Before we begin…
💥💥 Our window for expressions of interest for development partners (donors or programme implementers) is open. If you interested in applying internet of things (IoT) for a data revolution on your work, then let us know: https://forms.gle/GxFcGfTAi3zkd1br7
Our deadline is Monday 9 Sept (23:59 UK time) 💥💥
🎙️🎙️ Interested in finding out more? We are hosting a webinar to share and discuss the findings of our marketplace of ideas and our work on this pilot. The webinar is on Thursday 12 Sept (12:00 UK time)
Register to attend here: https://zoom.us/webinar/register/WN_zTIM9sRAQlq1c-P9q2TASw 🎙️🎙️
By way of introduction…
On Frontier Technology Livestreaming (our UK Dept. for International Development programme exploring frontier technologies to solve the biggest challenges in development), we work with a huge range of ideas.
Our technology use cases vary in which frontier tech they are using, where they are operating, and also what stage of maturity they come to us. Sometimes, DFID Pioneers come to us with a proposed solution, early prototypes, and evidence of traction. Other times, they come to us with a wicked problem and an idea. The Internet of Things (IoT) for a Data Revolution pilot certainly sits in the second camp.
As a first step, we convened a group of experts via an open invite to a marketplace of ideas, to help us better frame the problem and think critically and creatively about the potential solution.
Our initial hypothesis is that:
If… we embedded Internet of Things (IoT) connected buttons, sensors or other devices in development programmes, then… we could collect real time, accurate and consistent data that would enhance accountability and decision making.
We know data can often take too long to gather, be inconsistent across a programme and redirect money and other resources away from doing the actual work. This reduces both the accountability of these programmes, the quality and speed of decision making, and ultimately the desired impact.
Below are nine key insights from our marketplace of ideas — reframed as questions that we will keep at front of mind as our work evolves.
Our 9 key insights
1. How do we complement (rather than duplicate) what is already out there?
A central insight we gained was just how much activity is happening already using IoT for measuring impact — from monitoring air quality or water contamination, disease prevalence, to supply chain and inventory tracking.
We want to learn from and work with these existing projects. At the same time, we believe there is a gap for us to generate new evidence. This gap could be the sector (e.g. humanitarian response) — or type of data (e.g. beneficiary feedback)
2. How do we design for multiple users?
We see three ‘users’ (i.e. people who would engage with) for this tech. One is the beneficiary of a development programme. A second are people on the ground who monitor outcomes right now. And a third are those who make decisions as a result of the data.
For this work to be a success, we need to build with all three users in mind.
3. How can we think through the design challenges on the ground?
Would the use of IoT take away from local relationships? Would it remove the ability of people working on and benefitting from the development programme to engage with data as they collect it? Would it take away from local ownership of a programme?
Questions like these need to be considered proactively for us to create a wider, net-positive impact.
Additionally, how do we design against perverse incentives? For example, if IoT is tracking volume of goods delivered to a place through manually pressing a button, how do we guard against it being pressed too much (to overstate results), or too little (to get more delivery)?
How do we ensure the tech does not introduce bias to the data? For example, if IoT buttons are tracking use of HIV self-testing kits, will only ‘enthusiastic’ users report back usage, giving us false positives?
Finally, how do we ensure that this isn’t another burden for people on the ground?
Using sensors that don’t require any input, or even existing data collection mechanisms and using the data they generate better, might be one solution to these challenges — albeit one that introduces its own limitations. To give an example discussed in the marketplace, TFL don’t have any buttons on usage — they get this by looking at oyster tap-ins and outs.
4. How do we think through the design challenges for users of the data?
There was widespread agreement in the room that data today is not efficient enough to inform programming. Either there’s too much data, or it’s not available quickly enough to be used to make evidence based decisions.
One participant talked about a refugee programme that used IoT to predict migration patterns. It was predicting two days in advance, but the data took two days to arrive, so it became real time rather than predictive and therefore did not influence the programme.
This was our ‘meta problem’ — the risk that data generated from IoT simply adds to this pain, and makes worse a tendency to not use data at all in decision making. How can the data start the important, often difficult conversations that lead to change on a programme? And how can the data be useful to all types of influencers, from donors to local officials to politicians?
We discussed whether we need to make sure there is one automated indicator, that everyone agrees is our one metric that matters.
5. At what scale of data is technology justified?
We never throw technology at a problem for its own sake. If you only need a small amount of data, people might be better. At what volume or type of data does tech become value additive rather than a distraction?
6. Which programmes stand to gain the most?
Different programmes have different M&E requirements. What types of data, and what volume of requirements, are optimal for using IoT? In addition to this, and just as important, is targeting programmes that have the flexibility to change as a result of data.
One idea from our marketplace that demonstrates this is linking payments to the results that are captured using IoT sensors or buttons, for programmes that could or do operate on this basis. Could payment then also be delivered in an automated way?
7. How do we mitigate privacy and security considerations?
At its best, IoT has the potential to enhance anonymity and privacy, as data can be collected without human interaction. This is great — and we’ll need to make sure it’s not illusory. Any privacy (or security) breaches risk undermining trust in this and other technologies as a whole.
The technologists in the room and online also emphasised privacy and security as a legacy issue. Even after the data has been collected and acted on, how do we make sure it is kept secure and not compromised?
8. Will the tech be able to operate in or adapt to frontier contexts?
For the pilot to be a success, the technology will have to prove itself in a range of environments — extreme heat, wetness, and so on. If the tech was originally designed for a stable, indoor environment we will have to show that it can adapt.
Connectivity was also discussed. This relates to the coverage of internet access in different countries needed to transmit data; but also to how much this data costs and whether there are any political or regulatory restrictions.
9. To what extent should and can our data and systems integrate and be inter-operable?
Everyone in the room agreed siloed data was a problem. Could we integrate into existing systems and better leverage what already exists, rather than creating parallel systems? Or how do we make the data accessible and usable to others?
In parting, all participants recognised the opportunity while discussing the constraints and nuances. As with all tech innovations, we agreed that the answer isn’t a categorical “yes this will work” or “no this won’t work”. Rather, that this idea can add value and complement existing systems if we design and execute mindfully.