After reading an earlier post on the role of  open data in development,Tobias Denskus calls for more critical debates around open aid data. His article is really worth reading first; he draws our attention to four issues in the current state of aid transparency:

1. Everybody in the aid sector seems to be on board with it. This suggests that the ideas of aid data and transparency are too broad, no longer innovative or don’t challenge the fundamentals of development. Where’s the debate around the small print?
2. Many international development decisions are political: better aid data leading to more convincing arguments is unlikely to change this.
3. Data and numbers can only have a limited effect on the fundamental issues of how development “thinks, reports or learns” and aid transparency favours the quantitative over the qualitative.
4. Most aid data is old data and its usefulness beyond research is limited. Complex bureaucracies might struggle to respond to outsiders engaging with their data and even if we have more current data, how much “real-time” influence is actually desirable?

What do we mean by “aid transparency”?

Everybody and their mother seems to think aid transparency is a good idea – inevitably, it means different things to different people, all depending on the occasion.

I imagine this will continue as other public transparency initiatives gain popularity, with the details and intentions of each blurring into the kinds of wooly development rhetoric we’ve all seen. When you add to it the more controversial examples of public transparency that come along and dunk the entire endeavour into disrepute, it is vital to be clear on what we’re talking about when it comes to aid transparency.

Taking a simplistic view of development information, I think there are interesting, interconnected “aid transparency” activities in each of these three areas:

1. What and where are the opportunities to improve lives?

The World Development Indicators are freely available as part of the World Bank’s Open Data Portal and offer measures for 209 countries with data from 1960 up to 2009. Local communities and initiatives like UN Global Pulse are using new technologies and approaches to “reduce by an order of magnitude the delay in availability of actionable information on household-level impacts and vulnerabilities” We’re gradually getting a clearer, more timely picture of humanitarian and development needs and opportunities.

2. What’s being done about them?

Sam Moon and Tim Williamson define aid transparency as the ”Comprehensive availability and accessibility of aid flow information in a timely, systematic and comparable manner that allows public participation in government accountability.” With the International Aid Transparency Initiative (IATI) – there’s now broad agreement among donors on publishing the details of “who is doing what, where” that represent around 75% of the approximately USD 120 billion in global aid. This is open data on past, present and planned aid activities.

3. Is it working?

The UK is emphasising the sharing of measurable results as part of its aid transparency agenda (Nancy Birdsall’s thoughts on this are worth a read)  and I expect other donors to do more of the same.  There are good  examples of community and beneficiary feedback systems in Akvo and part of Global Giving’s work which offer both quantiative and qualitative data on the impact of aid funded work.

With this broader view of aid information in mind, let’s consider the issues Tobias raises.

Is “aid transparency” becoming development jargon?

In the West, the driving ideas behind government transparency have been knocking around since the Enlightenment; by this measure, the current focus on aid transparency is overdue rather than innovative. Having said that, I think the idea of open data in development is innovative and I don’t believe most people who “like” aid transparency fully appreciate what open data means.

For many, aid transparency boils down to “look at this table in our annual report” – if we didn’t move on from this, the impact on the fundamentals of development would be negligible. Since September 2008, IATI has gradually established consensus amongst donors on exactly what data to publish, common definitions to make data comparable and an electronic (XML) standard to make it easily shareable.

So, at least when it comes to “who’s doing what, where”, the data piece of aid transparency is well defined, clearly bounded and boils down to organisations agreeing to publish information (that they already have) in a standard form so different people can easily use it.

Development decisions can be political, will aid data change this?

There is a strange tendency among people working in development, faced with political decisions which make aid less effective, to assume that these political forces are beyond our control and influence.  Yes: development decisions can be political; access to aid data is one thing which maychange the politics.

There’s a growing body of evidence (see Rosemary McGee & John Gavent’s IDS paper) demonstrating how transparency impacts things like the quality of service delivery. There’s not as much discussion around overcoming the political hurdles in building a working feedback loop from those who ultimately benefit from aid to those who make the policy decisions.

Jorg Faust’s paper “Do Less Transparent Donor Countries Allocate Aid Differently?“ shines some light on these issues. His hypothesis is that “higher levels of political transparency – the ease with which the public can monitor the government – are conducive for limiting the impact of special interests on policy-making.” He finds that political transparency in donor countries has a significant impact on how they allocate their resources for development assistance. In the study, more than half the variation between donors in how well they allocate aid can be explained by how transparent they are. From a policy perspective, this study confirms that donor transparency leads to more effective aid. It is an example of how access to data can change the political forces.

There is potential for behaviour change when political decisions are made in an environment where they are publicly documented and those making them can be held to account. At the very least, with open aid data, there will be more people researching and advocating for changes in development (including those directly benefitting from it) who can make a better informed, politically sophisticated case.

