This week Chatham House hosted an event for the co-founders of UNICEF Innovation, Chris Fabian and Erica Kochi. The group is responsible for identifying and scaling technology for development to support UNICEF's programming objectives.
For instance with improving numbers and accountability on birth registrations in Nigeria with simple infrastructure to collect and distribute data. There were also some provocative examples of building solutions at the point of need, as opposed to in a lab well removed. A point made recently by SciDev.Net's columnist Henrietta Miers.
There is no doubt they have had some eye-catching successes.
But it was their failures that really caught my attention.
The team claims that it has a 95 per cent failure rate. This does not scare the people in the team. They see it as a natural consequence of working with innovation and indeed of working with complex systems, like the environments in which children live and learn.
The problem, they say, is that the rest of the development community is not more open to failure.
Fabian, who has worked in the private sector in Egypt and Tanzania, says it strikes him as strange that agencies would prepare projects and budgets claiming to know not only what they will be doing in five years' time but also how the world will be responding to their work.
This is a bracing argument. And one also made by complex adaptive-systems theorists.
The problem, as an audience member from Save the Children pointed out, is how one strikes a balance between responsible planning and this non-linear reality. This is particularly the concern for anyone who is receiving funding in an environment of austerity.
The UNICEF team talked about a few things they are doing to change the DNA of aid so it was more supportive of innovation.
First, admit that we fail. Perhaps the most accessible was what they called 'Failure Friday'. The team reflects on their week and shares one success, one failure, and one thing that confused them. The idea is that you institutionalise a culture of openness and learning.
The team also said that the problem with innovation in development planning is that we tend to think big. So by the time we recognise something is not working, we have already invested a great deal of money, time and reputation. The play pumps are a good example, the project behind them had reportedly raised US$1 million by the time it was clear they would not work. Instead, we need to plan to fail fast and fail cheap as we go.
The team also talks about having a portfolio of options, areas of work that they want to monitor and test. This is not the same as having a five-year strategy but more like a set of preoccupations that you want to invest in.
Finally, they are looking at changing the structural relationship between development and the private and technology sectors. So they have started a 'venture capital-like' project within UNICEF and will be bringing Silicon Valley incubator firms to UNICEF funding meetings. Reflecting an awareness that development goals need more than the development industry.
It might be easier for UNICEF to do this than many other agencies. They have an enormous unrestricted funding base through their individual donation campaigns and they work in an area of such strong universal appeal - providing tangible help to needy children - that they have enviable resilience and space to take risks.
But the truth is any development agency would need a better excuse than UNICEF's funding mix not to start 'Failure Friday'.