Co-design in global development data Initiatives with Dana R. Thomson

Co-Design in Global Development Data Initiatives Transcript

0:00:00.5 Tom Llewellyn: Hey, Tom here. We’re in the middle of our end-of-year fundraiser and could really use your support. If you appreciate this show and are in a position to donate, please click the link in the show notes to make a contribution today.

[music]

0:00:19.5 Dana R Thomson: We are really looking to ensure that the data we produce are generative, that they generate new understandings among stakeholders, specifically in our case, the communities and local government. So what do decision-makers have to say at the moment about the priorities that communities have articulated? And what data or information are local government city managers already using to address the priorities of communities? And can we actually model data or generate new data that facilitates new understandings between these kind of groups?

0:00:54.8 Tom Llewellyn: Welcome to another episode of Cities@Tufts, brought to you by Sharable and the Department of Urban and Environmental Policy and Planning at Tufts University with support from the Barr and Shift Foundations. Today’s show features a lecture from Dr. Dana R Thomson, who discusses co-design in global development data initiatives. In addition to this podcast, the video, transcript, and graphic recordings are available on our website. Just click the link in the show notes. And now, here’s the host of Cities@Tufts, Professor Julian Agyeman.

0:01:38.8 Julian Agyeman: Welcome to our last Cities@Tufts Virtual Colloquium this semester. I’m Professor Julian Agyeman and together with my research assistants, Deandra Boyle and Muram Bacare, and our partners, Shareable and the Barr Foundation, we organize Cities@Tufts as a cross-disciplinary academic initiative which recognizes Tufts University as a leader in urban studies, urban planning, and sustainability issues. We’d like to acknowledge that Tufts University’s Medford campus is located on colonized Wampanoag and Massachusetts traditional territory. We’re delighted to host Dr. Dana Thomson for our last colloquium. Dana holds a BA in geography from George Washington, an MSC in Public Health, global public health from Harvard School of Public Health, and an MSC and PhD in Social Statistics from the University of Southampton in UK. Dana strives for open data, user-centered design, equitable partnerships that address historical inequities, and co-design of meaningful information.

0:02:45.5 Julian Agyeman: She’s just begun as Associate Director of Science Applications school at Columbia University, and she helped to establish and now coordinates the Integrated Deprived Area Mapping System, IDEAMAPS. And this is a network which brings together experts from traditionally siloed slum mapping traditions including community members, government officials, humanitarians and data scientists. Several projects within IDEAMAPS are building the infrastructure and processes necessary for slum residents and local experts to not only access the data they need in appropriate formats, but also to validate model outputs and apps for modelers to access continually generated validation data to improve models. Dana, a zoomtastic welcome to Cities@Tufts. Let’s go.

0:03:38.9 Dana R Thomson: Thank you so much, Julian and the Cities@Tufts team for having me today. It’s it’s quite an honor to be thought of as a urban thought leader. I would probably characterize myself as an urban thought learner. Today’s presentation is really the first draft in a set of presentations that I would like to be doing in my communities of practice, which are people who model data for usually development and humanitarian purposes at a global scale. And often the concept of co-design and participation in the process by people who are the subjects or the people experiencing the phenomena that we’re modeling and mapping, usually this doesn’t get incorporated fully. So I’m gonna be sharing with you some frameworks and language that we’ve been using within the IDEAMAPS network and hopefully bringing into my role now as Associate Director of Science Applications at Columbia University.

0:04:36.7 Dana R Thomson: Are you seeing my screen? Okay. I hope. There’s two things I wanna tackle today. One is co-design and social power. And this is really reflecting on why multi-stakeholder workshops and some of the kind of what we thought were best practices in participatory modeling were not enough in our initial IDEAMAPS projects and how we’re thinking about it differently now and still trying to articulate, and just underscoring that co-design and scaled data are actually complimentary, they can coexist. But how we do that in hearing the kind of voices and grounding our data and the multitude of experiences of people on the ground in very different types of communities even within communities isn’t easy. But we’ll talk about how we’re thinking about it in IDEAMAPS. So I’m actually gonna start today’s presentation before telling you about the exact data that I’m working with and the colleagues that I’m working with.

0:05:39.9 Dana R Thomson: I’m just gonna share with you what we’re thinking of as our current approach to modeling data in a participatory way and talk you through the elements of it, and then kind of work backward and say what’s different in most modeling initiatives today? Where do they fall short? So we are calling this a double loop process. It’s actually more than double loop because the idea is that it continues to loop and be iterative indefinitely, but it’s a process that really starts with communities. So the first top wedge are communities. These are people who are often marginalized. We might think of them as the experience experts. They are people who experience firsthand the phenomena that we’re trying to model. So in IDEAMAPS that would be deprivations related to being an informal settlement, insecure tenure, under serviced areas, areas with limited quality infrastructure, et cetera.

