humanitarian.info

I promised myself that I’d blog every single day while I was working in Georgia. It should be fairly obvious that I didn’t. I can’t say that I was super productive while I was in Tbilisi – for a variety of reasons, including particularly dysfunctional co-ordination, but also because of the basic difficulty of getting good information in conflict situations.

In a natural disaster, government agencies and international organisations are usually relatively comfortable to share information about the situation – but in a conflict, they clam up tighter than my wallet around Christmas. This is because natural disasters have fewer political implications than complex emergencies; while in a natural disaster the worst thing you can say about a government is that they’re negligent, in a conflict situation the government is usually a belligerent,

This means that timely / reliable / accurate information is hard to come by in Georgia, as Ivan points out and Ethan overviews (is that a verb?). I find it hard to get too worked up about the lack of “citizen war reporters” even though it is my fervent hope that the web is going to change the way we do business in both complex emergencies and natural disasters. My lack of work-up is simply because even if there were shedloads of citizen journalists covering these events, I would still treat them exactly the same as any other information source – which is to say, I wouldn’t trust them at all.

As an example, the single most critical humanitarian information issue in Georgia was the numbers and locations of people displaced by the conflict. This was problematic for a number of reasons:

1. Nobody had a clue how many people had been displaced by the conflict. There were multiple government agencies involved in looking after the IDPs (frequently a euphemism for “ignoring them”, of course), each with their own figures, none of which tallied with the figures that UNHCR or the Red Cross had; and of course nobody in the humanitarian community had bothered to sit down and agree on a number we could all work to. Lesson from Afghanistan, folks: your numbers are never going to be 100% accurate, and it’s better just to agree to a number and get to planning than continually be running after the latest figures – which are also going to be wrong.
2. Nobody wanted to talk about the IDPs left over from the previous round of conflict in 1992-93; a staggering 220,000 people (not 100% accurate, of course – just run with it!) have been rotting in terrible conditions for the last 16 years, and some of their stories can be found on IDP Voices. Nearly all of us who were new to Georgia found this astonishing, since it raised a rather difficult question: what the *&%$ has the government and the UN been doing for the last 16 years? It also confuses the picture because in purely humanitarian terms many of these “old caseload” IDPs were in a worse situation than the “new caseload” – and many of the “normal” citizens live in conditions as bad as either.
3. For both old and new caseloads, the main priority is ensuring their basic shelter, which comes under the Emergency Shelter cluster. Unfortunately the UN in Georgia had decided that they didn’t want to activate the cluster system (because it’s a bit of a hassle and you have to actually take responsibility for your actions) but they did want to use some of the cluster tools (particularly the ones that give you a fat sack of cash to spend). This meant that it was like stepping into a time machine to 2004 – you remember, when “co-ordination” was a competition to see who could hold as many meetings as possible with as few outcomes as possible.
4. Notwithstanding the co-ordination problem, nobody had a clue what to do with all them displaced. The government unveiled a not unreasonable resettlement plan for the new caseload at the start of September, but that plan rapidly ran aground on the harsh reality that the stock and state of public buildings in Georgia are likely not sufficient to house the IDPs according to basic humanitarian standards, even on a short-term basis. (Some interesting discussion on this at the Social Science in the Caucasus blog.) The question is whether that government plan can be reshaped into a more realistic framework that will engage the entire humanitarian community as well as being attractive to donors…

One of the things about shelter issues is that they tend to get worse the longer you leave them. Conditions deteriorate, particularly when people are housed in buildings that were never designed for residential use. In this case, many of the new caseload had been placed in schools and kindergartens around Tbilisi and other towns – which meant that we also had to deal with the fact that those institutions were needed for the start of the new school year. This was a particular tension for UNICEF, who often run a “Back To School” program – which wouldn’t look too good if there weren’t any schools to go back to. In addition winters in Georgia can get unpleasant, especially the closer you get to mountains, and thus another constraint on resettlement.

You might have noticed that there wasn’t much talk about information in this blog post. That’s because there wasn’t much information, as I explained previously. We got hold of the complete set of school locations from the MInistry of Education (shape files ahoy!) but nobody seemed that interested. We tried to persuade the different actors – Red Cross, UNHCR, Ministries various – to consolidate the figures for collective centres and the IDPs therein, but with little luck. Paolo Palmero from OCHA had gathered a lot of data during his 2005 visit, but none of it seemed to be circulating in the agencies.

