076 – Data Systems for Low Digital Literacy

[00:00:00] Lucie: Hi and welcome to the IDEMS podcast. My name’s Lucie Hazelgrove Planel, I’m an anthropologist and Social Impact Scientist and I’m here today with David Stern, one of the founding directors of IDEMS.

Hi David.

[00:00:18] David: Hi Lucie, really nice to be chatting again.

[00:00:23] Lucie: Thanks for taking the time. So I’ve been working for a few years trying to think about database structures, and there has been very little progress, partly because I’ve been talking to the wrong people, and there hasn’t been a need. There hasn’t been a direct pressing need, I would say. There’s been a need, but it’s not a pressing need, perhaps. And so I wanted to take the time to explore this idea of a database more.

And I’m not talking about a general database. I’m thinking in particular of a working farm database.

[00:00:51] David: You’re thinking very specifically, we had an episode recently with Polly Noyce and she was mentioning Manor House and you recently had a trip to Kenya where you visited Manor House. So I am deducing that you’re thinking very explicitly about the work that’s been happening in the thinking, which has happened about trying to help Manor House to get data which serves it, which helps them to put that into a database, which will be of use and of value to them as an entity and an organisation.

[00:01:24] Lucie: Yeah, exactly. You saw through my sort of general database to, yes, Manor House Training Agricultural Centre. Because I’d always heard of it, since joining IDEMS I knew that there was this plan, and I know the different areas of the farm.

So for example, the livestock area or the crop area. I know that they collect data. A lot of their work is research work about what they currently actually do, their own practices. And this has been some of my questions, like how they collect data, what data they collect?

What I was aware of when I finally actually got to be able to talk to the director, David Mwangi, and his colleagues, he assembled, it wasn’t him actually, it was Amos, who assembled everybody together for us to have a discussion about their sort of interests, their needs about data and a potential database.

And I saw that what they were talking about was really, to me, it was big aims, a lot of work needed to be done before achieving all of that. So I was amazed, yeah, I was surprised and amazed by the level of what they were discussing compared then also to what they’re, what stage they’re at now. So what I had never understood previously is the lack of data that’s actually being collected.

So my understanding at the moment is that I was told that the livestock department is regularly, I think every day, collecting data on ODK forms on their phones about, I think, what feed different animals are given. And I haven’t seen these ODK forms, so I don’t know exactly what data is being collected, but I can imagine. But many of the other departments aren’t collecting data. Some are in between. Some are collecting data on paper. They have wonderful books, for example, I was looking at the, it was between the crop and the kitchen. I think it was the crops data, looking at what vegetables they were giving to the kitchen.

So every day, how many kilos of kale, how many kilos of other leafy vegetables. And I also heard that as part of this data collection system that’s currently being worked out on the ground there is that they have a complicated procedure. Especially, I’ve just mentioned the kitchen there, and this is the additional complication that it’s not, Manor House is not just a working farm, it’s also a training centre.

They have guests who come for a week long stay or something, who come to be trained, who come to explore agroecological solutions to farming practices, and the kitchen needs to provide food for them. And ideally that food often comes from the farm itself.

So there’s a really interesting circularity of data, or not only data, sorry, circularity of food, and produce, which is currently being done by some paper ticket form with receipts, and it sounds very complicated.

[00:04:15] David: Yes. So you said you’ve been hearing about this for years, and I’m afraid, this has never been a formalised project and there are already, even some open, there are digital tools out there to help working farms, which are used often in very high resource environments.

But part of what you are articulating here quite nicely is the fact that actually a big part of the problem isn’t the database. It is the fact that the people who are working on this, this is all the idea of collecting data, how you collect data, this is all very new. So there’s huge capacity building component that goes alongside this.

This has these training. It also, of course has students. And so it’s actually a student centre as well. And so actually getting something which has the right mix of things has to be custom built in some sense. And this is where that’s been the ongoing discussions for many years now. How can we help, at very low cost, we’d love to actually have the money to be able to put time and effort into this and build something which could be reusable and so on.