Are data and numbers enough to make development better?

We’ve argued that aid transparency is necessary but not sufficient for making aid more effective. To unpack an aspect I find interesting, consider Tim O’Reilly’s vision of “Government as a Platform”. The premise is that by opening up its data, the state is able to improve the way that problems are dealt with at a local, national and international level. The state acts as “a convener and an enabler” of the civic action that can take place when modern internet technologies are combined with government-provided data. We now know that opening up government data multiples its economic and social value; I see no reason why this shouldn’t hold true in development.

Another parallel here is with access to mobile phones.  Does access to mobile phones, by itself, reduce poverty? No. Can access to mobile phones enable people to change their lives, whether through mobile banking, making agricultural markets work better, or better election monitoring? Yes. Access to data won’t, by itself, make development work better, but it opens up new ways in which people can help themselves.

The transparency agenda isn’t about using numbers to directly redesign the aid system (along with how it “thinks, reports or learns”). Instead, “it seeks to change the dynamics of the system to make it more responsive and more likely to converge by itself on solutions which better serve poor people in developing countries” as Owen notes in his response to Shanta Deverajan’s post on “Development 3.0”

In other words, information on aid transparency – and in particular, a better feedback loop between beneficiaries and donors – is something we need to build upon because it will enable ongoing, incremental improvements to aid as a whole.

How can data be used to change funding before it is spent?

At the moment, aid information is published too slowly and the majority of it is historic. On this theme, Robert Kirkpatrick of UN Global Pulse gave an excellent talk at this year’s International Conference on Crisis Mapping (ICCM) where he says, “When it comes to global crises we’re investing based on data that is 2-5 years old”. So, our picture of the world is out of date and right now, the typical data on “who’s doing what, where” in the aid sector is at least 12 months old, if its available at all.

Yet donors do have information about their plans: the typical project cycle includes stages of budgeting, planning, and analysis before money is disbursed, and there is no reason why more of this information should not be publicly available and subject to scrutiny.  The IATI mechanism allows this to happen – enabling donors to live up to their commitment, given in Accra, to publish forward looking aid information.

This will improve donor coordination, and it gives recipient countries an opportunity to shape the funds heading their way.  I also see this as part of the “Open Data, Open Knowledge, Open Solutions” idea that the World Bank has been promoting. It’s a step away from the slow-moving Washington consensus of development towards something more responsive, representative and fluid.

It remains to be seen how the dynamics of complex bureaucracies respond to this.  It seems likely that the knowledge that this information will be public will be self-discplining, encouraging more thorough and evidence-based analysis of planned aid programmes; and aid agencies will adapt, over time, to become more responsive to the expertise and views of outsiders.

Open aid data is just a tool.

Tobias raises some worthwhile points in his original post; I’ve offered a perspective that I hope strengthens the argument for open aid data but the call for more critical debate still stands. The work aidinfo and others have done to date has focussed on building the case for transparency in aid, getting the data out there and showing what’s possible. Now that the data’s coming, the next phase of aidinfo’s work is about the day-to-day application of this information, and in supporting the institutions and individuals who want to use it.

The goal is to improve lives: open data is just a tool, and how we use it to make aid more effective is really worth thinking about.

At the Open Government Data (OGD) camp this year, David Eaves, a Canadian open data and public policy activist, gave an excellent keynote talk highlighting some of the challenges that lie ahead in the world of OGD.

He made some interesting points which I’ll share and add to with a perspective from international development. In short: building international development open data portals will help to create a more effective, data-literate aid sector; openness fosters a culture of learning and improvement, and if you’re doing something interesting with data, talk about it.

We built libraries to help citizens become literate

Across the world, an increasing amount of government data is being released through online portals. Only last week, the UK government released details of all spending over £25,000. In due course, with initiatives such as data.worldbank.org, AidDataand IATI leading the way,  good quality data on international development needs, actions and outcomes will become readily available.

David’s analogy is important: just as libraries weren’t built for people who were already literate, open data portals aren’t just for a “small elite of hackers and policy wonks”. When the western world got busy building libraries in the 19th and early 20th centuries, they were built on the belief that they would act as hubs to help citizens become literate, and in doing so benefit society as a whole.

In the world of international development, opening up data and building portals that offer convenient access for users may seem like nerdy technical endeavours now, but these are the first steps towards a more effective, data-literate development sector. Fundraisers, researchers, policy experts, administrators, consultants, field workers, local staff, community activists and the individuals who are directly affected by aid will benefit from better access to information they can use.

We need a patch culture

Open data leads to behavioural and cultural changes. At the OGD camp last week, during a session on measuring the impact of open data, one participant spoke of a conversation between two British MPs. The conversation went along the lines of (paraphrasing): MP1: “Does opening up this data make any difference?” MP2: “Have you claimed less on your expenses since the parliamentary expenses scandal?”  MP2: “Yes, of course…” MP1: “Well then yes, opening up government data does make a difference.”

