Recently I was lucky to be able to attend not one, but two great events in Montreal.
The first event spanned two days and was hosted by the Canadian Open Data Initiative. In the form of a Codeathon, the Open Data for Development events brought decision makers and coders (geeks and wonks) from all over the world together to make a headstart in tackling some of the challenges faced within the open data community at the moment.
We split into a number of different groups to look at an array of different ideas and subjects, ranging from International Aid Transparency Initiative (IATI) data, open contracting, results tracking and reporting, and domestic reporting through to dashboards and visualisation tools for consuming these data sets.
With a good understanding of IATI data and business flows, I elected to join the team working on following the money chain through the aid flow and visualising this. It’s something we’ve been working on in Akvo for some time, so it seemed a very logical fit for my skills and knowledge.
We determined there were two distinct parts to our challenge – gathering the data and visualising the data. Not being a core coder, I opted for the former and set to work hunting down some IATI files and trying to find some chains.
We quickly discovered that one of the biggest challenges facing data analysts and anyone else trying to consume IATI data is that the level of quality and completeness of most IATI data sets out there is not really sufficient to make many high level conclusions.
Working with John Adams and Ben Webb from the Department for International Development (DFID) and with a huge amount of insider information provided by Steven Flower from the IATI Secretariat, we were able to pull out a chain of four organisations, beginning with DFID – the UK Ministry of Foreign Affairs. This went on to a UK based fund manager Triple Line who, due to the nature of their organisation, held two levels of IATI data internally. (Well, they did by the time Steven made some live correctional updates to their data, having recently worked with them to publish their first IATI set.) Following after was War on Want, who had a very well formed and complete IATI publishing effort. Because their file was published before that of War on Want, it did not directly reference the Triple Line funds, as would be the ideal case. Instead it actually referenced the funds directly from DFID. With the internal knowledge of the funding streams within the team, we were able to determine the linkages, and form the chain. Lastly we saw funds being disbursed to local partners in both Honduras and Kenya, where the chain ended.
At the same time we were also able to scrape the raw IATI files and find several other examples of chains of this size originating from DFID and going through Development Initiatives.
It was an extremely informative and useful exercise, and this provided the visualisation team with a good clean set of data to work with to be able to create a simple app to navigate between the levels and pull out key information about the aid chain. The app is still in development and will continue to be worked on by the team and the open data community. It’s open source and on Github already, so it’s available for anyone to contribute. You can also check out the Hackpad for the session here.
For me this was a great learning experience, not really code related, but more of the state of open data within IATI.
Firstly, the data has to be good quality – most of the existing publishers are doing a fantastic job, but key missing data elements, in particular the links between organisations in the aid flow, seriously impact the potential value of the data in itself. This can somewhat be overcome by working with people within the sector, who have the knowledge to be able to fill in the blanks, but this is not ideal and it means that making any conclusions is somewhat limited to events such as this one. In fact, without the linked data, each published file is pretty much only useful within itself, as the double counting issue means it cannot safely be combined with any other data set.
Secondly, the absence of results within almost all of the data sets means that while we can follow the money trail, we are really still blind to see what these funds are actually achieving. I think we’re a long way off being able to breakdown this data in granular detail, but the inclusion of some sort of results would be a huge improvement in starting to account for aid funds and see what we might be achieving with the money we distribute globally.
The second event I attended was the International Aid Transparency Initiative Technical Advisory Group, which I’ve written about in detail here.
Adrian Collier is the product manager for Akvo RSR, based in Amsterdam.