But the simple truth is, this has been something which has been part of our support to Manor House as part of our investment, if you want our social impact investment, of trying to help them to navigate this and to gradually build structures.

And I really like the fact that you’ve drawn out ODK as part of those structures. ODK is Open Data Kit and we’ve talked about this before on previous episodes. This is a fairly standard set of open protocols to enable data collection which works very well through mobile devices, getting it into a relatively well standardised form.

[00:06:07] Lucie: And you don’t, quite crucially, you don’t need to even be online to collect the data.

[00:06:11] David: Of course, that is crucial. And this is part of why this has been such a successful set of protocols which have been used in emergency response and all sorts of other contexts. All of this is something we’ve been working on the periphery of for years. But at the heart of what you’re describing is the fact that actually, there is this big underlying need. And Manor House is just one example. We work with partners across low resource environments, I would argue, who can’t buy the, and shouldn’t buy, the commercial tools which are already built to do this.

And they need really powerful, adaptable tools, which can not only serve their specific needs, but be accessible to people with very low digital literacy.

[00:07:04] Lucie: Yes.

[00:07:05] David: It’s the hardest of the hard problems for this. You can have a complex system if you have people who are highly digitally literate, but they need a complex system because the institution is complex, which works with people who have very low levels of digital literacy.

[00:07:22] Lucie: Yeah.

[00:07:22] David: What they do have in their corner is us and INNODEMS, our Kenyan partners. And this is why they’re even engaging in this. Because it’s such a hard problem and we are really interested in this because this isn’t just a problem for them. If we can slowly understand how to support them, the need for this is immense. Institutions and organisations in low resource environments, this is a big need, and we need to develop technology differently to be able to really meet that need.

[00:08:01] Lucie: Can you just give a few more examples as to what other sorts of institutions would have these database needs?

[00:08:08] David: So let let’s think about the fact, what they are.

They’re a working farm, they’re a training institute, and they are an education institution. So let’s talk through each of those three things,

[00:08:20] Lucie: okay.

[00:08:21] David: Most working farms, in Western Kenya are smallholder farms. There are big farms, but there are many smallholder farms. Now, if you’re a big farm in Kenya, you might be on the boundary of where you could start using some of more commercial systems, where if you’re profitable enough, then it may make sense that you could actually bring in and start using some of those.

But even there, human labour is relatively cheap. compared to the technological cost. And so the ideas of precision farming, which are taking off around Europe and, in high resource environments, these are less incentivized in contexts where human labour is cheaper and technology tends to be more expensive and less reliable.

And so there are, even if you just take those farming those larger farming, and they are in some sense farming communities in that case because there’s such a large amount of human labour. There isn’t really the tailored products which would be really useful to them.

And once you start getting down to smallholders, there’s really nothing. And there’s very little, there are apps being created which are trying to do some things on this. I have not yet seen something which I would say yes, that nails it. Any smallholder farmer could use this to be able to manage the flows within their farm in a really consistent way.

Okay. And as a working farm, there is already a whole sort of area there of, from smallholders up to bigger farmers in low resource environments, where although there are lots of tools for working farms out there, I would argue that there is still a need, which is part of a gap, which if there was something which was really appropriate for this, then we would be encouraging Manor House to be using it.

But the simple truth is that even just as a working farm, there isn’t, I would argue, something which is really tailored for their content. And then you can go to the they’re an academic institution. They train students, and so you have all sorts of similar academic institutions from major universities down to schools or, technical training institutes and so on.

And some of them have systems, particularly the bigger, universities, particularly the big private schools and there are many systems that are used for this, student management systems and so on. So there are systems that exist, but again, if we look for Manor House is too small for any of those, and so they do pretty much everything on paper.