0:06:42.9 Dana R Thomson: So in our case, these are community members, and we start with communities to ensure that whatever we are modeling is relevant. Is it what are… We want to listen in our first phase, we being people who do data science, who produce data, who secure the funding to generate data. We are listening first to communities, that marginalized population, to understand what are the realities they face and what are their priorities? How are they already addressing those priorities and what data or information are they already using? What does the systems of information look like? And knowing from their perspective, who is holding the power and the purse strings because those are also really important people to engage in the process.

0:07:33.4 Dana R Thomson: The next step is engaging with those so-called power holders. In the context of IDEAMAPS, these are usually city managers, people in local government who are making decisions about the phenomena that we’re modeling. And in this phase or in this step, we are really looking to ensure that the data we produce are generative, that they generate new understandings among stakeholders, specifically in our case, the communities and local government. So what do decision makers have to say at the moment about the priorities that communities have articulated? And what data or information are local government city managers already using to address the priorities of communities? And can we actually model data or generate new data that facilitates new understandings between these kind of groups? And along the way, we’re actually accumulating stakeholders. So we’ve got community in the first step. In the second step, we’re actually thinking about communities and local government. And in this third step, we’re actually bringing in perspectives of basically all key stakeholders.

0:08:44.3 Dana R Thomson: So in the case of slum modeling, these might be community-based organizations or NGOs. It might be local academic perspectives or private sector service providers. So across stakeholders, what are priorities? Are there some commonalities among the different stakeholders about what is important about slumness or informality or security of tenure, for example? Where are there contradictions? Often the data that city governments want, maybe boundaries of what are called slum areas locally, are data that community members do not absolutely do not want to be mapped because it creates a risk of being fined, harassed, or at worst case, evicted from their homes. And so what are some of those tensions? Where are the priorities aligning and contradicting one another? And what are some of the sensitivities that we start observing among the different stakeholders that we really need to be mindful of during a modeling process?

0:09:46.5 Dana R Thomson: And now, finally, people who are working with a modeling team, I’m calling them modelers, but in our case, it’s a very diverse team. It includes lots of people who are not modelers, but who are academics from social sciences or people who do a lot of participatory qualitative work. In this case, it’s the team that are modeling the phenomena. This, in other modeling initiatives outside of our project, it’s not academic led. It might be led by a national science foundation or a private sector company, but what are the inputs and methods that would be feasible to model the kind of phenomena and priorities that are coming out of these conversations? We’re now integrating and listening to all of our stakeholders. What can we actually do with them? And how might our model data actually leverage existing data or supplement existing data or data systems that we’ve learned about from our stakeholders?

0:10:46.6 Dana R Thomson: And in this step, we’re actually gonna start producing. We’re actually gonna start creating a draft version of our model, of an initial model that we can bring back to stakeholders and start what we call the outer loop of this double loop process, which starts with validation. And that validation is actually a really key part where we kind of center this validation in our process. So we’re bringing it back to communities one-on-one with specific community and maybe members that we’ve engaged with before. Did the model results surprise them in some way? Are areas showing up on a map that they didn’t expect to be classified as informal or vice versa? Did they think that the model results are wrong and all models are wrong? Some are just more useful than others. But we wanna know from that community perspective with the specific stakeholders we’ve engaged so far, where specifically are the models wrong?

0:11:37.9 Dana R Thomson: And if we can communicate what data went into these models, maybe we’re not gonna explain machine learning modeling techniques, right? But we can actually talk about what data went into our models. Do they have any insights about why the models might not be performing well? We often think of the expertise existing with the people who hold PhDs, who receive funding from big grants, but there’s a lot of expertise in the field and in the communities that we’re modeling. So let’s start asking them for their feedback. Now, again, bringing the communities and now the so-called decision makers along. So these are our city managers, how can they obtain this data? Is it interpretable to them? Is the map make sense to them? Does it make sense to model access to transportation as a continuous value or as a binary value?

0:12:39.4 Dana R Thomson: These are some of the detailed kind of questions that we’re gonna start asking at this step. And how might these model data fit into whatever existing decision-making systems already exist? How might it fit into existing data sets decision-making processes, and in what formats do they need these data? Are there community members who might need data, pre-populated posters that could be printed and brought to a community meeting? Maybe this is where data are actually being used primarily for this type of user so that they can discuss as a community what their priorities are and plan their own actions and plan advocacy campaigns and meetings with specific local government departments. And maybe a poster is the right format for them. They’re probably not gonna be opening a mapping software and downloading a spatial file and summarizing it in their own way and producing maps, whereas maybe a local government does want to do that. Maybe they need data in that kind of format. So let’s talk about how access and format of data can be improved for whatever initial concepts we have.

0:13:49.7 Dana R Thomson: Next, we’re bringing all of these stakeholders along to an all stakeholder kind of engagement. This might be a workshop, but we’re really understanding here, impact across different stakeholders and starting to think about transferability and scalability and kind of meeting multiple priorities at once. So are the data accessible and in the right format for a broader set of users? Maybe the UN Habitat wants to take informal settlement maps and use that to calculate an indicator like SDG 11, which is the percentage of population living in slums and informal settlements across all countries and across the globe. Can this set of stakeholders be… Can their priorities be met with the data? And if so, what format do they need? And can the model data support change? This is really about change making. Can we actually support change among the stakeholders who are in the room? Something that would not have been possible without these data.

