“If all You Have is a Hammer…” - How Useful is Humanitarian Crowdsourcing?

This article was originally published on MobileActive on 20 October 2010. At the time there was a lot of debate about the article because people misinterpreted it as a blanket dismissal of crowdsourcing, which it plainly was not. The MobileActive website is no longer… Active, but people continue to ask me about the article, so I am re-posting it here. I stand by most of my critique, particularly as many of the points were later confirmed by Ushahidi’s own evaluation of the Haiti project. I am not posting it to renew the discussion, but because I believe the points it makes about new technology still stand, particularly at the end of the article: For while the tools that we now have at our disposal are important, we have a responsibility to use them for the right tasks.

The original article was accompanied by an Editor’s Note: In this article, guest contributor Paul Currion looks at the potential for crowdsourcing data during large-scale humanitarian emergencies, as part of our “Deconstructing Mobile” series. Paul is an aid worker who has been working on the use of ICTs in large-scale emergencies for the last 10 years. He asks whether crowdsourcing adds significant value to responding to humanitarian emergencies, arguing that merely increasing the quantity of information in the wake of a large-scale emergency may be counterproductive. Instead, the humanitarian community needs clearly defined information that can help in making critical decisions in mounting their programmes in order to save lives and restore livelihoods. By taking a close look at the data collected via Ushahidi in the wake of the Haiti earthquake, he concludes that crowdsourced data from affected communities may not be useful for supporting the response to a large-scale disaster.

1. The Rise of Crowdsourcing in Emergencies

Ushahidi, the software platform for mapping incidents submitted by the crowd via SMS, email, Twitter or the web, has generated so many column inches of news coverage that the average person could be mistaken for thinking that it now plays a central role in coordinating crisis responses around the globe. At least this is what some articles say, such as Technology Review’s profile of David Kobia, Director of Technology Development for Ushahidi. For most people, both inside and outside the sector, who lack the expertise to dig any deeper, column inches translate into credibility. If everybody’s talking about Ushahidi, it must be doing a great job — right?


Ushahidi is the result of three important trends:

  1. Increased availability and utility of spatial data;
  2. Rapid growth of communication infrastructure, particularly mobile telephony; and
  3. Convergence of networks based on that infrastructure on Internet access.

Given those trends, projects like Ushahidi may be inevitable rather than unexpected, but inevitability doesn’t give us any indication of how effective these projects are. Big claims are made about the way in which crowdsourcing is changing the way in which business is done in other sectors, and now attention has turned to the humanitarian sector. John Della Volpe’s short article in the Huffington Post is an example of such claims:

“If a handful of social entrepreneurs from Kenya could create an open-source “social mapping” platform that successfully tracks and sheds light on violence in Kenya, earthquake response in Chile and Haiti, and the oil spill in the Gulf — what else can we use it for?”

The key word in that sentence is “successfully”. There isn’t any evidence that Ushahidi “successfully” carried out these functions in these situations; only that an instance of the Ushahidi platform was set up. This is an extremely low bar to clear to achieve “success”, like claiming that a new business was successful because it had set up a website. There has lately been an unfounded belief that the transformative effects of the latest technology are positively inevitable and inevitably positive, simply by virtue of this technology’s existence.

2. What does Successful Crowdsourcing Look Like?

To be fair, it’s hard to know what would constitute “success” for crowdsourcing in emergencies. In the case of Ushahidi, we could look at how many reports are posted on any given instance — but that record is disappointing, and the number of submissions for each Ushahidi instance is exceedingly small in comparison to the size of the affected population — including Haiti, where Ushahidi received the most public praise for its contribution.

In any case, the number of reports posted is not in itself a useful measure of impact, since those reports might consist of recycled UN situation reports and links to the Washington Post’s “Your Earthquake Photos” feature. What we need to know is whether the service had a significant positive impact in helping communities affected by disaster. This is difficult to measure, even for experienced aid agencies whose work provides direct help. Perhaps the best we can do is ask a simple question: if the system worked exactly as promised, what added value would it deliver?

As Patrick Meier, a doctoral student and Director of Crisis Mapping and Strategic Partnerships for Ushahidi has explained, crowdsourcing would never be the only tool in the humanitarian information toolbox. That, of course, is correct and there is no doubt that crowdsourcing is useful for some activities — but is humanitarian response one of those activities?

A key question to ask is whether technology can improve information flow in humanitarian response. The answer is that it absolutely can, and that’s exactly what many people, including this author, have been working on for the last 10 years. However, it is a fallacy to think that if the quantity of information increases, the quality of information increases as well. This is pretty obviously false, and, in fact, the reverse might be true.

From an aid worker’s perspective, our bandwidth is extremely limited, both literally and metaphorically. Those working in emergency response — official or unofficial, paid or unpaid, community-based or institution-based, governmental or non-governmental — don’t need more information, they need better information. Specifically, they need clearly defined information which can help them to make critical decisions in mounting their programmes in order to save lives and restore livelihoods.

I wasn’t involved with the Haiti response, which made me think that perhaps my doubts about Ushahidi were unfounded and that perhaps the data they had gathered could be useful. In the course of discussions on Patrick Meier’s blog, I suggested that the best way for Ushahidi to show my position was wrong would be to present a use case to show how crowdsourced data could be used (as an example) by the Information Manager for the Water, Sanitation and Hygiene Coordination Cluster, a position which I filled in Bangladesh and Georgia. Two months later, I decided to try that experiment for myself.

