All right, so before I introduce our eminent speaker, I just have a few announcements that I've been given. So this is the iPad, GVU lunch lecture. So I just want to get an idea of the people that are here. So, can I get a show in for more accredited students? Okay, just leave it up for just a little bit. And then other students that are, I guess not accredited, Vibes, excellent. And then, interested guests. Hi, excellent. Thank you. A reminder that students that are taking this for credit, you must sign in by your QR code or manually. We're about to have a really fun, engaging talk from our speaker. So please, talk to your phones. Talk to your trash before leaving. and another reminder that meals are for our lunch lecture attendees only so don't dine and dash sitting right there tripping anyone before I introduce Alex I also want to announce that next lunch lecture next Thursday is Alex there's three of us there's at least three of them different speaker of health care and And we'll hear about that next week. So it is my absolute pleasure to introduce Duncan Alex Cabral. Alex is, as well as, a postdoctoral fellow in the Palomoba lab here at Georgia Tech. She is also an incoming assistant professor very soon in the Department of Urban Studies and Planning and Institute for Data Systems and Society at MIT. Not the one in Michigan, but the one in Massachusetts. A little old university that's come to get. She has a life-full experience, and I'll cover a little bit of it. A few highlights. She got her PhD from Harvard. She also spent time at Microsoft Research and Microsoft Xbox. She also was a high school teacher for STEM and math, if I remember correctly, as well as when she joined us, it's very important that leads a very large part of our lab focused on environmental justice, understanding environmental racism, of building out sensors and really working for us in the communities. And she has won many awards. She's going to go on to do great, amazing things like her to pay attention and learn as much as possible. So let's thank you for taking the help. Thank you, Josiah. It's great to see so many of you here, so many friendly faces, unexpected friendly faces, new faces. Really appreciate it. So yeah, I will dive in and try to leave some time for questions at the end. What this will look like, the motivation for my work, general approach, the three sort of subparts that I think make up that approach, and then this new section that I'm playing around with at the end. So the motivation for this work comes from a lot of things, and I always like to think about where we are locally. And so a few years ago, this was a news headline in Atlanta about a lead-tainted neighborhood becoming a major Superfund site. Now, a Superfund site is a really dangerously polluted site. Usually, it used to be some sort of manufacturing site, for example, and there's thousands of them around the country. But actually, a relatively small percent get to be these national priorities list. So, this makes this particular Atlanta neighborhood special in a bad way. Now you may wonder where this is it is shockingly close to us. So here we are I'm not tall enough but there we are Georgia Tech and this yellow diamond taken from the EPA Superfund map shows where that lead site is west site lead. Now next to this over here I got a map from this fun little site called census dots where it shows a dot for every person with a color based on their race. And you can see that it is pretty much surrounded by all black neighborhoods. You can also see crazy segregation. We'll see that a little more later. So this is West Side Lead, a super fun site, very, very close to where we are now. And now this isn't just an Atlanta issue. So I looked for an example in another city. Our friend Moy can't be here, but he's from Macon. And I found that there's two of these national priority list Superfund sites on the outskirts of Macon. And perhaps unsurprisingly for many of us, again, these are located near majority black neighborhoods, while the white neighborhood gets to be pretty far away. And this issue is way more than just Superfund sites. So for example, here's census dots for all of Atlanta. One thing you can see is how crazy segregated it is all black neighborhoods here all white and others up there and now here is a map of food insecurity and so this is based on the city health dashboard they took information from the CDC and mapped the percentage of families that are facing food insecurity and you can see that almost all of them are in that same area where there's mostly black neighborhoods and we see this another things. So fresh produce, for example, is not equitable across the city. Health is a big one. This is looking at the percentage of residents with obesity. Again, you could see in primarily those neighborhoods. And even as crazy as life expectancy. So your race can affect where you live, can affect the food that you're given, the soil that you're exposed to, the air that you're exposed to, and then even how long you are expected to live. And if that is not crazy to you, I don't know what to say. I think it's absolutely insane. And it's clearly based on these historical redlining issues, what we call in the U.S., where certain races live in certain areas and are then exposed to way more negative things than others. Now, all of this has sort of culminated in a movement which many refer to as environmental justice. And there's a lot of definitions for this. I pulled one from the American Public Health Association, which is really just environmental justice, the idea that people and communities can live in safe and healthy environments, that it should be equal, and that they should be involved in those actions. Now, a big thing in actually moving towards environmental justice is that we need data to do so. So people need to be able to, quote unquote, prove that they live in a neighborhood that has bad soil, that doesn't have enough food, that has poor air, etc. And with these data, people can then, one, identify the issues and the locations where