Summary version: this response showed yet again the importance of investing in information resources before an emergency hits. That doesn’t just mean getting loads of satellite images (although UNOSAT did some impressive work on damage levels) but investing in relationships with government, relationships that can be leveraged quickly to mutual benefit. It means having a basic picture already in place – locations of schools, for example – that you can then overlay new data on top of – such as the estimated IDP numbers in those schools. This really needs a collective approach – one agency alone isn’t sufficient to achieve success, although you need a focal point for the effort – but it continues to make me wonder if we should be thinking about setting up an organisation that collects and disseminates operational data like this.

At least that would avoid me feeling like a numpty, turning up at meetings with my tiny spreadsheet of schools that might need some watsan rehabilitation…

So it all kicked off in Myanmar this week, except that it didn’t, because the military regime has managed to bungle the response to Cyclone Nargis. We could get into a long discussion about the whys and wherefores, and there’s some frightening talk about the “right to respond” over-riding sovereignty, but let’s stay focused on technology. At least it’s relatively non-controversial, except that it isn’t, because Myanmar is one of those places where internet access is a non-starter, where satellite telephones are essentially illegal and where the technology infrastructure (e.g. suppliers and maintenance) is close to zero. What that means is that we’re going to be extremely limited in what we can do on the ground. So what is happening?

• A Sahana instance is being set up for the use of anybody who needs it, with the support of INSTEDD and possible uptake by NetHope members.
• Direct Relief International have done up a KMZ file of health facilities in-country, based on the WHO 2002 Global Health Atlas.
• OCHA are prepping a HIC to support the existing Myanmar Information Management Unit, who have already put out some W3 maps. UPDATE: MapAction have also deployed in support of OCHA and have maps available on their website.
• UNOSAT have also got their sat on with a KMZ file of the cyclone path and the usual satellite mapping.
• Ditto ITHACA, who have released a series of satellite maps showing the impact of Nargis.
• ReliefWeb’s info stream on Cyclone Nargis is of course like drinking water from a hose, with their map filter probably most useful.
• The WorldWideHelp blog roars into action with all the news that’s fit to blog.
• A couple of the mailing list discussions that I’m on are talking about ways in which we might leverage cellphone and/or satellite phone communications if they become available, particularly for tracking relief and relief personnel.
• Digital Globe and Geo-Eye have hopped the NASA satellite for an updating KML layer on the cyclone.
• Microsoft apparently have a team on standby to deploy the refugee tracking software that was developed for Kosovo (no reference yet). Microsoft are focused on supporting the HIC, and are ready to respond to other requests from the humanitarian community.
• Telecoms sans Frontieres are also on standby out of Bangkok, waiting for access to free up.
• Also Infoworld points out that – with regards to early warning – IT didn’t fail Myanmar, people did.

I’m nowhere near being deployed for this one (particularly as the government apparently is still refusing entry to foreign aid workers), but all of this makes me feel that we’re headed in the right direction. However until the government lets agencies start doing their jobs with less restrictions on movement and communications, we’re not going to see the benefits – another example of how the technology can be rendered much less useful when the political environment isn’t supportive. I leave you with the words of Surin Pitsuwan, Secretary-General of ASEAN in his speech last Thursday:

In spite of the technology that we have, in spite of the power that we have, in spite of the network that we have, we still lose lives needlessly… So it is more than just the power of technology, it is more than just the transformation of society through technology, it is certainly a shift in paradigm here in the minds of our people and particularly our leaders. Because if you don’t have that shift, millions and billions of dollars worth of technology cannot deliver effective, timely relief to people when they need it most, because we have reservations about opening up our borders for cooperation, because we have hesitation about cooperating with the outside world, because we have mistrust of the outside world.

The ConnectivIT lab at the University of Colorado has done some fascinating research in the last couple of years, which I’ve been meaning to blog about, but never quite got round to. Such are the workings of the web that these things always come around if you wait long enough. So I’ll preface this blog entry by saying that their work is well worth reading, and that my beef is not with them but with media coverage of technology.