3. In Which I Look At The Data Most Carefully

The only crowdsourced data I have is the Ushahidi dataset for Haiti, but since Haiti is claimed as a success, that seemed like to be a good place to start. I started by downloading and reading through the dataset — the complete log of all reports posted in Ushahidi. It was a mix of two datastreams:

  • Material published on the web or received via email, such as UN sitreps, media reports, and blog updates, and
  • Messages sent in by the public via the 4636 SMS shortcode established during the emergency.

I was struck by two observations:

  • One of the claims made by the Ushahidi team is that its work should be considered an additional datastream for aid workers. However, the first datastream is simply duplicating information that aid workers are already likely to receive.
  • The 4636 messages were a novel datastream, but also the outcome of specific conditions which may not hold in places other than Haiti. The fact that there is a shortcode does not guarantee results, as can be seen in the virtually empty Pakistan Ushahidi deployment.

I considered that perhaps the 4636 messages could demonstrate some added value. They fell into three broad categories: the first was information about the developing situation, the second was people looking for information about family or friends missing after the earthquake, and the third and by far the largest, was general requests for help.

I tried to imagine that I had been handed this dataset on my deployment to Haiti. The first thing I would have to do is to read through it, clean it up, and transcribe it into a useful format rather than just a blank list. This itself would be a massive undertaking that can only be done by somebody on the ground who knows what a useful format would be. Unfortunately, speaking from personal experience, people on the ground simply don’t have time for that, particularly if they are wrestling with other data such as NGO assessments or satellite images.

For the sake of argument, let’s say that I somehow have the time to clean up the data. I now have a dataset of messages regarding the first three weeks of the response. 95% of those messages are for shelter, water and food. I could have told you that those would be the main needs even before I arrived in position, so that doesn’t add any substantive value. On top of that, the data is up to 3 weeks old: I’d have to check each individual report just to find out just whether those people are still in the place that they were when they originally texted, and whether their needs have been met.

Again for the sake of argument, let’s say that I have a sufficient number of staff (as opposed to zero, which is the number of staff you usually have when you’re an information manager in the field) and they’ve checked every one of those requests. Now what? There are around 3000 individual “incidents” in the database, but most of those contain little to no detail about the people sending them. How many are included in the request, how many women, children and old people are there, what are their specific medical needs, exactly where they are located now — this is the vital information that aid agencies need to do their work, and it simply isn’t there.

Once again for the sake of argument, let’s say that all of those reports did contain that information — could I do something with it? If approximately 1.5 million people were affected by the disaster, those 3000 reports represent such a tiny fraction of the need that they can’t realistically be used as a basis for programming response activities. One of the reasons we need aid agencies is economies of scale: procuring food for large populations is better done by taking the population as a whole. Individual cases, while important for the media, are almost useless as the basis for making response decisions after a large-scale disaster.

There is also this very basic technical question: once we have this crowdsourced data, what do we do with? In the case of Ushahidi, it was put on a Google Maps mash-up — but this is largely pointless for two reasons. First, there’s a simple question of connectivity. Most aid workers and nearly all the population won’t have reliable access to the Internet, and where they do, won’t have time to browse through Google Maps. (It’s worth noting that this problem is becoming lessimportant as Internet connectivity, including the mobile web, improves globally — but also that the places and people prone to disasters tend to be the last to benefit from that connectivity.)

Second, from a functional perspective, the interface is rudimentary at best. The visual appeal of Ushahidi is similar to that of Powerpoint, casting an illusion of simplicity over what is, in fact, a complex situation. If I have 3000 text messages saying “I need food and water and shelter”, what added value is there from having those messages represented as a large circle on a map? The humanitarian community often lacks the capacity to analyse spatial data, but this map has almost no analytical capacity. The clustering of reports (where larger bubbles correspond to the places that most text messages refer to) may be a proxy for locations with the worst impact; but a pretty weak proxy derived from a self-selecting sample.

In the end, I was reduced to bouncing around the Ushahidi map, zooming in and out on individual reports — not something I would have time to do if I was actually in the field. Harsh as it sounds, my conclusion was that the data that crowdsourcing of this type is capable of collecting in a large-scale disaster response is operationally useless. The reason for this has nothing to do with Ushahidi, or the way that the system was implemented, but with the very nature of crowdsourcing itself.

4. Crowdsourcing Response or Digital Voluntourism?

One of the key definitions of “crowdsourcing” was provided by Jeff Howe in a Wired article that originally popularised the term: taking “a job traditionally performed by a designated agent (usually an employee) and outsourcing it to an undefined, generally large group of people in the form of an open call.” In the case of Haiti, part of the reason why people mistakenly thought crowdsourcing was successful, was because there were two different “crowds” being talked about.

The first was the global group of volunteers who came together to process the data that Ushahidi presented on its map. By all accounts, this was definitely a successful example of crowdsourcing as per Howe’s definition. We can all agree that this group put a lot of effort into their work. However, the end result wasn’t especially useful. Furthermore, most of those volunteers won’t show up for the next response — and in fact they didn’t for Pakistan.

