109 – Mathematical Scientists as Impact Activators

[00:00:00] David: Hi and welcome to the IDEMS podcast. I’m David Stern, founding director of IDEMS and I’m here with my co founding director of IDEMS, Danny Parsons. Hi Danny.

[00:00:17] Danny: Hi David, nice to be back on the podcast.

[00:00:19] David: Yeah, it’s great to have you. We haven’t done one in a while and we’ve got a really good reason to do one now because we’re going to be giving this impact activation workshop next February and it got us thinking what do we mean by impact activation?

And this is something which has really come out of our experiences, but it’s something which we’ve now made part of IDEMS more generally in the Impact Activation Fellowships. And it’s really central to a lot of what we’re trying to do.

[00:00:51] Danny: Yeah. It’s something we’ve related to Mathematical Sciences and people who’ve studied Mathematical Sciences. And it’s about impact, impact activation and showing that there’s a role for people with a mathematical sciences type background to be involved in work that’s really impactful in development type work and other areas.

And yeah, the workshop is a really nice opportunity to bring people who might be interested in this with a sort of mathematics background, but maybe aren’t aware of what the opportunities, aren’t aware of how their skills can be useful.

Yeah, I think it’s a really nice thing. We’ve obviously had I think, a lot of success, I would say, in bringing in people with that background through this Impact Activation Fellowship role that we’ve created and some really excellent people. There’s no shortage of people with the skills who we found that can contribute to the work and the work that they can contribute to.

[00:01:52] David: And the demand for people with our sorts of skills is huge in almost every area where we start collaborating. Because of the value we bring, the demand is there, but it’s a latent demand and people wouldn’t think of it at the outset. And if they did, they’d struggle to find people with the ability to enter in as we do.

[00:02:22] Danny: Yeah, I think a lot of organizations in impact spaces wouldn’t necessarily always think of mathematicians as being the people that they need and to support them. And also the people with the mathematics background themselves probably aren’t seeing many examples of where they’re directly, obviously, all areas of mathematics have big influence, but that’s using your sort of skills more broadly rather than specifics. And so that we’re not talking about sort of high level mathematics research, which is going to have impacts, but on more longer term, this is getting involved in things directly.

[00:02:59] David: And this is the key thing that in some sense, what we bring is that having those high level mathematics skills that you could be doing with maths research, but impact activation is about saying this is not about maths research, this is actually saying you need a whole nother set of skills to be able to enter in to other people’s domains and contribute there rather than focusing on the mathematical set.

And I think that’s critical because actually quite a lot of people, if they went to a mathematician, the mathematician wouldn’t help them because the incentives in academia for mathematicians is to do maths research, and that’s correct. But that’s often at odds with what we would consider the activation part, is all about moving away from thinking about doing maths as maths and thinking about simply what’s needed, and what can I bring to the table?

[00:04:02] Danny: Yeah. And I think a lot of it is a sort of more kind of core skills in terms of the way you think, you’re logical thinking, critical thinking, we talk about this with people in our team who want to ask questions and understand things, they don’t want to just do things that they’re told, they don’t want to do something unless they understand it.

And that’s a good thing, they really get that deep understanding. So you can see how it’s not obvious, necessarily on both sides without the experience of this kind of relationship, how you use a mathematician and how a mathematician should become useful. I think it’s different to a lot of other fields where there’s a specific application for that, whereas mathematics, you can see how it can be useful in all areas, but then it’s hard to see that clearly because it’s related to these core skills when you’re thinking about the impact.

[00:04:52] David: And I really liked the way you brought out this sort of questioning element because I still remember very vividly when I first entered into agriculture and started working with experts in agriculture. And I was useful almost immediately in a way which I’d never expected, which was simply that I would be in a group of experts and I would say I’m sorry, I’m a mathematician, I don’t understand what you’re saying because I’ve heard you saying this and I’ve heard you saying that and I don’t know that I understand the difference. It sounds like you’re disagreeing with one another but I don’t know that I’ve understood what the difference is. To me it sounds like this and actually just feeding that back into the room sometimes de-escalated differences .

I wasn’t seeing the difference and just by being able to be a non expert, sometimes just being able to listen really carefully and precisely and try to feed back what I’d heard, enabled people to articulate more clearly, what they had in common and what was actually different. And quite often, there were subtle differences, but there was much more in common than there was different. And that was already useful.