0:14:50.7 Dana R Thomson: And in the next iteration for the modeling team, we’re thinking about scalability. So in all of the initiatives that I work with, I’m thinking about data that come from open data sources, from things like earth observation which have very high resolution or detailed information about the whole globe usually. So can we take the experience and the models and the feedback we’re receiving in a specific context and make a model that actually might be accurate and valid and usable somewhere else for a a similar set of stakeholders. So the first thing we’re gonna do is actually reproduce. We’re going to create a new version of our model for the teams that we’re already engaging but also start validating with new groups of stakeholders. So we’re gonna revalidate with the current stakeholders and then transfer, validate and improve models in new settings, taking on board maybe slightly different sets of priorities in very different contexts and sensitivities.

0:15:53.2 Dana R Thomson: And so now we’re thinking about data pipelines and what kind of workflows. We’re thinking about the workflows that we need, but also the kind of standardized harmonized inputs that we would need to make something scalable. So that was a lot of detail and a lot of high level stuff, but I’m now gonna tell you about IDEAMAPS and how we’re using this approach now and why our previous attempts really fell short. So IDEAMAPS, as Julian mentioned at the top, is the Integrated Deprived Area Mapping System. We are a loose network of people from different slum mapping traditions that found ourselves siloed, frustrated whether we are community-based mappers or academic-based AI modelers or census survey folks in government or the NGO sector. All of our approaches to mapping slums and informal settlements to date have provided pretty limited coverage globally. Very little comparability among our methods and a lot of inaccuracies.

0:16:57.3 Dana R Thomson: And so we really wanted to create citywide surface maps of deprivations and assets that define slums or the absence of slums. And we want that data to be able to be produced routinely because our cityscapes are very dynamic. We wanted it to be accurate or valid at local level, and we wanted it to be scalable, meaning that it’s somewhat comparable across settings to be able to support use cases like the SDG monitoring. And so our overall network really has the objective of supporting participatory slum upgrading from within slum communities, supporting urban planners and people thinking about integrating low-income communities with the rest of the city, acknowledging the assets that low-income community members bring to cities and working on that participatory upgrading process, facilitating that and also supporting local SDG or other development indicator monitorings, understanding how vulnerable populations have changed over time and are we addressing and dealing with vulnerabilities in a sensible and effective way.

0:18:09.3 Dana R Thomson: There’s a whole family of projects that have grown out of the IDEAMAPS network. This is actually just a few and there’s actually a few more branches that need to be added to this. But the current project that I’m talking about today is one of the more recent projects. It’s a three-year project funded by Bill and Melinda Gates Foundation.

0:18:28.0 Dana R Thomson: We’re just finishing year one, more or less, and we call ourselves the Data Ecosystem Project. It’s being co-led by University of Glasgow, University of Lagos in Nigeria, African Population Health Research Center in Nairobi, University of Twente in the Netherlands, and George Washington University in the US, also University of York is involved, and now myself as Columbia University involved. And the goal of this project is developing not necessarily the datasets, but we’re actually reframing our focus, which is to develop the platform and process for participatory modeling to take place, really facilitating equitable exchanges of data between the people who experience, who have the experiential expertise of living with deprivation or with assets, and the modelers who can start producing this data at scale.

0:19:26.2 Dana R Thomson: And the idea is that we can provide the means to create citywide surface maps of assets, deprivations, routinely and accurately at scale, and allow different models to be incorporated in this platform over time. So from the very beginning, when five years ago when this network was coming together organically, we’ve accumulated members, stakeholders and users. We’re often usually the same people. We’ve accumulated members from a number of different backgrounds who care about or have an interest in slums, informal settlements, and inadequate housing. So those are first the community members or community-based organizations that provide representation in deprived communities. There’s also local non-governmental organizations that might exist nationally or in the city, local businesses from informal sector all the way up to the very formal global sector.

0:20:22.7 Dana R Thomson: We’ve got the international non-governmental organizations who, as many of you might know, often divide themselves arbitrarily into kind of long-term development, supporting projects, and like kind of humanitarian short-term response projects, but that’s all related in my mind. We’ve got local government, we’ve got national government, depending on the context. One or both of these governments have a lot of say at what happens in the city. We’ve got international agencies like UN agencies. We’ve got local academia, global academia, the distinction being for a given context, which institutions already have established relationships and are based in a similar context or the local context versus academic teams that are outside that context. And in academia, there’s tons of ways that we divide ourselves by discipline and sector, but there’s two groups that are really important in our IDEAMAPS initiative in this case, which are like the open science people and the data science people.

