Episod 21: Max Malynowsky - Offline is the New Online

What if humanitarians had an offline-first mapping tool as reliable as a Garmin GPS? In this episode, Max Malynowsky — software engineer at the OCHA Centre for Humanitarian Data — dreams out loud about a future where field teams can sync trusted, up-to-date geodata anywhere, even with near-zero bandwidth.

From the chaos of contested admin boundaries to the quiet genius of ODK and XLS forms, Max and Hugo unpack why the hardest part isn't building the app — it's building the data infrastructure behind it. If you've ever tried to print 20,000 settlements or wished for a universal translator for geodata, this one's for you.

Links: 

Max's LinkedIn

HDX

OCHA Centre for Humanitarian Data

Transcript

SPEAKER_01 0:01

Hello and welcome to 15 Minute Maps, where I ask my guests to let their minds roam free and come up with a new idea for their dream map. The first known maps of the world were created thousands of years ago, just a flat disk surrounded by water. And today we are mapping the furthest reaches of the known universe. In between lie a myriad of mapping possibilities. What if we could do away with resource limitations? Think beyond the conventions of time, space, and political boundaries. What kind of new maps could we dream up? Joining me today from snowy Amsterdam is Max Malinowski, software engineer at the OCA Center for Humanitarian Data. Previously, he's worked for the Canadian Natural Resources Center, the World Food Programme, and Space for Good. However, if you've ever used a humanitarian boundaries layer, then it's more than likely that Max had something to do with it. Max, thank you so much for joining me today.

SPEAKER_00 0:55

I've been trying to keep up with uh all the guests that you've had on recently. You've got some great people coming on.

SPEAKER_01 1:00

Well, you actually got a shout out uh in the Sven Schmitz-Leuffen episode uh when we were talking about the gap map. So I'm glad that we finally get you on in the new year. Now, um, for the last couple of years, you've been working on a little side project, I believe, called Field Maps, and you've been feeding in almost directly to the OTA and HDX uh humanitarian data platform. Um you now work for OTRA, so I'm guessing you're doing that in official capacity now. But I was wondering how do you navigate the broad array of pitfalls and controversies that surround you know such a sensitive topic as global admin boundaries?

SPEAKER_00 1:39

Yeah, um it is like that's a it's a really hard space to um to try to find a solution for. And I think it went like it's been going on for years um with um different kinds of approaches coming, um, with people coming up with like different kinds of approaches. Um ultimately, when it comes down to it, um if you're making a map for a particular country, that country um decides for itself how it wants to represent its own national borders. And frequently that conflicts with the way that its neighbors also do. Um, so if you're in a particular country, um, their their point of view is really that it like that is the correct one. Um it's when you start looking at like a regional level or a global level, it becomes really difficult. Um and the concept that really helped me kind of wrap my head around that is this concept of like worldviews, um, where like different nations will recognize different dividing lines at international borders. Um, if you're the United States, you have a certain kind of foreign policy that um uh sets um like certain boundaries around like say Kosovo or uh around Morocco. Um and those lines may not be recognized by other nation states. Um there can be like a US world view that's very well documented in uh a project that they have called the Large Scale International Boundaries, um, which is kind of the starting point for the field maps, international boundaries that I used. Um, and as close as possible, like I tried to use um somewhat of an international consensus um around um the final version that um I made for that project. Um but there is no single worldview that you can say, like, this is the worldview. It really kind of depends on your use case.

SPEAKER_01 3:38

I mean, that actually sounds like an interesting map by itself, is you generate a map which somehow represents every single uh varying worldview of a global boundary where you probably have overlaps almost like country Venn diagrams.

SPEAKER_00 3:52

Well Yeah. There's a there's a really oh the really interesting um visualization that uh Mapbox makes. They make a product called their um their Global Boundary Explorer. And if you go on to their site for that, um you can just like flip a switch that shows like this is the US point of view, or the Argentinian point of view, or the Chinese point of view. Um and it kind of helps you a little bit understand like how those particular nations view the world.

SPEAKER_01 4:18

I mean, that sounds incredibly fascinating. I'm surprised I hadn't heard about this uh this thing that Mapbox have done. Well, look, Max, I'm super interested to hear what your map idea is because you've been in this realm for so long. If you're ready, then I'll count you in in three, two, one. Over to you.