Yeah. So almost all their student registration, everything is a sort of paper based system. And going to a digital system would be an immense task for them. And actually their needs are very specific, they have relatively small number of students.

So compared even to secondary schools, their needs are pretty specific. And so they don’t fit into a big category. So having something that customisable that, which would really suit them is unusual, but there’s so many education institutions that don’t fit the moulds of what is available.

Now there are big efforts happening and there’s actually very interesting research happening for high schools and schools in general to be able to understand how can we get school management systems into schools, which would help and enable this sort of digital data. And that’s people are thinking about that at the continental level, even within Africa, of this sort of thing.

But it’s something where the Gambia, for example, has got very advanced in that and there is progress that’s been made. But these are hard problems, again, especially when you’re an institution which doesn’t, which is, where this isn’t standardised, and where you might not have access to something which is really suiting your needs.

And so it’s just a hard problem. And then, of course, training institutions, many of the training institutions are one offs. They’re original in many different ways. And so actually those tailored solutions for training institutions, again, you’d need something which was adaptable in these low resource environments, which can’t afford to pay much for it.

And it’s something where there are. If I’ve missed it and I could have, then I, if somebody can tell us, please share with us what it is that they should be using. But we’ve not found open solutions which are really almost turnkey and yet have this sort of, boundary between working well for people with low digital literacy and being highly customisable.

It’s a hard problem. Now, we don’t claim to be getting anywhere near that. We don’t have the funding to do this as a project I’d love to take on and actually drive forward, but it is something which what we’re trying to do is to support this to be grown in some sense from within the institution.

And that’s a, this is a, an approach which has been very slow, as you say. But I’m delighted to hear that the discussions that they were happening, having, were more advanced than you expected, because that’s the ambition I want them to have. And then I want to try and help, support this slowly to help the technology come up to support them, based on their needs.

And building with them. So it’s something which, emerges and we’re building technologies, which could support this. The whole point of something like ODK is if you’re using those sorts of templates this could then be carried across because it’s in a format, which is then a, it could be embedded into different things and so on.

And so actually this isn’t tying them into one thing. We could be built compatibility with those formats into other systems. So this idea of being able to use these open protocols so that they can be into operability with something which emerges. Now, my, my guess is what they will eventually need is probably a system with a server based database served by a combination of, let’s say, app and chatbot to be able to enable data connection in certain forms.

That’s my guess of what would really serve the web to be able to accommodate for the low literacy users. The chatbot is really good because it can live within the technologies that people already use. Things like WhatsApp, Telegram and the apps can be installed on their devices and they can just learn how to use that in a way which then serves them and not only does the component that ODK does, which is the collection of data.

But the app can then close the loop and give them the feedback so they get the benefits of having had the data. Yes. And that’s going to be really important. But that app, building out such an app to serve them, that’s a huge amount of work. Especially if we’re thinking of it not as something just to build them.

Which they can’t afford as an institution, but something which could be built to be open and serve multiple institutions. And that’s a whole nother big project. Oh, I’d love to take on that challenge. But it is a challenge, and enabling it to emerge, if we were to take that on as an organisation right now, we’d get ahead of them quite quickly.

Whereas, actually, having this slowly emerge, and then pushing. This actually means that it’s coming from them.

[00:16:01] Lucie: Now one question I have is that, so to me, before going to an app and everything like that, before going to digital technology.

And I like paper, so to me it makes sense to do it all on paper first and to understand to me it would help me understand, what are my needs? What data do I actually want to collect? Which data should go where? And things like that. But it’s, I get the impression both from them and from you that you want to go straight from ideas to me, end result.

[00:16:34] David: It’s not ideas to end result. So what is important here, and there have been places and there have been groups who have been using paper for years. But paper is really problematic because you can’t use the data in its paper format. And so you either need a digitisation to put that into a form where that data can actually be properly used.