The media coverage of Ushahidi focuses mainly on this first crowd — the group of volunteers working remotely. Yet, the second crowd is much more important: the affected community. Reading through the Ushahidi data was heartbreaking, indeed. But we already knew that people needed food, water, shelter, medical aid — plus a lot more things that they wouldn’t have been thinking of immediately as they stood in the ruins of their homes. In the Ushahidi model, this is the crowd that provides the actual data, the added value, but the question is whether crowdsourced data from affected communities could be useful from an operational perspective of organising the response to a large-scale disaster.

The data that this crowd can provide is unreliable for operational purposes for three reasons. First, you can’t know how many people will contribute their information, a self-selection bias that will skew an operational response. Second, the information that they do provide must be checked — not because affected populations may be lying, but because people in the immediate aftermath of a large-scale disaster do not necessarily know all that they specifically need or may not provide complete information. Third, the data is by nature extremely transitory, out-of-date as soon as it’s posted on the map.

Taken together, these three mean that aid agencies are going to have to carry out exactly the same needs assessments that they would have anyway — in which case, what use was that information in the first place?

5. Is Crowdsourcing Raising Expectations That Cannot be Met?

Many of the critiques that the crowdsourcing crowd defend against are questions about how to verify the accuracy of crowdsourced information, but I don’t think that’s the real problem. It’s the nature of an emergency that all information is provisional. The real question is whether it’s useful.

So to some extent those questions are a distraction from the real problems: how to engage with affected communities to help them respond to emergencies more effectively, and how to coordinate aid agencies to ensure and effective response. On the face of it, crowdsourcing looks like it can help to address those problems. In fact, the opposite may be true.

Disaster response on the scale of the Haiti earthquake or the Pakistan floods is not simply a question of aggregating individual experiences. Anecdotes about children being pulled from rubble by Search and Rescue teams are heart-warming and may help raise money for aid agencies but such stories are relatively incidental when the humanitarian need is clean water for 1 million people living in that rubble. Crowdsourced information — that is, information voluntarily submitted in an open call to the public — will not ever provide the sort of detail that aid agencies need to procure and supply essential services to entire populations.

That doesn’t mean that crowdsourcing is useless: based on the evidence from Haiti, Ushahidi did contribute to Search and Rescue (SAR). The reason for that is because SAR requires the receipt of a specific request for a specific service at a specific location to be delivered by a specific provider — the opposite of crowdsourcing. SAR is far from being a core component of most humanitarian responses, and benefits from a chain of command that makes responding much simpler. Since that same chain of command does not exist in the wider humanitarian community, ensuring any response to an individual 4636 message is almost impossible.

This in turn raises questions of accountability — is it wholly responsible to set up a shortcode system if there is no response capability behind it, or are we just raising the expectations of desperate people?

6. Could Crowdsourcing Add Value to Humanitarian Efforts?

Perhaps it could. However, the problem is that nobody who is promoting crowdsourcing currently has presented convincing arguments for that added value. To the extent that it’s a crowdsourcing tool, Ushahidi is not useful; to the extent that it’s useful, Ushahidi is not a crowdsourcing tool.

To their credit, this hasn’t gone unnoticed by at least some of the Ushahidi team, and there seems to be something of a retreat from crowdsourcing, described in this post by one of the developers, Chris Blow:

“One way to solve this: forget about crowdsourcing. Unless you want to do a huge outreach campaign, design your system to be used by just a few people. Start with the assumption that you are not going to get a single report from anyone who is not on your payroll. You can do a lot with just a few dedicated reporters who are pushing reports into the system, curating and aggregating sources.”

At least one of the Ushahidi team members now talks about “bounded crowdsourcing” which is a nonsensical concept. By definition, if you select the group doing the reporting, they’re not a crowd in the sense that Howe explained in his article. This may be an area where Ushahidi would be useful, since a selected (and presumably trained) group of reporters could deliver the sort of structured data with more consistent coverage that is actually useful — the opposite of what we saw in Haiti. Such an approach, however, is not crowdsourcing.

Crowdsourcing can be useful on the supply side: for example, one of the things that the humanitarian community does need is increased capacity to process data. One of the success stories in Haiti was the work of the OpenStreetMap (OSM) project, where spatial data derived from existing maps and satellite images was processed remotely to build up a far better digital map of Haiti than existed previously. However, this processing was carried out by the already existing OSM community rather than by the large and undefined crowd that Jeff Howe described.

Nevertheless this is something that the humanitarian community should explore, especially for data that has a long-term benefit for affected countries (such as core spatial data). To have available a recognised group of data processors who can do the legwork that is essential but time-consuming would be a real asset to the community — but there we’ve moved away from the crowd again.

7. A Small Conclusion

My critique of crowdsourcing — shared by other people working at the interface of humanitarian response and technology — is not that it is disruptive to business asusual. My critique is that it doesn’t work — not just that it doesn’t work given the constraints of the operational environment (which Ushahidi’s limited impact in past deployments shows to be largely true), but that even if the concept worked perfectly, it still wouldn’t offer sufficient value to warrant investing in.

Unfortunately, because Ushahidi rests its case almost entirely on the crowdsourcing concept, this article may be interpreted as an attack on Ushahidi and the people working on it. However, all of the questions I’ve raised here are not directed solely at Ushahidi (although I hope that there will be more debate about some of the points raised) but hopefully will become part of a wider and more informed debate about social media in general within the humanitarian community.

