this year's State of the Map were denied visas by EU governments at the last minute, including Guensmork Alcin, an OSM organizer from, yes, Haiti.

7. The rise of the pro-am humanitarian.

Another interesting trend to appear in the past year of Crisis Mapping is the rise in prominence of the "pro-am humanitarian". The quintessential example is Todd Huffman, who runs, with a partner, a beer-for-data program out of a compound in Jalalabad, and has the formal support of... no one at all. He works as a consultant for a number of organizations, and seems to usually get his travel expenses paid, but ultimately humanitarian relief is, in his own words, a hobby and not a profession. This distinction, Todd claims, gives him the freedom to move quickly, and to accomplish things that would be difficult or impossible for others with more substantial backing.

To be fair, this breed of humanitarian worker is hardly new, and Todd is hardly the only example. Everyone in MapAction, everyone involved in the Humanitarian OpenStreetMap Team, and many of the people involved in Mission 4636 were and are committed volunteers, amateurs with professional-grade skills. These "pro-am humanitarians", and the organizations they work with, have demonstrated an ability to accomplish things that would be very difficult for any UN agency or large NGO to manage, on account of their size and speed.

The future, I think, of "Crisis Mapping", whatever it may become, will lie in these two sets of crisis relief participants, professional and pro-am, in learning to build trust, and in starting to work together in a way that leverages the respective strengths of each. We have seen this work once before, but a little haphazardly. Now we need to take the much harder step of learning how to make the miracles happen reliably and on time, before the next time they are needed.

Currently
· Idibon
· Humanitarian OSM Team

Elsewhere
· Twitter
· Flickr
· Github
· LinkedIn

Previously

Before That
[2010] Jan Feb Mar Apr
May Jun Jul Aug
Sep Oct Nov Dec
[2009] Jan Feb Mar Apr
May Jun Jul Aug
Sep Oct Nov Dec
[2006] Jan Feb Mar Apr
May Jun Jul Aug
Sep Oct Nov Dec
[2005] Jan Feb Mar Apr
May Jun Jul Aug
Sep Oct Nov Dec
[2004] Jan Feb Mar Apr
May Jun Jul Aug
Sep Oct Nov Dec
[2003] Jan Feb Mar Apr
May Jun Jul Aug
Sep Oct Nov Dec
[2002] Jan Feb Mar Apr
May Jun Jul Aug
Sep Oct Nov Dec

© copyright 2002-2013 Schuyler Erle * [email protected]
All original material on this website is licensed under the Creative Commons.
= still powered by Blosxom (after all these years) =