Now, I was already established, as a mathematician and a statistics educator. So I had the confidence that I am an expert in my own right. And I was able to go into that room and say, look, I’m an expert, but at other things, I’m trying to understand your subject. Can you help me? And that process was already useful.

[00:06:42] Danny: Yeah, that’s a nice example. I feel that there’s different examples like that where someone with a mathematics background can be useful, and some mathematicians would be good at that kind of thing that you described, and some maybe not as good, and there would be other things.

The kind of example I have in mind that I’ve noticed a bit for myself is related to what you said, but it’s being able to step back a little bit from what’s being discussed and what’s going on and look at something a little bit with a fresh eyes, but abstracting a bit is how I’d describe it and being able to look not just at how everything is at the moment, but saying why is it like this? And if I step back, why couldn’t it be like this?

It’s come back to the questioning again a bit, and sort of abstracting out a lot of the noisy stuff, which is the sort of important details of the application, but allows you to maybe ask questions that people hadn’t thought of and question things that have always been like that. And sometimes then you find out other things of why this is happening and sometimes it can lead to new things.

That’s a very mathematical thing. It’s almost like when you have these exam questions which talk about, you’ve got this many red balls or something, or you’re baking a cake like this, and that’s not important. Strip away that. What are you actually asking? What are you actually wanting to do? And that’s the kind of useful question, I think, in projects and big projects as well.

[00:08:03] David: And exactly this idea that once you’ve stripped away all that noise, you’ve got something which could be quite precise. But often when you feed that precise thing back, you’ll find that people who understand domain and the context disagree with it in ways which they can articulate. Where within the noise, they often can’t articulate it.

[00:08:25] Danny: Yeah. And I think then a lot of people wouldn’t be able to cope with that necessarily. Maybe one of the reasons people who study mathematics like it is because you don’t have that noise and it’s the sort of very logical and that simplicity that you can get down to.

So I think to work in this space, you also need to gain new skills as well and that communication and that teamwork, yeah, get into a new area is something that is going to be a learning part on the sort of mathematician side.

[00:08:52] David: Absolutely. To value people who don’t have that abstraction skill, to understand that they can’t efficiently and effectively communicate to you what you need to know. You need to slowly and patiently listen carefully through I want to say a sort of less clear description where quite often the stories that are told are what people care about, whereas from a mathematical perspective often we have a tendency of stripping the stories away to get to the heart of it.

[00:09:27] Danny: It’s knowing. Which bits you can strip away to help progress, and which bits are really important and need to stay there.

[00:09:34] David: Exactly, and valuing the stories. If you’re a mathematician who doesn’t value the stories, you’ll never be able to be good at this sort of role, because the stories are what matter in many cases. And if you just strip it away, and you get rid of that substance, then you might be able to solve a problem, which is abstract, but it often won’t solve the real problem because the real problem is embedded in the context.

[00:10:03] Danny: Yeah, and there’s this sort of compromising and back and forwards, you’re going to have to do things in ways that are not perfect but are needed and so on. So, it’s an interesting experience, I guess, people would gain in that.

[00:10:16] David: And this is where we talk about impact activation, because I think most mathematicians have the potential to contribute a lot, but that potential will lay dormant unless you build some of these other skills, and you think about some of these other skills, and how you could contribute them.

So I’m interested that you had this, I think, really nice abstract way of describing how mathematicians can build this abstraction. Can you tell a story about it? Can you actually give a personal experience where you found this happening?

[00:10:53] Danny: Umm, put me on the spot a bit now.

[00:10:55] David: I’m happy to do that, as you know.

[00:10:57] Danny: Yeah, I’m not sure I’ve got a concrete example in mind, but I think about it in relation to software development a little bit and thinking of how different software projects that we’ve been involved in. Maybe thinking about R-Instat and the development of that as a statistics software. There’s, I guess, ways of thinking that have helped us to do things that other software aren’t doing and to say this should be something that’s easy, shouldn’t it? And we could do this and make this something that’s easy to do for users. But, you know, why isn’t this normally done like this?

Maybe an example, trying to get slightly more concrete, is operations or statistical methods or processes, which would normally be separated out into this is this process, this is this method, this is this method. We’ve tended to think a bit more, well actually aren’t they just really the same method but done in slightly different ways or different options. And building in then that flexibility to be able to do them and to see them as being similar in how we build the software.

[00:12:00] David: Let’s try and get really concrete on that in some ways. So this could be, for example, with some of the summary system, that there’s sort of elements of that where, you know, all of these are easy things to do if you’re coding things up. And then there’s these very complicated summaries for the climatic, which are tailored products, but where in specific contexts we want to actually bundle that all together so it just becomes a few parameters that you’re actually setting and choosing for start of the rains or something like this.