Resources are always scarce in the humanitarian sector, and the question of which technology to invest in is a critical one. We need more informed voices discussing these issues, based on concrete use cases because that’s the only way we can test the claims that are made about technology. For while the tools that we now have at our disposal are important, we have a responsibility to use them for the right tasks.

AidCoin: a revolution in humanitarian financing


The World Humanitarian Summit has opened up a space for debate in the humanitarian sector. Questions about how humanitarian action is financed — where the money comes from, the channels through which it goes, and the way in which it is finally disbursed — have been the focus of much research and discussion, including by High Level Panels contributing to the WHS. We now have a much clearer picture of humanitarian financing, including the problems that it faces.1

A revolution is happening around the humanitarian sector, and our existing financing system is not fit for purpose. Many of the large number of recommendations about how to fix this system have been very good, but all have sought to tweak the existing system rather than accept that system is fundamentally flawed. This short paper outlines a speculative model through which new financial technology (fintech) might make possible a complete revision of humanitarian financing.

Why is this important?

Financing is a critical issue because to a large extent it determines incentives within the sector; and misaligned incentives are the primary reason why accountability in the sector remains a problem. It is important to note that this is a systemic and not an individual problem; many aid workers are committed to accountability, but operate in a system that works against this commitment. Accountability therefore is not merely a question of providing staff with more training and better resources to introduce accountability into their programmatic work — although obviously this will help, and should be encouraged.

While humanitarian organisations struggle to become more accountable to those they claim to serve, their actual obligations — financial and political — are to their donors, who tend to be institutional donors, rich publics, and (increasingly) private philanthropists. Needless to say, none of these three categories can claim to speak for disaster-affected communities.

The problem of accountability is deeply connected to another large problem: transparency. Attempts to track aid finance (such as OCHA’s Financial Tracking System, or FTS) have been frustrated by the complicated nature of aid flows, and the unwillingness of key players — particularly at the top end of the chain — to work towards increased transparency, mainly for political reasons. This has come at the same time as growing calls for increased transparency in a hyper-mediated world — which itself is one of the reasons why the WHS was convened, to address the systemic problems that are increasingly visible to both affected communities and giving publics.

In addition there are second-order questions of effectiveness — whether these funds produce the intended results — and efficiency — whether those results are achieved with a minimum of waste. It is impossible to adequately measure these two variables partly because of the first-order problems within the financing chain. Without being able to track funds from point to end point, it is very difficult to judge whether they have been effective or efficient, whether internally (according to the terms of contract between the donor and implementing agency) or externally (according to the understanding between the implementing agency and the affected community).

At the “bottom” end of the chain of humanitarian finance, cash distributions have become recognised as a legitimate form of relief assistance.2 This has opened up space for humanitarian actors to develop tools to assess local markets as a critical part of humanitarian needs assessment and programme development. Remittances have also now been recognised as a critical aid flow — primarily in terms of development, but also during times of crisis. In many of the countries in which we work, the financial sector is relatively under-developed; people rely on alternative channels such as hawala and innovative solutions such as mobile banking.


The humanitarian community (primarily the large institutional donors and large international organisations) establish a dedicated cryptocurrency, referred to for convenience here as AidCoin. All major funding is converted from its original currency into AidCoin at a pre-agreed rate of exchange and placed in escrow pooled funds: crisis-specific funds, regional response funds, and one global response fund. These pooled funds can be used to safely store funding and smooth out the response, thus enabling the system to respond more easily to “forgotten emergencies” — indeed it would be possible to earmark some funds for exactly that.

AidCoin ensures that every single unit of funding can be tracked from the point that it enters any given pooled fund, to the point at which it leaves. This can initially be as simple as funding going from e.g. DFID to Save the Children UK, and can leave the fund after SCUK transfers it to their country office. At this point the primary advantage is ease of disbursement: DFID places their money in the fund under a smart contract that enables them to release it immediately and with zero transaction costs when certain conditions are met regarding the situation on the ground and the capability of the agency.

DFID and SCUK agree with their major suppliers and logistics partners to use AidCoin for all transactions. These external actors accept AidCoin with the security that it is backed by actual money that can be seen within the fund, and can be redeemed in a currency of their choice. These contractual agreements are also supported by smart contracts that enable immediate release of funds when pre-agreed conditions are met. Once major private sector actors accept AidCoin under these conditions, smaller actors are also more likely to accept them — both horizontally (at the international and regional level) and vertically (in disaster-affected countries).

Implementing agencies are then be free to use AidCoin down the complete length of their logistics chain, to point of purchase, including national offices and partners. This cuts down transaction costs and removes exchange rate of losses while maintaining transparency, since (like Bitcoin) every AidCoin can be tracked in detail by everybody within the network while (unlike Bitcoin) everybody within the network is equally visible and the precise trail of funding is clear. The disadvantage at this stage is that it is hard to see how agencies can justify their complete set of overhead costs.