0:21:29.6 Dana R Thomson: And we also have private sector. So we knew that this is our kind of like landscape of stakeholders. How do we prioritize all of these different needs and use cases, et cetera? Well, we really focus on, we kind of intuitively focused on community members, local governments and international agencies and the use cases that they bring to the table. For community members and their organizations, they’re looking for fine scale data that characterizes the deprivations they are experiencing or the assets that they offer as a community to the city overall, and maybe even making comparisons between their community and a city average so that they can advocate for the services and infrastructure that really should be a public service and that they can work with specific local partners, namely government agencies and related private sector on upgrading projects, sustained upgrading projects. So they are seeking data and already collect a ton of data actually for these purposes, but that’s where we’re gonna be plugging in at the community level.

0:22:41.9 Dana R Thomson: Local government of course has lots of data around urban planning and might be able to supplement that data with more information around kind of citywide situation landscapes and in particular helping provide information that is current and up to date, modeling kind of between data collection activities ’cause not all data are current. You know, you might be urban planning department by be working with 10-year-old data or even older. And then from the international perspective, the current approaches to measuring SDG 11.11, which is the percent of population living in slums and informal settlements and inadequate housing. We’ve not had a good measure of this. There is currently a method that’s based on household survey and census data that kind of classifies what are called slum households. But if you look actually where these slum households are located, we find that half of so-called slum households or poor households are not living in slum areas.

0:23:45.7 Dana R Thomson: And the kind of needs of a poor household in a slum area are very different from a poor household not in a slum area. The relevant policies and interventions to provide wellbeing and opportunity to those different types of groups are vastly different. You might be looking at social safety net programs for at a citywide and national scale for those households that are very poor and very isolated from social networks, maybe living in better off neighborhoods. And you might look at area-level investments in infrastructure and services and really community-based interventions for “slum communities”, I’m putting it in quotes because it’s quite a derogatory term in some places. So we are looking to, so we are actually working closely with UN Habitat and other groups in this space to redefine how we even measure and think about this indicator so that it’s more accurate and usable for monitoring and decision making.

0:24:46.7 Dana R Thomson: Okay, so let me backtrack a bit and talk about what tends to happen in global modeling initiatives, at least in the space where we’re producing data for development or humanitarian response. You might have experienced this in your own communities of practice, but where I sit, I’ve seen a number of initiatives really start with the modeling team and thinking about what data are available, what’s feasible, and we are going to imagine what data, we’re gonna create some data and model it to solve a problem for someone else. And we’re gonna see what data is available, we’re gonna think about what’s feasible to make and what we can do at scale. And then we’re gonna pitch it to stakeholders to help them understand that they can use this data to solve their problem. And so this is how I would visualize that kind of workflow, which is really starting with what one team thinks is feasible, producing a set of models, iterating until you’ve got something that is produced globally, and then pitching that data to many different stakeholders to kind of solve their problem.

0:25:57.0 Dana R Thomson: This is what would be classically thought of as a top-down approach. And there’s a lot of problems with this in terms of uptake and whether or not these data actually get used. And in fact, a lot of these data, from my perspective, I see don’t getting mainstream into decision-making at city scale. They don’t get used by communities, even if they might be open access or available on GitHub or something. A lot of the data just become the source of a student project in the future or a global report that gets dusty on a shelf at some point. And so really not having the impact because, and really addressing the priorities that we intend to, the model are, people come into this workspace with very good intentions. They want to solve real problems. They want to provide power and insight for decision-makers in deprived or vulnerable populations. They’re trying their best, but this is not in a very effective way to approach global modeled data sets. What we were doing in some of the early IDEAMAPS projects is we’ve always had this sense of co-design and co-production. We are a super diverse team. We have people who hold very different titles and social power within our group.

0:27:21.5 Dana R Thomson: And so co-design and co-production concepts have always been at our heart and what we’ve always strived for. On the bottom, I have two rough examples of what you might see if you do co-production and the language that gets used here. Really on the left, the basic principles of co-design and co-production are participatory. It’s about power sharing. It’s about capacity building and working together all steps. And how that might happen is being shown on the right in another framework where you really iterate, building connections, aligning, discovering, designing, testing, implementing and learning and then hopefully repeating that process. So we tried to do all of this. We really did. We had one-on-one meetings with local government which was really around sensitizing and trying to build partnership with our project and try to get some voluntary buy-in and involvement. We had very good connections and still do with communities who are collecting data, using data very effectively for advocacy and internal upgrading processes and providing requested trainings on new tools and methods that they might find useful in their work and learning with them through training.

0:28:42.3 Dana R Thomson: We’ve held multiple multi-stakeholder workshops with a lot of thought about who is present, are we representing, mapping out kind of key stakeholder groups, identifying key people in the cities where we’re working, facilitating very thoughtful activities in those workshops, holding them early in our project, middle of the project, end of the project, making sure that we’re visiting the deprived communities that we are trying and aiming to model, making sure that our modelers have the opportunity to visit the places that they’re supposed to be modeling. So these are all things we were doing but to just map it back onto this framework and why we still weren’t seeing a whole lot of uptake by community and local users. This is what we think was happening. We think what was happening is that we were starting with these all-stakeholder workshops and really asking the question from the beginning to everybody in the room, what are your priorities? What data are you using? How might we plug in? Where are the overlaps and what are the sensitivities? That process really implies that everyone sitting around that table has an equal voice, that they hold kind of equal power to voice what they are thinking, to even understand what the question is being asked. The reality is that there are a lot of power inequalities in that room and the power dynamics that might exist, say, between community group members and local government representatives.