it's happening. Two, hopefully inform policy, whether it's at a local level or a national level. And then three, determine what sort of action may actually take place. And there are few, but there are success stories where this actually does work to some extent. So one example is in Flint, Michigan. Many of us may remember the water crisis that they had. And even though this is still ongoing, the community was able to quote-unquote prove that they were exposed to poor water. They were then given bottled water, and now actually a lot of the lead pipes have been replaced. In addition, a much more recent example is that some of us have heard of the XAI facility that's being deployed in South Memphis that a lot of environmental justice groups are upset about because it is a predominantly black neighborhood. And the Southern Environmental Law Center, which has an office just down the street from here, they are now going to sue Elon Musk and XAI because they're using gas turbines, which are not allowed. And so again, having these data can help, but it can be difficult to gather. And regulatory monitoring is not enough. So for example, this is a regulatory monitor for air pollution. It is large. It is expensive. When I say expensive, I mean hundreds of thousands of dollars. It requires special expertise for maintenance. The data might come in hourly or daily rather than continuous. And perhaps most importantly, these are often placed far away from polluting sources. And it might actually be that they are purposely placed far away from polluting sources, but that's a conversation for another day. Now why, or what does this look like? So interestingly, for example, in Atlanta, just about six months ago, there was a report that the air quality is getting worse, especially for ozone and particulate matter. But if you look at where we have regulatory monitors, they're so tiny, they're very tiny dots. You can see that we have at Georgia Tech, so we monitor ozone particulate matter here. We have one there, which I believe measures ozone, and then we have one up there in Hills Park. And so interestingly, those all say that the air pollution is good, but if the American Lung Association is telling us that the pollution is bad in particular areas where we don't have monitors, then how are we supposed to know what's going on. Now, another motivation, and this is depressing, is that regulatory monitoring data might not always live forever, right? So just this year, the Trump administration deleted a lot of this data from federal databases. We were fortunate that researchers and advocacy groups created their own backups, but this really points to a need that communities have for more than just regulatory monitoring. Now one thing that has started to serve this gap is hyperlocal sensors. So for air pollution, the most common one is called purple air. It's a small sensor that you can purchase for about $200 to measure particulate matter. And these have started to become quite popular because they are much lower cost, they can connect wirelessly, you can put them anywhere, and they collect data every minute. This is an example of purple air in Atlanta. So I captured all the outdoor sensors. And you can remember that EPA, we had the three locations. Here we have many more than three. So it seems successful. But the problem with these sorts of sensors is one, they're designed by engineers. I know that's probably a sore spot. That's a lot of us. But engineers aren't always the best at thinking through all of the community implications and all of the needs that communities have. Another is that they have limited or no maintenance. So they're like closed box systems where if something goes wrong, consumers don't know how to fix them or what they might do. They don't give a lot of visualizations. You usually just get a dashboard that's like here's your plot, but you can't really get granular into the information. You rely on users telling you where they place them, if they're outside or inside, which can of course cause some errors. And the biggest issue though is that they often show up in rich white neighborhoods. And so we're really then just seeing that these devices are not being placed where people need them, but rather by the people who have extra money to spend to put a device outside in their yard. Now, another interesting thing, if we think back to the West Side Lead example, is that this Superfund site was placed on the national priority list because an Emory University doctoral student gathered data and shared it with the EPA. And then a few years later, the EPA is now doing sampling. And so we can see that if we can gather data, even we as students can gather data and work to show it to the people who have power, we can make a difference. So ultimately, this all leads into my broader research question of how we can design these sorts of sensing systems with and for communities who are really facing the most negative effects of environmental injustice and racism. So the broader approach that I take, which I've referred to as co-designed hyper-local environmental sensing, really it pulls from a lot of different areas, primarily around sensing and cyber infrastructure and HCI, and then combining theory applications and engagement in these three sorts of areas that I look at, designing the hardware, designing the deployment, and designing the interfaces. And now it's important to take like a sidestep here just to talk about community engagement, because I think it's something that we throw around a lot, but it's not easy. And I want to be clear that a lot of the work and a lot of the talks we see where people are doing this, it sounds like, oh, I'll just go out and engage with communities. Great. This takes time. It sometimes takes years to build these relationships and to maintain it. It takes a commitment to sticking through with that relationship. A big one that I think Carl mentioned a couple weeks ago, too, it takes willingness to be a novice. And so we are not experts in living in these areas necessarily. So we need to