The final step extends AidCoin to disaster-affected communities. These communities benefit from the advantages of cash (flexibility and anonymity) with none of the disadvantages (insecurity associated with handling cash, protection from price volatility and exchange rate fluctuations). Recipient security could be ensured by using the same blockchain technology to generate unique IDs supported by biometric verification; this could be done without compromising recipient privacy.3 The danger of AidCoin versus hard currency is that AidCoin may limit the options for affected communities: for example it can not be easily transported across borders to a new country in which AidCoin has not yet been established.

While it may seem unlikely that these communities will adopt such a technology, it should be borne in mind that mobile banking is usually conducted without cash exchange, that voucher systems have been successfully implemented as part of wider cash programming, and ATMs have been successfully incorporated into relief distributions. All of these approachess can be incorporated into the AidCoin system without changing the processes behind them or the experience of the users. Once AidCoin is in use at community level, it enables a wide range of economic activities that might not be possible in a barter or cash economy. Entire economies — particularly within relatively bounded environments such as refugee camps — could be built on AidCoin.

It is possible to track AidCoins from the donor purse to the beneficiary purse, thus creating complete transparency without requiring any additional reporting. This transparency cuts both ways: donors are able to see exactly how their money was being used, but recipients are also able to see exactly where their products and services came from. It will also support coordination, since duplication will become much easier to identify’ and there will be an incentive for donors and agencies to fill gaps since this will also be much more visible.

AidCoins can be used by affected communities or individuals to purchase services and products from aid agencies, thus changing the incentive pattern and contributing to accountability. Since AidCoins are interchangeable between organisations, it is possible for affected communities to “vote with their wallets” in terms of who they purchase services from, thus encouraging improved service and competition within the aid effort. It is also possible to establish smart contracts between aid agencies and affected communities, so that agencies only receive funding if the community feels it has been well-served.

AidCoin also makes it easier for national partners or authorities to inherit these services as part of agency exit strategies. It is also possible to allocate AidCoins for specific activities — such as capacity building — that enable such strategies without fear that these funds would be misused. Legacy AidCoin could be integrated into national-level finance systems through standardised agreement between e.g. the UN and Ministry of Finance, although this will become complicated depending on local regulatory frameworks.

AidCoin could also be used to make earmarking funds far more effective, whether for geographic or thematic focus. Donors or agencies that want to promote e.g. gender-based violence as a critical issue could earmark their AidCoins, and subsequently see very clearly whether they were spent on relevant activities, and consequently how cost-effective those activities were. This would also identify areas which are currently not being paid enough attention, allowing donors and agencies to make more informed decisions about where to direct their funding in order to have most impact. Donor coordination would become much easier as a result.

AidCoin changes incentives around many aspects of the aid industry — for example, once the costs associated with e.g. a London headquarters become clear, it makes much more sense to locate more capacity in cheaper locations closer to disaster areas, thus contributing to decentralisation and capacity-building within the sector, which can be supported by decentralised decision-making processes made possible by blockchain technology.

What are the major challenges?

1. The entire humanitarian system cannot move to AidCoin in one go: the solution is to extend AidCoin in stages across the full extent of humanitarian action — wherever financing is required — until it spans the entire system.

2. None of this will be possible unless AidCoin has real-world value. While it stays within the humanitarian financing chain, there are no problems: as soon as any actor wishes to exchange it for other kinds of value outside the chain — products, services or other currencies — the humanitarian community needs to work out how to address the following scenarios:

  • If an aid agency pays its staff, both international and national, in AidCoin, how do those staff redeem that currency for real-world expenditure?
  • If a logistics company receives AidCoin upon delivery of e.g. tarpaulins, how do they use those AidCoins to pay the initial manufacturer of those tarpaulins?
  • How does a merchant in e.g. Myanmar redeem the AidCoin she receives from a refugee to purchase a jerry-can?

These are not insurmountable problems. Most monetary transactions — especially at scale — are run as virtual through the international banking system. The first two problems could be addressed through co-operation with that system, either to provide points of interface for AidCoin users, or to establish a parallel system that moves money in and out of AidCoin. The third problem is more difficult: on the ground, many people operate in a cash or barter economy; however it has already been pointed out that non-cash economies have already been created in e.g. Kenya (for example, paying in mobile telephone credit), and AidCoin might be no different.

3. Interoperability with other parts of the international system — particularly development and peacekeeping budgets — will be problematic, particularly for those who believe that humanitarian relief needs to be programmed in coordination with those other resources. In one way AidCoin would make it much clearer what is being spent on relief vs development — in terms of being able to disaggregate overall funding — and thus how to use the two funding streams in a complementary way. However unless AidCoin (or a similar approach) was also adopted in these other sectors, there would need to be the same sort of interface as discussed in point 2 above.

4. AidCoin does not address the challenges facing core humanitarian principles such as neutrality and impartiality. At present, the massive technical problem of humanitarian financing is a distraction from the massive political problem of humanitarian financing. As a technical solution with political implications, implementation of AidCoin will enable the humanitarian community to stop worrying about the technical questions and focus on those principles more tightly.

5. The disincentives for this system are huge. They remove many of the opportunities for agencies to recover operating costs — a legitimate concern — as well eliminating less legitimate concerns such as corruption within the supply chain. Both of these can be flipped, however: less operating costs through project budgets would force donors to accept core costs as legitimate expenses that have been covered up for political reasons; and agencies would no longer need to turn a blind eye to corruption, since the opportunities for fraud would be greatly decreased. The political costs to this system would be high at every level: it would force many of the systemic problems fully into the light. At the moment they are half-concealed, however, which is the worst of both worlds.