0:30:25.9 Dana R Thomson: There’s a whole history there that shapes what people are gonna say in that public space and how they’re gonna say it. There’s also the fact that these teams of researchers who include the modelers, our ears are the filter. Even though we have a very diverse team, which we try to learn, we try to listen, we’ll all sit in the same room with the same stakeholders, listening to their responses to the same questions and we’ll all walk away. The team members, we walk away with very different understandings of what just happened because we have different disciplinary backgrounds. We know different things about the local context. What happens is that these teams are generally weighted toward Northern countries, Western countries, academic, high income backgrounds and we don’t have the language. We don’t have the context to fully understand what people are saying from a community perspective sometimes. We’re not hearing what they mean. And so there’s a lot of problems actually with what we think is a best practice, which is an all stakeholder inclusive meeting that’s got a lot of rules around how we interact with each other, but by assuming that there is equality in that room, that everyone has an equal voice, we are actually not addressing some underlying issues.

0:31:46.6 Dana R Thomson: There’s actually a lot of inequality in that room and there’s things that we should be doing to address that inequality so we have a better starting point when we do get to an all stakeholder meeting. So we would start with our all stakeholder meetings, we would see what’s feasible in terms of modeling, we would bring it back to our all stakeholder meetings and try to think about impact and try to scale our models. But it was really only after our models had been iterated, being very centered in among our modeling team, that we actually really tried to bring it back to communities. And there’s still a pitch kind of involved that we’ve created this dataset that we thought like solved your problems or gave you the insights you needed. So here it is on a platter, let’s report it back to you, but it’s still actually not validated. It still might not be relevant. It still might not be useful. It still might not generate the kinds of interactions between say communities and city government officials that are really needed. We didn’t hear it well the first time. So that’s where our reflection has been.

0:32:56.6 Dana R Thomson: I think once we’ve been through this with a couple of projects and wanted to do better, and we’re really drawing, and we as a team are still developing our understanding of a lot of different communities of practice that use overlapping language to figure out how we can do this better and maybe provide some guidance, some options for other modeling teams, ’cause this applies far outside of just slum mapping. There are teams out there producing maps of graded population dataset, settlement maps, infrastructure maps, building footprint maps. There’s a lot of data being produced in global scale by models. And these lessons very much apply. So the overall framework that I started with and that we are kind of working with now and trying to articulate is drawing very much from a whole bunch of overlapping thoughts and processes related to co-production, justice, ethics in AI modeling, participatory processes, et cetera. And we’re combining it very much with lessons from the agile development community that’s really centering the user experience, so user-centered design, thinking about experimentation, rapid sprint cycles, lean processes where you iterate constantly.

0:34:24.3 Dana R Thomson: This community and the language they use is a bit agnostic to social power differences, but they do think of the user as somebody that needs to be learned from and that validation needs to happen iteratively, rapidly, and that they’re involved throughout whole processes. So both of these worlds provide a lot of ideas and language for us to draw from. The agile world has this concept of sprints and rapid deployment, and the co-production, co-design universe has this idea of multiple stakeholders and dealing with power across stakeholders and iterating. So this is how we’re operationalizing the double loop in our kind of team and our daily workflows in case it’s relevant to the work that you’re doing. First of all, our team structure is that we have different expertise in working groups. So there’s a team working on developing platforms, a mobile app website for actually doing fair data exchange.

0:35:28.5 Dana R Thomson: There’s a modeling team working on operationalizing models, and that’s one whole team. There’s a whole team focused on engagement. These are people who are academics who already have relationships with local government and community groups and are really facilitating a lot of, and helping to translate across disciplines and contexts for the rest of the team. And we actually have a whole stream focused to learning. This team will be thinking about training materials later in our project, but right now is actually focused on our internal learning processes, and it’s very much linked to the co-design team, which is really facilitating interactions across all of these groups. So what are we doing? Our team, this diverse team, global team across multiple continents, meets one to two times per month, very much the engagement team with our two technical teams, the platform and modeling team.

0:36:26.3 Dana R Thomson: And we bring in government liaisons, which I’ll get to in a minute. So that team, there’s internal learning happening all the time. And our modelers have also reached out to a broad community of practice of people who model for development, model development indicators, and they meet several times a year to exchange on methods and kind of overcome challenges around scaling, et cetera. What it was really important for us was to center communities. And the way that we’ve thought to do that is we work with two different communities in each of three pilot cities, supporting them to set up participatory action research group. These are groups that meet regularly once every month or two, where a specific set of say 20 people in a community define what are their priorities, decide what data exists already and what data they want to maybe generate, how they want to use that data to make change in their own community around the things they care about.