Or, essentially that data is only serving a very basic purpose. So you can’t easily do summaries based on data which is in paper form. You need to transform it into a digital format to be able to have it in a format. Otherwise, it’s a big human effort. Paper based systems have existed for centuries, have existed for centuries for farm and certain farms and so on.

So it’s not just that there’s nothing original about that, it’s that’s not actually taking advantage of that doesn’t really help. That would just be a lot of work. Whereas actually, if we can have the data to be digitised straight away, then we can very quickly build systems which provide the information in formats which are useful for different actors.

And that’s the key. One of the things I believe very strongly is that by jumping straight into digital data, then we can very quickly incentivize good collection of data because we can get the data to be used and useful much faster. And that’s the key.

Okay. If you start, if you think about it on paper, then, it’s so much work. It’s actually more work to digitise the paper than it would have been to collect it in the first place. And I would claim that collecting it digitally in the first place is potentially less work than just collecting the data, or than digitising the data. And so essentially, you’re reducing by more than 50 percent the workload involved.

Because to make the data useful for paper, you have to then digitise it.

[00:18:36] Lucie: The only sort of, the only things that I saw were the sort of, the summaries of vegetables going to the kitchen

[00:18:42] David: on paper.

[00:18:43] Lucie: Yeah, on paper.

[00:18:45] David: And so imagine if instead of that, that was digitised. And so you could actually then very quickly have graphs.

[00:18:53] Lucie: A yearly overview, yeah.

[00:18:55] David: Not just yearly overviews, but even more short term things that you could actually see if something’s gone wrong, this actually can become visible very quickly.

If the kitchen is potentially going to have a problem because there’s not enough veg or whatever it might be. Some of these things can very quickly then become visible in ways which may or may not be, you’re never going to use the data of the paper record. It’s just there for record keeping sake.

People don’t look at that data and use it very often.

[00:19:25] Lucie: Yeah okay. And then why jump straight to collecting data on er it’s something like ODK…

[00:19:34] David: Collect. That’s not looking straight. That, to me, would be an intermediary step. Okay. Yes, you’re mentioning the

[00:19:42] Lucie: Chat box and things. But why not Excel, is what I’m, is what my question is. Why not go by Excel first?

[00:19:50] David: Because Excel means you need to have a computer with you or something, you’d have, that’s for your data entry, if you’ve done a paper record.

[00:19:58] Lucie: But you can access Excel

[00:19:59] David: on your phone? You can, but it’s but then you need to know what format it’s in and so on.

It’s not actually helping you, so that’s more work, than it would be to do it in OBK or something. So actually getting good data in Excel would be harder. But

[00:20:15] Lucie: ODK just produces Excel forms,

[00:20:17] David: is my understanding of it. So ODK, you have information, questions that you answer, so a well designed questionnaire in ODK is easier to fill out than finding the right columns to fill things out in Excel, and so on.

You’re actually, you’re guided through. And my favourite example in Manor House for this is the chickens. We know about the data collection for the chickens.

[00:20:42] Lucie: No, I know that there was an experiment recently…

[00:20:46] David: Yes, okay, so let me talk you through this, because this is this is what I, this is what I hoped when we started this, and this only happened very recently, as you said.

They had this experiment from chickens. And actually in the end, they first of all, they started doing their collection forms and they were terrible. And in the end, we worked together on them quite a lot. And we got to something, which I am so excited by. And the team, the local team, were finally understanding, and they did a brilliant job because they understood that actually, done well, the form helps them.

[00:21:21] Lucie: Absolutely, yeah.

[00:21:22] David: Actually, it’s this element of this form we then talk through. Okay, when you go… We had an expert, and when you go and you observe the chickens, what do you look for? What do you observe? And the point is, by the time we’ve gone through and actually talked through, What they observed as somebody who, I don’t know what to look for with chickens, as somebody who looks at chickens for signs of all sorts of different things, then the form was something which they can now use to train students.

How to observe things and how to act on a daily basis with the chickens. And this is great!