AidCoin is not the answer for every problem faced by the humanitarian community — for example it does not address the ongoing global shortfall in humanitarian funding — but it does address many of the significant issues. It requires more research and development, but more importantly it requires pilot testing to begin as soon as possible in order to start learning. The technology that makes this type of approach possible will develop quickly and may move in unexpected directions: the only way to take advantage of it is to start moving quickly and keep momentum. The humanitarian community has historically not been good at adopting new technologies: and unlike previous technologies, this one represents a major shift in the system.


This note is intended as a basis for discussion about the potential role of new fintech in the humanitarian sector. It does not provide a comprehensive analysis of financial flows in the sector, or a rigorous model for the application of fintech in disaster settings. Accordingly constructive comments are welcome to critique and improve both the analysis and the model — bearing in mind that papers in the humanitarian sector are usually far too long-winded!

1See, inter alia: Making financing work for crisis-affected people, Charlotte Latimer, July 2015; Looking Beyond the Crisis, Future Humanitarian Financing; Financing in Crisis?, Rachel Scott, June 2015.

2See the website of the Cash Learning Partnership for more resources; the ODI HPN paper “Cash transfer programming in emergencies” gives a good overview, although it is now 4 years old; DFID has convened a High Level Panel on Humanitarian Cash Transfers to contribute to the WHS.

3See specific proposal in “Distributed crypto identity as a mechanism for legal empowerment of the poor and stimulating local economic development”, Gavin Chait, 18 January 2015; and wider discussion in “Cheap ID”, Vinay Gupta, 2006

Time under water

You think of these migraines as something outside yourself.

When you wake from that afternoon hibernation, you think to yourself: “Is it gone yet?” No, it hasn’t gone yet. It rides you like guilt, bearing you down. It throbs and writhes just beneath the skin of your scalp, leaning against your eyeballs with all its elbows. It makes you weep when you accidentally look out of the window into the bright sunlight. It’s a monster, announcing itself early in the morning with that faint ache around the eyes, that nausea on an empty stomach, that thirst that you feel too late and now cannot be quenched in time to stop it.

When it eventually hits you, you will lose the day. You can’t hope to beat it; you just have to survive. Survival means what survival has always meant; curled into the fetal position in warmth and darkness, reliving any memories that can take you away from that place, from that pain. The migraine turns you into a monster: the vampire, seeking the darkness, sleeping during the day; the zombie, shuffling around the house when you become desperate for food. It wants nothing more than to make you a monster like itself.

When it’s especially bad, you pray that you might die (and sometimes you even mean it), but always you survive. Your mind keeps working all the way through, running away at its pace until finally you fall asleep, but that sleep is not refreshing. You wake up with ashes in your mouth, feeling as if your skull has been hollowed out. You are light on your feet finally, after that zombie shuffle you had before, but only because your brain still reels from the impact.

It doesn’t kill you, but it doesn’t make you stronger either; it only reminds you that you are at the mercy of the monster. It wants nothing; it’s just a reflection of the brain misfiring, somehow, somewhere. The monster is your own mind, and the only lesson it has to teach is that you are at its mercy. You’ll forget that lesson, of course. You always forget that lesson, until the next time, when you wake up with that faint ache around your eyes; and the monster, eating the precious day straight from your table.

Es-Tu, Charlie?

Je-suis-CharlieI’m deeply sorry for those killed in the attack on Charlie Hebdo and its aftermath, and for the pain of their families and friends. I have to say these things in advance of anything else I write, because – well, because those are the things you’re supposed to say. But I didn’t know any of those people, or their family, or their friends. I didn’t read their magazine, and I probably wouldn’t have read it even if I’d been in Paris. My sympathy is strictly mediated, and for public consumption.

In all honesty, I’d rather save my sympathy. I’d rather ration my outrage. I’m worried by attacks on freedom of speech, but this wasn’t an attack on freedom of speech. Charlie Hebdo was completely free to publish what it wanted; the attack on freedom of speech comes now, in the aftermath, as those in power use it as an excuse to control us more deeply than ever, and people start to wonder if what they published was worth the risk.

Obviously the editor-in-chief thought so: he said so explicitly. Maybe the cleaner at his offices felt differently, or the policeman guarding him; both of them died as well. So far I’m still with Voltaire: I may not agree with what Charlie Hebdo said (or the way they said it), but I will defend to the death their right to say it. Yet I’m also with Holmes: Charlie Hebdo’s right to swing its fist ends where my nose begins – and I’m prepared to extend my Holmes a bit further.

You can swing your fist in whatever direction you want, but – for the purposes of satire, at least, for the purposes of ridicule – there’s a common expression in comedy circles: don’t punch down. Punch up, by all means: target the powerful, and the rich, and the famous; but don’t target those who don’t have the same cultural resources with which to defend themselves. I wouldn’t say this is an absolute rule, but it’s a useful rule of thumb.

So why did Charlie Hebdo punch down? Because Charlie Hebdo didn’t even realise they were punching down. They saw ISLAM in big block capitals, MILITANT ISLAM that went against the very grain of Charlie Hebdo’s secular origins in the republic; they saw the imminent threat that’s been painted by our governments and our media, the threat that is not just at our doorsteps, but cleaning our kitchens, or stirring in our banlieues.