0:37:25.4 Dana R Thomson: We as the global team are involved to the extent that we can provide some basic facilitation, very high level facilitation and some resources. What we get out of that is learning a lot about the context, the language. We start developing our vocabulary to hear what communities are saying when we get to the multi-stakeholder workshops. And we really get to learn in multiple ways what priorities exist in the different community, in the cities and the types of communities that we care about.

0:37:56.0 Dana R Thomson: It’s obviously a small sample. We also have built relationships with local government departments and have received nominations of one person from each city who joins our team in the co-design workshops frequently. We compensate the local government department for that person’s time. So we kind of buy out some of their time to be part of our project from very early on and they’ll continue. And then we do hold our multi-stakeholder workshops, but we’re in a much better position as a team to hear what different groups are saying and be aware of the different dynamics and power inequities that exist. But we’ve also taken some steps so that the specific stakeholders who come to these workshops have greater equity. There’s a lot more agency when they come to sit at the table to speak their mind and to draw on the data they’ve collected.

0:38:51.4 Dana R Thomson: And they already can relate to our general aim of the project in different ways. So I don’t have enough time to talk about the added value but hopefully that comes up with questions. I hope it’s obvious. I’m going to stop here. That’s my contact information. Feel free to reach out. Check out the ideamapsnetwork.org. And again, here is the model that we are currently working with, kind of our first draft of it, what we’re calling the double loop process for participatory modeling. So I look forward to your questions. Thank you so much.

0:39:24.8 Julian Agyeman: Thanks very much, Dana. What a fascinating presentation. We do have one question in the question and answer, but I am going to take moderator prerogative. Two questions, they’re really quick ones. You use seemingly interchangeably the words co-design and co-production. Is there… Do you see a difference or are they part and parcel of the same thing? And I’m asking generally because, again, I float between the two, but generally co-production is what I use.

0:40:00.9 Dana R Thomson: I think I’m just going to close. I’m not sharing now. So I think that people who build these frameworks will say co-production is the general concept and co-design is one part of a co-production process. To be honest, I’m still learning a lot about the different vocabularies that we use. For example, I was based in the UK before I came to Columbia. I had not come across this idea of environmental justice, which is something that is a common phrase that’s used in a lot of disciplines here in the US. But for me personally, that was a new set of overlapping ideas. I’m getting familiar with that vocabulary, and I think that this is where we’re at right now in the state of our fields, is figuring out what some of the common language and concepts and approaches are, where we can apply that in our different disciplines more systematically, intentionally, and relate to each other a little bit more as we’re doing that work.

0:41:00.0 Julian Agyeman: Right, yeah. And the second question, again, just a brief one. I didn’t hear you differentiate between, say, quantitative and qualitative data. What’s the relationship between those? And I’m thinking especially perhaps with communities. What do you… I’m a big believer that story animates science. What is the relationship in different types of data in your work?

0:41:33.3 Dana R Thomson: That’s an excellent question. The platform itself and the kind of initial intention of the team is very much focused on quantitative data, which is the production of very detailed maps about citywide assets and deprivations and tiny little grid cells. That’s what we are working on. But our team has always included qualitative researchers and people who generate qualitative data, community members, why is that? Because we need to learn from each other in our team. If you’re going to do this kind of interdisciplinary work outside your team, you need to reflect that a bit in your team and you need to have a way of of integrating knowledge and exchanging knowledge and qualitative methods provide some of those tools. Right? The qualitative pieces is really helpful as well because there’s a lot of qualitative data that a lot of the data that comes from communities is qualitative. Some of it can actually be harnessed in a quantitative way and incorporated. And the beauty of modeling is that it provides…

0:42:49.7 Dana R Thomson: Modeling in general is a kind of way of integrating lots of diverse data from different types of data, different geographic coverages and scales. And so thinking about how do we take advantage and notice information that would otherwise not be kind of on the radar for quantitative data scientists.

0:43:09.8 Julian Agyeman: All right. Okay. We have a question from Alice Song. Alice asks, so how can community members be compensated equitably to enable their capacity to liaise and participate with the PAR and government teams?

0:43:25.6 Dana R Thomson: This is a question that comes up all the time, and we care a lot about this within the IDEAMAPS network and community. I can talk about this one project, the data ecosystem project, but there’s also some spinoff projects under the umbrella of IDEAMAPS that are thinking on this issue of compensation. Within the context of this particular research and data production grant, the compensation is for stipends, so travel costs and basic stipends for someone’s time to come to a meeting from… In a community participatory action research group meeting. There’s also funds available to collect data and act on that data so that the teams can decide how they want to use that money. But what happens all the time is that community-based collaborators are often viewed as people who collect data for us.