[00:21:57] Lucie: I saw an early version of those forms and I found them so confusing because I didn’t know if we were looking at the whole flock or if we were looking at just one hen because it then is looking at the chicks and he was like is this the total chicks?

Is this one hen’s chicks? And it meant from one, one level to another. I was never able to have those conversations!

[00:22:13] David: Exactly. And that led us, those were the initial thoughts and what they were just doing was putting down all the different things. And so we then structured that to say, okay, what are the different levels?

What are the different levels? Okay, you’re going to observe something at the flock level. You’re going to observe something at the individual hen level. You’re going to observe some things and so on. And going through those different levels and actually understanding. What you’re observing on what, that was a really powerful exercise, which then led to these forms, which now they are using to train the students.

So the students are able to take over some of the responsibilities of looking after the hens and making those observations.

[00:22:52] Lucie: Yes. And I saw that. I saw that in person.

[00:22:54] David: Yes. And now that data, of course, because that data comes straight away, the student might not realise the full implication. of what they’ve observed.

But they might observe something which needs to be acted on, it then goes up, the data could come through, and then we can build the observation systems which feed back to the people who are responsible for it, so that they get the warnings, or they get the flags, or they can see the things they need to see.

So even if they didn’t make the observations, students made the observations, they get the information. That’s where this becomes so powerful. Now, ODK is a fantastic tool and I’m really delighted with the way a number of members of the team at Manor House have started to really appreciate and engage in this and it’s been so rewarding to see that growth.

I feel a lot of that’s happened relatively recently, but it to me is an intermediary step. You don’t want to get stuck in ODK because the whole point is that actually, although ODK is a good first step and it’s helping them and it will always be useful, really what they need is something which will give them feedback.

It’ll actually have you know, elements of the analysis of that data, feeding back to people who need to know. And that’s what we really want. And this is where I hope things are going with this sort of database and with what we’d hope will come out of this.

[00:24:26] Lucie: Okay, that’s that’s great. And thank you. You’ve… so I came out of my discussions with some of the staff at Manor House, with more questions as a, and you’ve managed to, you’ve explained away some of my questions. And so I now need to work on trying to think of a structure for a database for them!

[00:24:44] David: But in some sense, the structure for the database isn’t the key at this point in time.

This is where the starting point, if the starting point is a collection of ODK forms, that’s where, that’s what we should really be working towards. Because the ODK forms at this point in time, they will translate into tables in your database. So they are the database that you’re collecting.

So I would argue, and this is what I’ve been doing with some of them on some of the experiments, thinking through the data you want to collect and how this can be made useful as a process of designing ODK forms is in some sense designing components of an underlying database in a way which is very flexible and adaptable to their needs because they have, we, they have people there in Kenya who can author these for them, who can edit them. They become theirs and that’s what’s so important. We don’t want a database that’s which they are then tied to in a way which they can’t change.

And this is why the ODK form process is so important, because if they need to change it, they have the ability to do so.

[00:25:53] Lucie: Exactly, yeah.

[00:25:54] David: We don’t want to design a database which is outside of their power to change. That’s, that would not serve their needs, but this is a really complicated concept, because, in some ways, they need something which is relatively standard, but in other ways, what they need is very custom.

Yeah. And it’s gonna change over time. The example we have of the chicken, which is related to this experiment, that experiment is going to be relatively finite in time, because they are finding pretty quickly that if you use the I think it’s Tephrosia that they’re using, if you use the Tephrosia, it is reducing seriously disease and pest pressure on the birds, and so the birds are doing so much better in the experimental part where you’re using this than the other part.

But pretty quickly they’re going to say, okay we have enough data to actually draw these conclusions as academic research and we will now be treating the whole flock because it is making such a difference. And then on the back of that, my hope is, they will then say, but what we don’t know is whatever it is they don’t know next.