SPEAKER_00 4:36

Yeah, um, so I've been thinking about this, and um I've spent a lot of time in the field with my time with Reach Initiative. Um, and the one thing that I really would have loved to have um during my time there would be an offline first app that could connect to any platform through open data standards, um, no matter like where the source is coming from, and receive timely and relevant updates to synchronize humanitarian map data for an offline environment with like very little bandwidth. Um, the way I'm kind of thinking about this is it would be kind of like a humanitarian version of like a Garmin GPS device or like an offline hiking map. Um, whereas like there, I think the outdoor industry is really mature in the way that they kind of handle using map data in the backcountry and offline. Um, and like these devices and apps, they can connect with so many different platforms for trails and routes, places like Strava, AltTrails, Kamut. Um, and you can just get this kind of like there's a whole ecosystem of like of routes and trails and uh like water points and campsites. And you can get all of this by just like you know, like clicking a synchronize button or like a download data set. Um they have things like you can access satellite imagery, weather forecast, wildfire alerts. And like I just like I think like the like the capabilities that like the outdoor industry has, like, it's so rich and it gives you so much information to create like informed decisions in um like in that kind of environment that you found find yourself in. And I would love for like the humanitarian environment to have a kind of tool like that that's so easy to use and works just as reliably. Um, like in our particular context, um, there's like more that you need than I think just like trails and water points. Um, there's things like administrative boundaries, operational presence, settlement locations, road conditions. Um, I remember in like episode three of your podcast, um, there was Nathaniel Raymond talking about like a standardized method of like gathering, storing, and like sharing data between organizations into like a common operational picture. Like, I would love if there was like an app that could just synchronize all of that, where you could just like take all of that with you and trust that that was like the latest and up-to-date data before you headed out to the field. So, like that's kind of like the that's kind of like what my dream map would be.

SPEAKER_01 7:05

So you mentioned that the outdoor industry, obviously, I mean, they've been going at it for years and years and years. I remember being a Boy Scout as a child and having the scout leader walking around with a giant GPS device, collecting the data and collecting the information, which we can do now on our phones, you know, no problem. But that data is relatively fixed. The trails have been created, the water points are there. Given in, let's say, a flood situation, or even in a conflict, and actually talking of conflicts, how would you go about dealing with those kind of conflicts in the collection of the data? One person can collect information in one sense and another person in a different sense. How do you marry those two uh differences?

SPEAKER_00 7:49

I think like we the two those two categories that you described, like conflict and and flooding, um, I think both of those are a little bit distinct. Um like one of them is um uh is is somewhat sense, it's usually sensitive data. And um that's the kind of data that may not be publicly available in a way that you would like find openly available on a platform. Um, but things like flooding should be. Um, like flood data, like there's a lot of models out there that produce different kinds of results. Um, I think we're still in the process of um taking all of these global models um that produce these kinds of data sets and um coming to a kind of consensus on um exactly what works best for each country. Um, and then um kind of like sharing that all in like a very similar way across the world.

SPEAKER_01 8:44

So not long ago, we were working together on a project with UN DSS and it was an online and offline map. Is this something that you've kind of had brewing for a while? Because you you also mentioned that you had been working in the field with Reach. Is this really like a pipe dream, or is this something that uh is genuinely achievable now?

SPEAKER_00 9:05

Um I I think there's still a little bit of uh I think it's still a little far out. Um I think like there's there's I think like two parts of this. Um there is like there's the shiny app itself, um, which I think like we're we're getting pr quite close. Um there's some really interesting tools out there um from like um projects like MapLibre, which are giving tools to create your own open source uh map apps with. Um but as I think what we're a little bit farther from is the infrastructure to power that. Um, and kind of like creating like our um like to use like a transit analogy to like to build the roads and the rails, um, and then put like cars and trains on them in order to deliver this kind of data with like a really robust and efficient network. Um so that's kind of like what brought me to HDX is to like really work on like the infrastructure and like the ecosystem behind it. Um so I think like that's the that's the hard work that um we're still um we're still trying to make some progress in. Um to go back to like the um uh like a particular um experience that I had with REACH. Um so I was working um in REACH doing a nationwide survey program um in major population centers, asking people who had just arrived there what conditions were like in the places where they were coming from. Um so we called this like the hard to reach program, and this helped us inform, um gave us like early indicators for things like hunger or um sanitation issues. Um so often like the places these people were coming from were like really small settlements, and we had a lot of difficulty like translating this description of where they're coming from into like something we could just like plug into a map with like a latitude-longitude coordinate. Um, we had this like really big database of like 20,000 settlements that um that map action had put together from a whole bunch of different sources. And like you couldn't find this on OpenStreetMap because it still needed a bit of cleaning. Um, this is kind of like a bespoke data set that was like quite good quality, but didn't really kind of exist in like a formal platform. Um so we tried like printing this out, but um like 20,000 points is like a little bit difficult to put on paper. Um, so in the end, like at the time, like I was a little, I was quite well versed in using web technologies. So I built um something that's now called like a progressive web app that let um the web browsers of all of our field officers download a copy of the website in which they could like work in an online map in an offline environment, including all of the data behind it. Um so this let them have kind of like a static um copy of that map. Um and they could just like um like without needing any kind of experience using really complicated desktop software, they could just open up the URL and just browse any location in the country and have really useful reference data freestyled, like roads and rivers and administrator boundaries. Um, so it worked well as like a proof of concept, but it was quite difficult to maintain over time to push updates to. And that's really, I think, the hard part is like um making it scale to other countries. Um, like when we had some other country teams ask, like, if could we have a map of that in our locations? Um and I kind of looked into it, and like the hardest part is like the data for those other countries are in completely different formats. And so like it was really easy just to make a copy of the map for another country, but then to get like the data from that, from like those countries into the app was really hard. Um, and here at HDX, um, we have a lot of activities that uh standardize data. Um, Happy is our humanitarian API um that um both programmatically provides data, but also does a lot of standardization services um to clean up data. Um and that's a lot of work. Um, so like that kind of like standardization and service provision, like that's the really hard problem that we're still in the process of solving right now.