[00:12:35] Danny: Yeah, Yeah. We were then able to think about the building blocks of summaries and think of them that way, so that then you can construct any summary because you’re basing them on some fundamental building blocks, whereas if you just did them, each one separately, then there’s only so many because each one is just the whole summary. Yeah, it’s being able to abstract back to building blocks and things like this, which gives you this great power and the flexibility.

[00:13:07] David: So let me see if I can actually articulate this as a sort of story in some sense. Essentially, the summary system that has been built into R-Instat, there’s two ways that people generally tend to think about this. If you want to do summaries or if you’re thinking of coding, then you just can do summaries, it’s easy. You know, you code them up and you code a summary up individually. If you want to build a tailored summary, then you write a function. You sort of want to do a start of the range, so you have a start of the range function with a number of parameters.

And stepping back to say coding up summaries is something which isn’t going to give the flexibility, isn’t going to be easy enough for the people who just need to be able to do certain tailored summaries. But if we just write the tailored summaries, then people are going to get tied into what those specific functions can do. And those functions would be black boxes.

And so within R-Instat, you actually have this rather strange language which has emerged, which is this sort of summary system, where, if you do a start of the range summary, you don’t get a script to do it in some sense, there isn’t just a single function, but there’s this sort of summary language or summary system which has emerged, which actually has all the components in a way which is not easily readable, but is something which now could be tweaked and could be changed and could be adapted.

[00:14:42] Danny: Yeah. And one of the advantages of that then is that it’s able to do things that users haven’t yet even asked for, but which they might ask for in the future. Oh, can we do it like this, but slightly different? And we can say yes, because it’s just slightly changing this part of the building block in this thing. And so it’s that sort of abstraction and thinking ahead.

[00:15:04] David: And I think what’s really interesting on this is that I would argue that at the moment, even though this is probably 10 years in from when we started working on that, that summary system hasn’t yet paid off, but I still believe it will.

And this is I think part of the key, that in some sense what that level of abstraction gave is it was about understanding that, okay, going to the extremes of just writing a function to do these individual things, versus actually just writing the code and the script in a normal way to do it where each one is an individual process. In the long run, as more people want to do these things and as we’re going to do those, actually building out that system and that approach to doing it is going to have reusability in a way which is going to be important. And as a mathematician, that level of abstraction to be able to do it from both is an interesting instance.

Now, I don’t feel that this is a great example, which is reaching hearts and minds, but this is an example of abstraction related to software development, where the way we develop software is different.

[00:16:15] Danny: Yeah, and it’s not unique to mathematicians, obviously.

[00:16:18] David: No.

[00:16:18] Danny: That would be what good sort of software architecture we’d discuss as well.

[00:16:22] David: And this is where good software architecture, just like good algorithms, tend to become very mathematical.

[00:16:28] Danny: Yeah.

Another sort of example of a sort of skill that I feel is used quite a lot, especially in our work, which comes from very mathematical places, looking and seeing connections between things which are different. We have a lot of different projects and we work in different areas and so things can look very different but being able to see similarities, and this again relates to abstraction because you’re seeing what’s the same in different contexts, is I feel a big part of what we do.

I mean it comes down to some of the software we’re building, where we’re thinking about not just implementing for a specific context, like for the parenting work with apps and chatbots, but thinking what are the underlying things that we want to build, which could then be used in other cases, which are not related to parenting, but are related to apps and chatbots.

I think that’s a big part. I already have one specific example for that…

[00:17:24] David: Go for it.

[00:17:25] Danny: …which I had in mind, which is a small example. It was when we were organizing workshops related to agriculture projects in Niger. And we wanted to have workshops with both farmers and people who worked with farmers, and with students in the mathematical sciences, and bring them together because this was about data and agriculture and also how mathematicians can be useful.

And I remember, one of the ideas coming up of how we should structure the workshop is why don’t we have the sort of mathematicians with the farmer group people together for part of the workshop because this is exactly what we did. We had workshops where we were looking at procurement data in completely other countries, but in another context where we had mathematics students again with experts in public procurement, looking at public procurement data.

And that kind of worked quite well. And then we brought that to another context, and actually led to us writing a paper on that kind of approach in general, of bringing experts with mathematics students together. And that was sort of, I remember, almost like a light bulb, oh, isn’t this just similar to how we did in the procurement context, and why are we separating out these two groups when there could be interesting stuff when they’re together.