David’s point is a little broader. Governments, development agencies and donors are constantly showcasing their success: they want to be seen as getting things right the first time, and so can end up being rather secretive. As a result, people (and the media) jump to the worst conclusions when they do find something out – since institutions don’t actively disclose what they’re doing, when someone does spot a mistake, the reaction is “Gotcha! You’re clearly evil and were trying to hide that.”

A culture of openness can calm things down. In the world of open source software, if someone finds a problem with the program, they file a bug report, the developer is grateful (and perhaps slightly embarrassed), a “patch” or a fix to the code is released and the result is better software for everyone. Open data would encourage this kind of patch culture in international development – a mistake or a problem if spotted, could be seen as an opportunity to learn and improve rather than front page news.

Let’s balance advocacy with action

The biggest user of open data on international development will be the international development sector itself. Sure, journalists and the general public will make some use of the information but those who stand to gain the most are donors, civil society organisations and recipient country governments. Over the last two years, a lot of political effort and advocacy work has gone into getting aid institutions to publish data in a standard format. This needs to be rewarded with tangible examples of what can now be done. The supply of information is assured if there’s a demonstrated demand.

Competitions like Apps for Development are a good way of encouraging use of data but I’d encourage anybody who is thinking or is already using open international development data to shout about it! We’d be happy to talk more and feature interesting examples on our website.

With a good set of examples, patterns and common practices, the use of data will become embedded in development practice. Better information will make aid more effective: data portals improve data literacy, an open culture promotes learning, and using information to add value in the aid sector will ensure data remains open and will promote the release of more.

(This post originally appears on the aidinfo website)

Competitions have been a constant feature of the open data movement.

From the early days of  the UK government’s Show Us a Better Way challenge to the World Bank’s recent Apps for Developmentcontest, the model of these competitions has evolved with the availability data and we’re starting to see some online platforms that help to commoditise the tools and processes involved.

Most open government data competitions now look something like: “here is some data; here are the rules; these are the prizes; do something neat” Outside of open government data, there are plenty of examples of data problems that have been solved by making the data public and letting a subtle combination of incentives and game mechanics do the rest – see the Netflix Prize for a good example.

With this in mind, I’ve come across 3 interesting, slightly different platforms:

ChallengePost

As a “Marketplace for Challenges” ChallengePost invites users to “Find, Solve and Post” challenges. The World Bank’s Apps for Development contest is running on the ChallengePost platform as are several other corporate and governmental contests. The site incorporates a “social leader board” and allows others to pledge support (money or just appreciation) for a particular contest.

Kaggle

Kaggle describes itself as a “a platform for data prediction competitions that allows organizations to post their data and have it scrutinized by the world’s best data scientists.” and supports two competition types: predicting the past and predicting the future. The team behind it have a finance background but the platform seems reasonably suited to any statistical / analytical work.

Hypios

On hypios.com you’re using their “secure marketplace that facilitates problem solving by connecting Seekers to a global community of Solvers”. In contrast to the two platforms mentioned above, there’s much less open data here – it’s more about letting organisations securely outsource research problems, plenty of which are data-driven.

So what’s this got to do with international development?

Earlier in October Robert Zoellick described the Bank’s Open Data, Open Knowledge and Open Solutions initiative. This kind of thinking needs to be more deeply embedded in the world of foreign aid. Although only part of the Bank’s initiative – I think using data competitions was initially worth it for international development organisations wanting to get people interested in their data.

Right now, I generally think that small, specific contests are worthwhile. However, the overhead of curating data, running, promoting and supporting a large contest should be spent on developing capacity and long-term relationships with the already willing people who want to use newly available data that’s emerging as a result of programmes like the International Aid Transparency Initiative.

I think the online platforms described above can help organisations design and promote specific challenges with less effort and to build valuable external relationships. I would love to hear about any more.

(This post originally appears on the aidinfo website)

The Governance and Social Development Resource Centre (GSDRC) published a research report on participatory monitoring  and evaluation (PM&E), and beneficiary feedback for the Aid Effectiveness Team at DFID.

You can visit GSDRC’s website to download the report.

The report responds to the query:

“Please identify the existing literature on participatory monitoring and evaluation, with a particular emphasis on gaining wide-ranging beneficiary feedback. Comment on the coverage, scalability, risks, benefits and applicability”

After offering an overview and literature review on PM&E, the report highlights  examples of technologies being used to facilitate beneficiary feedback and notes that:

Although ICTs present opportunities for scaling up beneficiary feedback mechanisms, the advantages of these tools are largely unproven. It is important that ICTs are not viewed as a ‘magic bullet’ and that they are carefully tailored to context.