From that perspective, ISLAM looks more powerful than it is (which is exactly what MILITANT ISLAM wants you to think) so you think you’re punching up: but of course context is everything. Context doesn’t excuse, but it does explain. Context is the detail of the picture, rather than the broad brush strokes. Context reaches into the past to pull out reasons by the root. Context spreads its arms to avoid arrogance and embrace ambiguity. Context forces us to think, and think again.

What else happened on the day that Charlie Hebdo was murdered? Well, we don’t know the exact demographics of French prisons, because the French government doesn’t keep those statistics; but if they’re the same as they were in 2008, then the prison population in France is about 60-70% Muslim. That doesn’t excuse Charlie Hebdo’s murderers; but if you don’t think it’s relevant, then you’re going to be continually surprised by the real world.

headlineImage.adapt.1460.high.1420993506639What else happened on the day that Charlie Hebdo was murdered? In Nigeria, Boko Haram finished its battle for the town of Baga: the town was burnt to the ground hundreds, maybe thousands, of its in residents killed, and the rest fled. Nobody will commemorate them outside French embassies across Europe. That doesn’t make the murder of Charlie Hebdo any less shocking; but if you aren’t equally moved by Nigerian deaths, perhaps you should ask why.

What else happened on the day that Charlie Hebdo was murdered? In 1957, the French Army took over the Battle of Algiers from the civilian police force. The battle eventually cost the lives of up to 3000 Algerians, with another 3000 missing, and the date marked the point at which torture became institutionalised in that early war on terror. That has nothing to do with the muder of Charlie Hebdo, of course, unless you think that the history of French colonialism has no relevance to power and powerlessness.


Was it MUSLIMS who killed those people in Paris? Are such broad brushstrokes very useful in understanding the situation, or remedying the problem? Or are they a convenient way of avoiding looking at the context, of wondering whether this is an imperial hangover for which there may be no cure? Invisible Arabs, like those in French prisons; invisible Muslims, like those who disappeared during the Algerian war of independence; invisible Others, like the Nigerian dead of Baga.

In some small way, I’m more afraid of publishing this than I am of criticising Islam. The Empire doesn’t treat its dissidents well, but it does it quietly; at least Al-Qaeda makes it abundantly clear what your punishment will be. If I don’t express enough solidarity, then the internet will be outraged, because context-free outrage is the currency of the internet. Still: I went to the demonstration last night in my city, stood by the signs that said Je Suis Charlie, and worried about freedom of speech.

Nothing excuses these murders. No reason is good enough for these murders. No history justifies these murders. No religion justifies these murders. No politics justifies these murders. Yet the murders had their reasons, their histories, their religion, their politics. It’s too easy to dismiss them, and it’s too easy to turn on ourselves, and we need to avoid taking those easy routes. Maybe Je Suis Charlie, but Nous Sommes Pas Charlie; and that’s our political problem, not theirs.


Horse Phantoms

A couple of months ago, I published Four Species of Ghost, a knife crime on the body of psycho-geography, in Kamenzind Beograd Issue 5. Four Species is partly an homage to Iain Sinclair, whose work has been hugely influential in how I understand the city as a work of literature, and that’s probably a little too obvious.

It’s hard to escape the shadows of your literary influences, but it’s harder to escape the consequences of poor urban planning. As you’ll gather if you read the article, Savamala is a mess; trapped between a rising river and a falling hill, it’s suffered from decades of neglect that no amount of artisanal bike repair shops can fix.

Walking in Savamala now, I feel the hot breath of the Four Horseman of Gentrication upon my neck; and first among the Horsemen is Beograd na Vodi (warning: link may contain satire), the biggest experiment in social engineering ever attempted in the Balkans. Savamala Future doesn’t look anything like how I described it in my article: in fact it looks worse.

Beograd na Vodi

This is not my beautiful house.

Don’t believe me? Take a look at the billboard outside the main train station, a vision of Belgrade that looks like it’s just flown in from Dubai, and isn’t planning on staying too long. I’m sure it would be a great place to visit, but you probably wouldn’t want to live there – but then I’m not sure that the government really intends people to live there.

Large-scale urban development projects such as this speak the language of the public good, and they spend public money; but the real ambition is to attract outside investment (in this case, more Emirati money rolling into town). The problem is that the only way to do this is to wipe out what came before: the rich and ragged history of one of the oldest parts of the city.

One good thing has come out of Beograd na Vodi so far – the renovation of Geozavod, into something that looks a bit faux but is finally being put to use. Of course, that made my article outdated before it was even published; but that’s the great thing about psycho-geography. It’s not about what’s there right now, but what was and what will be. Savamala survives, for now.

Writing off Iraq

This month the White Review published my non fiction piece Every Night is like a Disco: Iraq, 2003, which only took me 11 years to write. I had it in notes, in my head, for that whole time; I always intended to write something, especially after Granta published Fly Away Home: Iraq, 1998. So why did it take so long?

I’d been writing about my working life ever since I started, cribbing from my notes, emails, reports, and anything else I could get my hands on. I’d published a couple of pieces in small magazines, and was thinking (optimistically) about pitching a full-length book, to be titled “Life in the Colony”.