SPEAKER_01 13:20

So you mentioned HDX again, and to go back to the Sven episode, the ICRC episode, once again, it kind of feels like what the humanitarian cluster system is missing is a data cluster with a certain organization at the top who can decree this is how it should be done. Why can't that happen? Why is that such a difficulty? Or does it exist already?

SPEAKER_00 13:46

Yeah. Um I I'm not a big fan of like the one-stop shop concept where you kind of take all like you take a copy of data from um a whole bunch of different places and kind of consolidate them in one place. Um one for like two reasons. Like uh one, it kind of creates a single point of failure. Um and two, usually when you create a copy of something, it's almost like instantly out of date from the source. Um and I really like a federated approach to doing things where um you can have different sources of truth that all kind of communicate with each other in the same way. Um in the in the geo world, um, that kind of like connective tissue um is emerging with um the standard called stack. Um, and on the open web in general, um, RSS I think is a really good success story on how distributed um data syndication happened. Um, like in the early aughts, um, there were like a lot of engineers trying to figure out like how do we push updates from a whole bunch of blogs and websites in a way that people can just keep up with. And um in the end, like RSS wasn't a really complicated standard. Um, and it had the backing of the New York Times, which I think like pushed help pushed it over the edge. Um, and then now like almost all podcasting is powered by this standard. So you don't have to have like a single source of truth for podcasting. Um, it comes from all these different sources, but they all communicate off the same standard.

SPEAKER_01 15:17

So it's almost like we would need some kind of globally accepted universal translator for geospatial data.

SPEAKER_00 15:25

I think more than like some kind of translator, I think we need like a little bit of like a paradigm shift. Um like there's there's a really great article pushed out by development seed called um like rethinking crisis data 2.0. Um and it kind of frames that we're we're kind of stuck in this. Um for a long time, we'd be kind of stuck in this like desktop-oriented way of doing things. Um, like you, you just like you get your files from somewhere and put it on your computer and then you do something with it. Um and it I I think that like we there's this shift to doing things um to like using data in the places where they are stored. Um and I think like this requires, I think, like rewiring a lot of different habits of like rather than just like, no, I need to make a copy before I can use something to be able to use something directly from where it's stored. Um, like if you wanted to watch video a video online or listen to music online, you don't download a file and then wait till it's finished downloading and then open it up to watch it. You can just click on a link and then you just kind of start watching something. Um, and then everything kind of just streams in the background. Um, but that's not kind of one of our habits. Although when it comes to especially working in the field, um that's not a habit that I think does work um given the way internet can be really difficult to come by in those places. So I don't know.

SPEAKER_01 16:53

Yeah, I mean, borderline impossible to do in the field. Well, Max, look you have one minute left. Do you have any final thoughts that you want to leave the audience with?

SPEAKER_00 17:03

Yeah, final thought. Um to get back to that map app that works offline. Um, I think like a a lot of this conversation kind of like um has been pointing to the fact that it's kind of a hard problem to solve. But I think we have solved this problem in reverse. Um, when we look at kind of like not looking at the consumption of data, but looking at like the production of data, so like data collection. Um, I think we've got a really great success story um and that exists in the version of ODK and Kobo. Um, like the open data kit has been, I think, just like low-key. Like it really runs the, it really powers a lot of the humanitarian ecosystem. Um, they like the community has come together and has adopted the this standard called like XLS form, um, which is like it's not perfect. Um it's kind of just like a it's a programming language that you can use in Excel. So it's like it's really accessible to people who don't know how to code. Um and I look at that success and see like all of the teams that I've used that have like deployed and used it. Um and I think like it really, it really shows like that we can like solve these kinds of hard problems. Um, but sometimes like these are problems that we do need to like solve ourselves uh by building kind of our own tools that really kind of like work for our particular contexts. Um so I I look at that and I see like it's I think I can think we can like we can keep doing we can keep doing like uh so I I look at that and I think that yeah, we can keep doing cool stuff. Um it just requires resources, it requires us to be really innovative and um to to try new things. Um and yeah, I'm excited for um what uh what the future could look like um in like five or ten years from now.

SPEAKER_01 19:06

Max, your time is up. Thank you so much for your wonderful insights. We did stray from the original chatting path, but that's kind of what I like about the conversational style of this podcast. Now, if people want to learn more about what you do, or if they want to get in contact with you, what's the best way to do that?

SPEAKER_00 19:22

I'm sure you're gonna leave a LinkedIn link somewhere in the show notes. Um and then um uh I'm working at the Center for Humanitarian Data. Um, so um you can uh do a search for us. Um, and then um all of our data is on HDX. That is uh that is the lifeblood of what I do, everything on there.

SPEAKER_01 19:42

We'll have a link to HDX for those people who don't know what it is as well. Max, have a wonderful day. Go out there and enjoy the snow, and uh I look forward to seeing what you do in the future.

SPEAKER_00 19:53

All right, I hope we can catch up soon sometime in uh sometime in Geneva.