And then they go into another experimental phase. This is really at the heart of what the institution is about. It is also a learning institution. So the observations, that they are trying to make from this experiment of using Tephrosia on the chickens, has come out of the farmer research networks who had made observations like this and who have now tried this and who have substantial evidence at a farmer level, that it is effective.

And so they’re no longer wanting to test this because they all believe in it and they’re now adopting it. And this is something where now getting the research evidence to back that up, actually to be able to quantify some of the things that they’ve observed.

That’s part of what’s happening in the experiment, but it’s happening because the farmers who Manor House is working with as part of these farmer research networks, they have evidence, which is not yet scientific evidence, but it’s evidence which is convincing them about the effectiveness of these treatments.

[00:28:18] Lucie: Yeah. Yeah.

[00:28:20] David: My, my hope is if they as an institution embrace this, then because there are this combination of doing bits of research, they’re training students, they’ve got these, they’re a training centre, they relate to these farmer research networks and they’re a working farm.

If this just becomes part of their processes, the data that’s going to come out eventually is going to be extremely useful for research, as well as for them, and helping them as an organisation. And so my hope is, and this is why we’ve never forced this, we are responding to them and supporting their needs on this.

This episode, to me, it didn’t need to be about Manor House. It became about Manor House because we’ve already had an episode on Manor House, and you were there recently, this is a recent experience, and this is how we work.

Trying to support organisations, and this is not because there’s a grant to do this is because this is something we recognise is a challenge in so many organisations and we want to be the people to support this. We want to solve some of these problems. That’s that’s our ambition as an organisation in many ways.

To build the tools that will really help organisations like Manor House to have a step change in how they can do that and use data.

[00:29:43] Lucie: And understand their own processes fundamentally, yeah.

[00:29:46] David: And learn and document and experiment all these different things, which at the moment are, if you’re in a high resource environment, these are now possible because you have data everywhere and data is omnipresent.

They’re not always possible because you might not have ownership of that data or access to it. And that’s again, this is where working in these low resource environments, the value of that data is arguably of less commercial value, which means that there’s less pressure to actually for somebody to take it from you.

And so actually we can give ownership back to Manor House for this data and they can be using it for their benefit, which is what we believe in terms of building community tech. So this comes back to our long term goals about community tech, how we want to build community tech. It’s a really big challenge.

And it’s something where it’s something so simple, this is where it, I think when you came with this sort of as a topic, having discussed it with them, it feels like a very small topic, but it’s a really big, hard topic because to go further than the sort of Kenyan organisations, I believe pretty much any community organisation internationally, it’s the same thing.

If you have an organisation which is not fully commercial, but it’s serving community in different ways, then these same questions about data collection would come up. We were very recently at a very interesting community centre, Braziers Park just not far between Reading and Oxford, near a place called Wallingford and a wonderful community that is a learning community. A big part of what they’re trying to do is learn. But again, the question is, if you really want to learn, you want to be part of global learning, what data do you have? An archive going back I think it’s the 1950s, so over 70 years, and incredible information.

But they’re just starting a project to digitise it. If it’s all on paper, it’s useless. Nobody can access it unless they go into the room. It’s not accessible. If it’s digitised, then they have the paper records of many different things, of their, weekly meetings going back for 70 years. Amazing data, but without digitising it, it’s very, it’s inaccessible.

It’s a huge mammoth task to access. Whereas once it gets digitised you can find ways to use it and you can maybe work on that data in such a way that it becomes more accessible, more easily accessible, to be useful for something. And this we think is the same case in community organisations all over the world.

So we would love to be collaborating with such organisations to say, how do we build systems? How do we get, build tech that enables community organisations to have that data and to use it in ways which are currently not really possible?

[00:32:54] Lucie: Yeah. Yeah. I think that’s a really nice place to end.

[00:32:57] David: Thank you for bringing this topic up. It’s been a really nice discussion. Thanks, David. Thank you. Cheers.