[00:18:41] David: Exactly. Yes. Let me just reframe that example because it’s a beautiful one. We had some work related to procurement and in particular the analysis of procurement data to be able to identify red flags and instances of potential corruption. And that piece of work was very interesting piece of work, which we then got to this point where part of what we were doing was training, they were really data scientists, to be able to look at this procurement data and analyze it and identify these.

And of course, they could do the technical bit of the analysis, but they didn’t really know what they were looking for. And then we organised these workshops, which then put them in little teams with people who had actual experience in procurement or in corruption.

[00:19:36] Danny: And this came from doing them separately at first, where we met with procurement experts separately after we had done some sessions with the mathematicians, and this was good, but wouldn’t it have been great if they were together, because they could both learn a lot from each other?

[00:19:49] David: Exactly. So getting to that point where actually we actually had them in the room in the same way, there was a sort of process for workshop which emerged. And then it was a year or so later that we then had an opportunity to do data visualizations with farmer groups. And again, we had this sort of relatively large amount of data that was coming out. And so we thought we’ll get data students to be able to analyze that data and produce some of these visualizations. We can train them how to do that and then they could create this at scale that we could then share with the sort of farmer representatives.

And then of course the light bulb moment that you mentioned was to say but this is exactly the same as the process that we saw in the corruption work and hence we should take that same approach of putting them into the same room together.

And it’s quite nice that this example is an example of impact activation, I would argue, because the learning that came for the maths students, the data students, about what it means to then work with people, what to look for, what they care about, was huge, and this worked very well. In fact, one of those is, of course, now part of our team in Niger.

[00:21:01] Danny: Yeah and I remember from the public procurement workshops, one of the people working in public procurement in the government or one of the countries was saying, I didn’t really ever realised how useful someone even with data skills, let alone mathematics could be in the work that we do.

And yeah, I didn’t think of this in advance, but it’s a nice example because it links back to our workshop early next year on the impact activation where we are bringing together sort of people, mathematics type people, interested in impact activation.

[00:21:31] David: And I think what we are going towards is that if we really wanted to do the sort of impact activation that we’ve experienced in other cases, we can do something which is context independent, which is what we’re going to try and do next year. But it’s so much more fun when you make it context specific and you actually bring people in from a particular context and you actually build those links and you actually get it to grow from there.

And in some sense we’ve got more experience of that, that’s easier to do. And I think what we’re hoping is to get to this element where we can prime people to be part of that process in the future. But I do love those opportunities to actually put these two groups together in the same room.

[00:22:19] Danny: Yeah, and see those learnings together. Yeah, I’m excited about this workshop coming up. I think, we haven’t said, talked about, just the standard routes for sort of a mathematics person, people see it as research and teaching, or into something like finance and banking, and that’s fine for some people, but a lot of people would really be interested in these impact kind of routes and to be able to just show what we’re doing on that and how people can be useful.

[00:22:49] David: It is this element that I believe there’s so many elements of work related to social impact in our societies where more people with strong mathematical skills in those areas would help us build to more sustainable futures, and actually would contribute.

It’s not going to bring the solutions, of course not. But it’s about being part of these transdisciplinary teams. And I think part of the reason I’m so passionate about this is I do believe that a lot of inefficiencies that sometimes grow in the sort of third sector could be helped if there were more people with mathematical skills in that area. There’s good evidence of this because the few people who there are, who come from this area, tend to be successful.

[00:23:42] Danny: Yeah, and I don’t think there’s a shortage of people who would be interested in this.

[00:23:45] David: Exactly.

[00:23:46] Danny: It’s just knowing about the opportunities and there being the opportunities for them.

[00:23:49] David: Creating a pipeline for those opportunities, because if everything is ad hoc, you’ve got as an individual going down that road, I mean, we’ve both paid quite heavy prices in certain ways to go down that route. And it’s too much to ask for that to scale. Whereas there’s no reason why it has to be as hard as it is at the moment to find your own route.

[00:24:14] Danny: Yeah. I’m looking forward to seeing what comes from the workshop.

[00:24:19] David: Yeah, exactly. And my hope is, of course, that this is just the beginning. That this is actually an approach which could then lead to something more. Anyway, this has been great to talk about, and nice to have you on an episode again.

We should do this more regularly.

[00:24:37] Danny: It’s nice to be back on.

[00:24:38] David: Thanks.

[00:24:39] Danny: Thanks.