Their latest research, published in New Scientist magazine under Emergency 2.0 is coming to a website near you, suggests that Facebook “is more effective than the emergency services“. This is the sort of headline that makes newspapers and blogs in Magic Future Kingdom soil themselves with excitement, so it’s received wide coverage, which is good – more people interested in these issues means more visibility, more activity, more resources. Unfortunately the coverage in the Daily Telegraph shows the problem with this coverage:

Within just 90 minutes of the first deaths, however, a web page accurately describing the events appeared on web encyclopedia Wikipedia. Twenty minutes after that, Facebook users had set up a group called ‘I’m OK at VT’, which allowed students and staff to reassure the wider world that they were safe. A Facebook discussion was also begun which authoritatively listed the victims and whether people were feared dead rather than confirmed dead.

I’m fascinated to know how we judge the “accuracy” of the Wikipedia entry. Since Wikipedia policy is that nothing should be posted without a citation from a reliable external source, the “accuracy” of that entry must have been wholly dependent on – guess what? That’s right, reports from the media or emergency services. Check the wikipedia page if you want to see for yourself – and if there weren’t any citations, then how on earth can you tell if it’s accurate or not?

We can chalk these statements on journalistic shorthand. The real problem with this is that there is absolutely no accountability for Wikipedia, Facebook or other social media. If those reports weren’t accurate – if you went to Facebook, read that your son was dead and later found he was alive, for example – then that’s a lot of trauma that nobody will ever take responsibility for. For the emergency services, it’s a bit more serious than that – if they get it wrong, they get sued to oblivion, people lose their jobs and their credibility goes out of the window.

How do you know when somebody is dead in a situation like Virginia Tech? When they’re officially declared dead. Who officially declares them dead? A medical professional, a member of the emergency services. The idea that in Magic Future Kingdom we’ll just automatically know when somebody is dead is ludicrous – maybe their Twitter stream will stop or something?

The article does make good points, more rooted in the research. People on the ground are the source of a lot of information, and technology makes it easier for them to get that information out. It’s also likely that the more people you aggregate, the more accurate the information will be, which I think is Leysia Palen’s point about how these events show “socially produced accuracy”, i.e. a version of the wisdom of crowds. Yet there are limits to that accuracy, and there is a question about how useful that information in terms of actually dealing with the emergency – of which notifying relatives is only one small part. The need for a central authority that can route all this information is a foundational point of effective disaster management – so what are the implications of these developments for effective disaster management?

Sanjana also makes a point which I’d agree with entirely:

Of course, what it means is that Facebook, in the US, with reliable broadband wired and wireless coverage, with a ubiquity of PCs, where everyone speaks, reads and comprehends English, where Universities are well connected, where everyone has laptops and where everyone and their pet Chihuahua have a Facebook account, the platform can on occasion get more information out quicker than emergency services.

Time for us to shell out a few bucks for a New Scientist subscription and read the actual research article, rather than the press coverage. Plus, I need to get back on Medication 2.0 or something.

An email from CEO Eric Rasmussen tells me that INSTEDD is finally flying in radar (and apparently I’m mixing metaphors, unsure of whether INSTEDD is a whale or a plane). In his words,

InSTEDD has been invisible, a rumor and a ghost, for the few past months, but we surfaced today in a media call with Google.org in the launch of their first-ever Initiatives.

I’ve known Eric virtually for a couple of years, although we’ve never managed to actually meet in person. He’s a very solid choice for CEO – his thinking on civil-military affairs was always more lateral than I expected for somebody in his position, and I think that it reflected his willingness to listen and learn from others. That open attitude will be the single most important tool in INSTEDD’s box, at least in its early days, and in fact that’s why they are only launching now – for the last few months, the team has been travelling around meeting with various actors and hearing out what their needs might be.

If you don’t know anything about INSTEDD, I recommend that you visit the website, as it gives you the right flavour. It’s moved away from Larry Brilliant’s 2006 talk at TED and morphed into something that’s somehow broader yet less ambitious (a good combination if you can manage it). Yet I still harbour doubts about how successful it will be, doubts which don’t have anything to do with the goodwill or competence of the staff (or the fact that they’ve got a ton of funding behind them, not least from Google.Org).

Where do my doubts come from?  I’ll freely admit that part of my concern is rooted in my essential antipathy towards advice from experts (yes, including myself – it keeps me on my toes); part of it is based on my lack of faith in the power of technology to save us from ourselves; and part of it is linked to my sense that the entire system of disaster response has been built with a fundamental flaw at its foundation, and trying to make it more effective may simply be throwing good money after bad.