Owen raises a related “biomimetics for development” idea in a presentation arguing that we should not try to design a better world: we should concentrate on building better feedback loops. Aidinfo have argued elsewhere that the beneficiary feedback loop, alongside donor transparency, is a key part of the aid effectiveness equation.

We’re keen to learn about other good examples and models of beneficiary feedback – do get in touch if you’ve got any to add.

(This post originally appears on the aidinfo website)

A question landed in my inbox yesterday and it got me thinking:

We want to make a very simple map showing the top destination of aid funds, regional donors and NGOs… I was wondering if you knew of any simple and free open-source tools that can allow this sort of mapping.

International Aid is Complicated

Visual displays of information can be eye-catching, easier to remember and a concise way of communicating a complex set of data.

Now, the international aid architecture cannot be described as simple. The diagram above, taken from “Trends and Issues in Development Aid“, clearly shows this.

The arrows in the picture represent a flow of moneyand a flow of information.  Aidinfo works to change the way this information is published, shared and used because doing so will make aid more effective.

With initiatives like IATI, data.worldbank.org andAidData paving the way for more detailed aid information that is open, timely, findable and usable, you could soon find yourself able to use information visualisation to communicate insights about the world of international development that just weren’t possible before.

If you get it right.

More data’s coming but poor visualisation would make it useless

Edward Tufte is credited with coining the term “Chartjunk” which refers to visual elements in charts or graphics that are not necessary to comprehend the information being represented. I use the term more liberally to describe examples of poor visualisation of information that defeat the purpose of visualising it in the first place: they make it harder to understand the message.

The use of statistics in international development is generally no better than in other fields – let’s think about how to do better.

Literacy in data and design

Owen blogged about a great set of guides on how to use and understand statistics from the British House of Commons library. Before attempting anything else, I’d make sure you’re comfortable with that material.

Following this by reading Edward Tufte or Stephen Few will certainly put you on the right track, but for something a bit quicker, I’d recommend this short video of an ignite talk by self-confessed information visualisation enthusiast Matthias Shapiro. When designing a visualisation he suggests:

1. Ask a specific question
2. Gather your information
3. Apply a metric

Without a clear question in mind, the chance of producing something pointless greatly increases. Once you’ve got a question, focus on gathering only the information that’s relevant. Finally, take a piece of that data and apply it to an appropriate visual dimension (e.g. bigger numbers represented by larger circles)  Above all, tell a story. These seem to be key ingredients to a good visualisation.

Putting it into practice…?

So with that in mind, how did I reply to the initial question in my inbox?

The good thing about blogging is that you can say what you wish you’d thought of in the first place take a little time to reflect on an idea and then open it up to readers to see what they think. The questioner was looking for a tool that would let them map some aid information and I recommended looking at the Exhibit Framework which is part of the SIMILE project at MIT.

For a bit of fun I thought I’d use Exhibit to see what I could do in 60 minutes to show how much aid the ten largest country donors give annually. Click here or on the image to the right to get to the actual visualisation.

In short, it’s not bad but not particularly good either.

I wanted to show the relative size of overseas development assistance coming from the top 10 countries both on a map and in a table. In many ways, this information is better communicated using a bar chart but for the purposes of making a point, this map’s useful.

Some other thoughts:

• The map might give the false sense that the US gives as much as Europe combined – check the table to see that this is definitely not true.
• In the tables, the % of 2009 GNI figure is in many ways more interesting: 0.2% for the USA looks huge but 1.12% for Sweden looks tiny. The map doesn’t reflect this.
• Using billions instead of millions would improve readability.
• Comparing circle areas is harder than comparing bar heights.
• Adding more information in Europe would make the map too crowded at this zoom level.

Part of the problem was my lack of familiarity with Exhibit and the 60 minute time limit, but I think there are also elements of the tool that make it easy to forget or hard to apply good design principles.

I wanted to see what was possible for the non-expert to produce – I assume the original questioner doesn’t have a background in programming or graphic design but still wants to visually communicate a message. There are technology and design organisations doing excellent visualisation work for international development clients, but there are many cases where presenting data would make a huge difference, but the resources to bring someone in just aren’t there.

You can download the source data (taken from OECD via this blog) and all the Exhibit files to try it yourself.

Opportunities and Pitfalls

It’s great that we’re getting more data on international development along with free tools that can help visualise it.

However, after 60 minutes trying this out I’d say:

• Begin with basic data and design literacy and then worry about fancy stuff.
• Be aware of the limitations and constraints of the tools you’re using – some make it easy to ignore good design principles.
• Remember that there may well be better tools for the job - Maslow’s hammer isn’t just for hammers.

What tools and techniques and resources have you been using to visualise aid information recently?