After Iraq – well, after Iraq, three things happened. One was that I was filled with remorse – an attentuated strand of survivor’s guilt – and I felt that I would be trading on the lives of others. I hadn’t earned that right, not by a long shot, and the best thing to do would be to shut up. This feeling continues today, and publishing Disco was quite difficult for me.

Second was that I was angry at the humanitarian sector, and particularly the United Nations, for the abject failure of judgement that was Iraq. After working in Afghanistan, I didn’t expect much from the US and UK governments, but I had hoped that the UN had learned something. They hadn’t, and I’m not sure they’ve learned it now.

Third was that there was a snowstorm of books after Iraq, often from people who’d been in far worse situations than I, or had done far better research and reporting. Iraq became a coin in the reputational currency of the Beltway and beyond, to be traded for power and (often) sex; and I didn’t want any part of that. Aid memoirs? I felt like I needed to grow up.

The memory wouldn’t go away. It was a turning point in how I approached my work, in how I viewed the entire sector. It represented something more for me than just another story in the collection, and I never forgot those people along the airport road. Here was Iraq under a microscope; or more likely Iraq under a magnifying glass, burnt by the sun.

So this summer I sat down and wrote it in a couple of days. Long gestation, quick birth; I guess it had been writing itself in the back of my head. It was an awkward child – too long for short stories, too short for longform. I had a vague idea that The White Review, whose tastes trend extremely catholic, might want to publish it, maybe? And they did.

Now it’s out, I feel a sense of relief. I can get on with writing other things, maybe even from that same period: I have half a piece from Liberia, that same year, “A Day Late and a Dollar Short”, but who knows if I’ll ever finish it? That’s the problem with writing: you can make all the plans you want, but sometimes the words follow their own schedule.

Your Berlin is a Wonderland

Thanks to the Wonderland crew, I spent last week in Berlin touring a cross-section of the city’s most Berlin-ish urban developments. Berlin isn’t the city it was – the relentless teeth of global capital and recession angst have really started to bite in the last few years – but it retains some core values that didn’t fall with the Wall.

Those “core values” include a measured pragmatism that acknowledges that financial benefit isn’t necessarily a social good. The urban landscape (London, I’m looking at you) is increasingly scarred by borderline psychopaths who want to re-shape the city into a playground for themselves and others like them. Berlin has a stronger immune system than that.


A good example of this is a financing policy in which tenders are not granted to the lowest bid; instead a cost is set by the authorities and the most interesting proposal given the tender. Finding the money for “interesting” proposals is another thing entirely, of course – the bizarre role of Swiss pension funds in supporting alternative culture should not be under-stated.

Now obviously I’m a sucker for socially-conscious + environmentally-friendly buildings – the reason I was at Wonderlab was to represent Pametnija Zgrada – but I also retain my core personality trait of being impossible to please. My question is a simple one: does it work? That is: does it live in the real world, rather than in an urbanist’s wet dream?


The Berlin answer is: yes, but only just. Nearly all the projects we visited had benefited from investment that covered site acquisition at reasonable interest rates, so they could start making profit quickly. The inevitable & painful dilemma: alternative cultural approaches are being subsidised by extremely mainstream financial tools (and not just pension funds).

I liked all of the projects, especially ExRotaprint, which seems to get the balance Goldilocks right. Yet it was clear that these types of project don’t necessarily translate well to other contexts – definitely not to other countries (like Serbia) but probably not even to other cities in Germany. We can take inspiration from them, but probably not replicate them.


We spent the rest of our days exchanging experiences between exchanging experiences between projects, a crazy-quilt mix of different approaches to the same problem: the disconnect between people and place. As the world goes urban, this isn’t just an excuse for architects to hold workshops, but the concrete problem that’s going to shape the century.

This is mirrored by the online problem of privacy; to a large extent these twin problems are urban problems. One thing that was noticeable by its absence, then, was any discussion about the smart city – yet Berlin is geared to be a tech hub, and all of the workshop participants were all heavy technology users. I guess that discussion will have to wait until next time.


We were hosted by DAZ, whose internet leaves a lot to be desired, but whose hospitality was excellent.


(Apologies for the fully intentional John Mayer reference in title.)

Picture accompanying The Humanitarian Future

Danas Mirabilis

And so it came to pass that Aeon and Nature both published one of my pieces on the same day.

The Humanitarian Future was difficult. The original article was trying to be three things: a potted history of the modern humanitarian sector, an outline of what needs to change for ‘humanitarianism’ to survive in the now-times, and a brief personal account of my work. My editor Brigid Hains took one look at that and realised it wasn’t going to work. The final version doesn’t capture everything that I wanted to say, but, contrary to what the internet wants you to believe, it doesn’t have to. This is a small part of a wider discussion, across many platforms and in many fora. Go read.

The tiger waiting on the shore, by contrast, was easy. I had been enjoying the Nature Futures podcast for a few months, and wondered if they had an open submissions policy. They did, so I wrote the story in less than a day, and sent it off with no expectations. Editor Colin Sullivan accepted it immediately, and it’s printed more or less as I originally wrote it. Writers’ lesson for the day: you never know where you might find a platform for your work. The story is a science fiction / horror / family drama – a bit more obscure than most of the stories that Nature publishes, and it fills me with a feeling I can’t quite describe.