Having said that, INSTEDD’s philosophy is healthy.  It’s not proposing overarching systems that will solve the whole set of problems in one go, but a piece-by-piece approach that addresses key problems in the field as they arise.  So I welcome their launch, and I definitely welcome their presence on the scene – it’s definitely healthy for the sector to have more non-traditional actors bringing some new perspective.

And new tools, of course!  INSTEDD will be releasing any tech tools it develops, and it looks like on an open source basis (more specifically, the MIT license).  First out of the gate is a GeoSMS system, which is built on a bunch of existing Google tools and some GeoRSS.  And yes, we hope to discuss how Sahana might fit into these developments…

In the last few days of my mission to Bangladesh, I chaired a meeting of the Information Management Working Group in Dhaka, at which something curious and exciting happened. The Group is comprised of representatives from each of the clusters, the CDMP (a government body) and usually a couple of others (today, CARE – it’s really just a place to reach agreement on basic issues like geocodes.

But in the meeting we actually agreed that all the clusters would pursue a policy of open data – that all data collected during assessments would be shared in its raw form as soon as the collecting organisation had published its own report, through the offices of the Disaster Management Information Centre (DMIC).  This will make it possible for other agencies to incorporate that data into their own analysis, enabling better cross-cluster co-ordination.

As far as I know, this is the first time that this has happened in any humanitarian response operation anywhere (I could be wrong).  It could be that the cluster leads won’t be able to live up to this commitment, since there are all sorts of internal agency constraints.  Nevertheless I thought it was indicative that we are moving in the right direction, and that alone made my work in Bangladesh worthwhile.

The single most important co-ordination tool that you will find in any humanitarian response is the Who’s doing What Where, better known as the W3 or 3W. This is the basic tracking that shows which organisations (who) are doing which activities (what) in which locations (where). It comes in a limited range of flavours, and it’s usually a spreadsheet with a simple matrix by location.

For some reason, here they’ve decided to add an extra W – When. I have no idea why they decided to do this, since we’ve never been able to successfully get the first 3Ws; the decision was taken before I arrived, and I would have fought tooth and claw to stop it. To be fair, I’ve been asked for a chronological component in other places, because it’s really useful to know what plans organisations have.

The problem is, I think the W3 sucks. To be more precise, I think our current approach to the W3 sucks. Why do I think so? Because it never bloody works, that’s why. I’ve been through this process in every emergency I’ve ever worked in, and I’ve never seen it work. Sometimes we’ve managed to fake it, but that’s not the same thing as having a really solid information system that provides useful analysis.

There’s no doubt that knowing who’s doing what where is essential to coordinating a humanitarian response, so we need to rethink the entire thing urgently.  I chaired a meeting of the Information Management Working Group yesterday, and we reflected on the problems that we’d had implementing the W3 and how we’d dealt with them.  That will go into a longer document that Neil Bauman and I are going to write up, but here are the headline problems for me right now:

a. We set up templates for people to fill out, but they never use them. This is because the forms don’t resemble anything they use internally for planning their activities, and they don’t have a good reason to fill out our forms in addition to their own.

b. When they do use our formats, they inevitably fill them out incorrectly. This means that you have to work out what on earth they were trying to tell you when they were typing out their information. Is that really 18 water purification plants in one union location? Surely not.

c. We ask for quite a high level of detail – in my opinion, an unreasonably high level of detail – without really asking whether that much detail is useful.   As a result, people are often put off by the thought of that level of reporting – particularly if their own organisation doesn’t ask them for that much information, why should a co-ordinator?

d. As a result, most of the time we just get the information in whatever format they have with whatever detail they have it in already or can be bothered to submit it in. This means that we have to shoehorn it into our format (usually unsuccessfully) and then pretend that we’ve got some useful data.  In reality, the coverage is patchy at best and the accuracy is questionable – by the time you’ve typed it up, it’s already out of date.

e. A lot of the time – especially in the early stages of a response – people simply don’t know exactly what their organisation is doing. They know roughly what activities their organisations are going to undertake, and in roughly what locations, but precision is not their usual frame of mind.

Having said that, it’s clear that people and organisations do plan their activities, so the question is – how do we get a better picture of what they’re doing? That will probably be the subject of my next post, once I’ve finished cutting and pasting NFI distribution data into this bloody spreadsheet for the meeting today.