0:44:24.7 Dana R Thomson: They can extract data efficiently for us and that they’re cheap. They always get paid on different scales and in different ways than so-called experts who have institutional affiliations and fancy titles. And we really like to change that. Me and many of the people on this call will have had the opportunity to progress in our careers because we had training that acknowledged the skills we had before and gave us something to build on and gave us credentials. Community members get hired into data collection activities all the time and they get exposed to the same training again and again and again. They get paid the same amount after doing the same work for 15 years. They might get a bump up of a few dollars a day. It’s pretty gross, actually, the way that we just kind of, we don’t have those ways of building up that knowledge and expertise and they don’t compensate it. So one of the spinoffs of IDEAMAPS is working with a group in Nairobi called Community Mappers, which is trying to, and is successfully creating a model that’s more of like a, creating a model where they have the expertise to design research and carry out research, and they kind of contract to academic teams and have, they charge rates that are more commensurate with what a graduate student would be paid.

0:45:51.5 Dana R Thomson: And they provide real value, and they… We work with them to build co-design training materials that are relevant for a low-resourced context that are modularized, that can be done in a day without a ton of computing resources so that they can build on those skills and methods, they get certificates through this process. And there’s a lot more that can be done there but we have started that work and that conversation matters a lot to the people in our group. It is a great question that people are asking here.

0:46:23.6 Julian Agyeman: Great. Thanks. We have a question from Claire Riley. Thank you so much for sharing. This is an incredibly interesting and comprehensive program model. Are there links to NGOs, charities, local government and other groups who can use this data and collaborative group findings to offer support on the ground to these communities?

0:46:44.0 Dana R Thomson: Yeah, a lot of… So there’s a concept of boundary organizations. Again, another term that’s a bit new to me. Our funder, Gates Foundation has really introduced us to this concept. They have done research or contracted research to say what makes for effective data impact and use. And the presence of boundary organizations, something like an NGO whose mission is to advocate and elevate the voices of vulnerable populations are usually a very important element, part of the equation. And yes, working with… Involving the NGOs that do that kind of work is really critical. They’re always present at our kind of all stakeholder type meetings. Our initial meeting is always very open. Like we just do an open public call that people can Zoom in or apply to attend in person, those workshops. We try not to make it closed. We are always looking for hey, do you see opportunities to share data and resources and stuff? So I think that as potential data users and translators, that would be the role that I see NGOs playing at the moment. But also if there’s feedback here in this group drawing on your experiences it’s, I’d love to hear what your thoughts are on that.

0:48:11.5 Julian Agyeman: Yeah, I’m sure you’re going to get several emails from people in the audience. Paige Kelly from Shareable says, “This model is very inspiring. Thank you for this great talk. I’m based in Syracuse where there’s a big debate in the city about what to do with land that will open up when the highway that runs through the city is taken down.” The community that lives around the highway is pushing for community land trust for the land. The city, New York State, has said that they will make a financial decision. Sorry, the final decision, and will listen to folks. But it does feel like a bit of a closed-door process. Do you think the IDEAMAPS model could be relevant here and with the debate over what to do with land that opens up in a city?

0:48:53.4 Dana R Thomson: Yeah, I wouldn’t actually claim that the IDEAMAPS… This is IDEA… Like I said, we’re drawing from kind of rich traditions, overlapping traditions that provide, I think, some guidance about how groups might come together around an issue like this. And that co-production, co-design framework is really helpful. I think what’s always challenging is you really need facilitators, right? You need decent facilitators. And that’s maybe to the point that the previous person mentioned around these NGO groups often provide that role and that service for stakeholders who have different priorities in competition with each other. NGOs can be an arbiter and academic teams can also play that role. So depending on where you sit, maybe think about this kind of facilitated community forums and having to build the relationships with community and the relationships with the state government. Independence, it’s really, data exchange in our mind is really about trust exchange. And it’s that trust building. So data is just one vehicle for building trust. When you can exchange data, you’re kind of starting to build, you’re also exchanging insights and trust. And so data might be a vehicle for doing that, it might not be in that particular case.

0:50:27.9 Julian Agyeman: Thanks. We have a question from Mark Roseland who’s a professor at Arizona State University. Fascinating presentation. Thank you. While this double loop process addresses the inadequacies of conventional approaches re-stakeholder and community inclusion, I wonder if it goes too far in the other direction by not engaging city managers earlier in the process. Your process seems to have a big risk of efforts not being implemented because of this, therefore demoralizing all the efforts involved in doing this work. What has been your experience in terms of implementation?

0:51:03.6 Dana R Thomson: I think what I’ve observed throughout the IDEAMAPS project is if you really, if you start the process with those power holders at least in the case of “slum mapping”, what you end up with in a, I’m using again slum mapping in quotes, but what you end up with is a real ethical risk because the priority of city government… The most efficient way to get rid of slums is to knock them down, right? And that’s often what we’re measured and that’s what city governments think they want to measure is where are the slums because they’re responsible for monitoring them. A lot of our metrics like SDG 11.11 define that. There’s a lot of problems with just providing that data. High resolution satellite imagery and modeling can do that. It can do that pretty well at scale, but that puts a lot of… It creates a lot of ethical dilemmas. And I think if the issue, the phenomena that you’re modeling, if there are vulnerable population involved, I think it is absolutely necessary to start to center them in that process.

0:52:13.9 Dana R Thomson: I wouldn’t by any means diminish the importance of involving local government from very early days and maybe even start with both community and local government, maybe separately at first until there’s some language and some sensitivity is developed on your side as a facilitator to figure out where and how we bring them together, right? So it’s not to say local government doesn’t matter but when you have vulnerable populations involved, they have to be centered, otherwise you have huge ethical risks, very unintended consequences and it’s… We see it in the world of slum mapping, very model data-driven approaches that have produced maps that become basically, if it’s not the local government that’s going to use the slum boundaries to knock down whole communities, it’s the private sector, it’s the landowner, right? There’s a really complex web of ownership for buildings and of land and who gets to basically decide to knock things down. And so somebody will nefariously use that data if we didn’t start with vulnerable communities.

0:53:22.3 Julian Agyeman: Right, okay. Leslie Wilson is a retired emergency manager from DC, working globally as a vulnerable population specialist independently for the last 12 years. He says, “My slums are liberated Ukrainian villages. After 18 months, I finally got permission to move into a regional administrative town between Kiev and Chernobyl. Your approach would have great potential for breaking through barriers I keep running into. Pitching my resilient village concept is a tough uphill slog. Recently, I connected with the ICRC to bring landmine awareness training to some of my villages and towns. However, connecting with all levels of government is necessary but extremely difficult. Would love to find a university interested in adopting my region and help to get the attention of Ukrainian universities, government, et cetera.” More of a general point, but what about disaster areas. What about them?

0:54:25.0 Dana R Thomson: I think that I’ve forgotten his name, but this…

0:54:28.3 Julian Agyeman: Leslie.

0:54:28.5 Dana R Thomson: I think most of us who work on long-term development topics would appreciate that these are the areas that are… Areas that are vulnerable in the long run are also the areas that are usually most vulnerable in the short run to things like climate disaster or conflict or things that caused… They’re more vulnerable to disasters in the short run, which only compounds that long-term vulnerability. And so in areas that are prone to disaster might also be areas to focus on for those long-term development relationships. The partnership with local academia or NGOs or community groups, community-based organizations, which are sometimes formalized and registered with the local government, but sometimes are just a Facebook group or something, right? It’s just a group of people that come together around something that’s really important to them. Those types, establishing those kinds of networks and relationships sooner than later when you do in a disaster-prone area can go a really long way to mobilizing data and information. Data, and when I say data, I mean it’s inclusive that qualitative experiential information, that intergenerational information, et cetera.

0:55:55.7 Julian Agyeman: Well, Dana, I could go on asking you questions for a long time, but we’re coming up to one o’clock. If anybody’s got any final question for Dana, please, you can email her. I think Dana wants to make connections and spread this IDEAMAPS idea. So let’s be in touch with Dana. Again, Dana, thanks very much. Can we give Dana a warm Cities@Tufts thank you for the final colloquium of the season. Thank you, Dana, so much. Thank you. And we’ll be back with a fantastic program again next semester for all of those regulars and hopefully for some new people as well. Thank you, Dana. Happy holidays, everybody.

[music]

0:56:38.7 Tom Llewellyn: We hope you enjoyed this week’s presentation. After taking a few weeks off, we’ll be back with our next lecture in late January. Stay tuned for updates. In the meantime, we recommend scrolling back to listen to past episodes like Punitive and Cooperative Cities from Dr. Stacy Sutton in Season 2 or Public Space, Paradoxes, Possibilities, and Propositions from Dr. Vikas Mehta in Season 3. Cities@Tufts is produced by Tufts University and Shareable with support from the Barr and Shift Foundations and listeners like you. Shareable is in the middle of our end of year fundraiser and could really use your support. If you appreciate the show and are in a position to donate, please click the link in the show notes to make a contribution today. Our lectures are moderated by Professor Julian Agyeman and organized in partnership by research assistants Deandra Boyle and Muram Bacare. Light Without Dark by Cultivate Beats is our theme song. Robert Raymond is our editor.

0:57:34.0 Tom Llewellyn: Additional communications, operations, and funding support are provided by Paige Kelly, Allison Huff, and Bobby Jones. This series is produced by me, Tom Llewellyn. Please hit subscribe, leave a rating or review wherever you get your podcasts, and share it with others so this knowledge can reach people outside of our collective bubbles. That’s it for this week’s show. Here’s a final thought.

0:57:56.3 Dana R Thomson: And just underscoring that co-design and scaled data are actually complementary, they can coexist. But how we do that in hearing the kind of voices and grounding our data in the multitude of experiences of people on the ground in very different types of communities, even within communities, isn’t easy, but we’ll talk about how we’re thinking about it in IDEAMAPS.