hear from the people who are. And a big one is that building on existing relationships. So a lot of the times other people in our community may have relationships, and we need to then build onto those in order to engage with communities. Now, it also comes with several challenges. A big one is around pacing. So in academia, we often have deadlines for papers, for grants. We have students who are leaving. But communities are facing these issues, and oftentimes they take decades in order to actually address. Another one is trust. Particularly with researchers and with these communities, of course, there's been a lot of exploitation. There's been a mistrust based on what governments have done, what people have done. And so acknowledging the issues of trust and taking time to build it is super important. And then finally, there's this whole power hierarchy. The most interesting example I like to refer to is like IRBs and consent forms, for example. So you may be working with a community and thinking, oh, well, I'm going to turn this into a great paper. And then you'll do a study and interview them. And once you bring about a consent form, though, it really introduces this power dynamic that perhaps you didn't want in a community engaged scenario, which isn't to say that we shouldn't do it, but which is just to acknowledge that this is a challenge we will face. Okay, so back down to the main road after our detour. I want to take you through these different parts of my research approach and what that has looked like through a couple of projects. So the first is thinking about co-designing the hardware, which really constrains what data we can collect. So for this example, I want to tell you a bit about manumen, which is a wild rice that used to grow abundantly across a lot of the United States. It grows in the water, and it's a grain that is important to Native American communities who have been harvesting this for generations, and they rely on it for food, for the economy, spiritual and cultural events, etc. Now, there's several threats to manumen, including, of course, climate change, human use of water and land like boating and urban development, even birds for example and populations that can change based on climate change and human actions such as mining and pipelines. And so this image shows all the lakes and rivers with wild rice as yellow dots and then a proposed pipeline and a red line that would go through so many of those. Now monitoring this wild rice, though, it's also very difficult. So if you've ever been to Wisconsin, Minnesota, Michigan, what you may know is that these, I think Minnesota is literally called the land of 10,000 lakes. And so these are places that have thousands of bodies of water. So thinking about how you would actually go out to all these bodies of water, many of which are remote, is very difficult. It's time intensive to grab things like water quality samples or to look at the abundance of the wild rice. And even though there's satellite imagery, of course, it's pretty low resolution. You might not be able to really see what's going on there. And so an approach that we've taken is to gather this hyper-local environmental data, data where the wild rice actually is or may be, to sit within the bed. And these we've called macaque, which is a Native American Ojibwe word, the tribes we work with, for containment. Now in designing this hardware we first started by learning from field researchers and natural resources. So one of our grad students Eric spent a summer up north working with a lot of tribes and natural resource departments to learn how they do their work and how these sorts of sensors might fit in with their practice. We then built a pilot, which looked like this, and which we deployed last summer, 2024, for about five months with four of our partner organizations. And these collected data such as the temperature of the water at the surface and depth, air humidity, water pressure to be able to gather as a proxy for water level changes, acceleration for things like waves and boat wakes. and then of course you know connected data wirelessly and from the pilot what we found was okay well the devices themselves are pretty well correlated so we had two devices over here that were on the same lake a very big lake different ends and so we would expect to see that okay the temperature sort of changes similarly throughout the day great and the water level changes about the same throughout the season, that's great too. But what we also found was that there were a lot of issues with connectivity where some devices could just never connect to the cell signal at all to transmit data and a lot of the batteries ended up dying either intermittently or just early on in the summer. So for moving to V2 then, of course we had to learn from one, our failures, and two, trying to integrate some more of the local knowledge, especially that our tribal partners have. And so in talking to them, one of the quotes that really stuck out to me was this one where one of our partners said, well, seeds can be dormant for years, so maybe we need to know what the light looks like down there. I think Josiah and I were up at a meeting and someone told a story about a person who bought a lake house in Minnesota and went to build a dock there. And as they were digging up the muck to put in the dock, what happened? A bunch of rice started growing up that hadn't been there for 10 or 15 years and so getting a sense of you know maybe it's that the sun isn't reaching the seeds can we try to adapt for that so our hardware then tried to address this by also looking at light we added in water quality sensors as well and we learned from our failures we put in an sd card for storage in case the devices were offline we had solar charging to be able to actually have the batteries hopefully last all the summer and we added gps location in case devices got swept away by strong winds. Now what this looked like for this summer was interesting. We deployed a lot more sensors with a lot more partners. We had 19 total and the biggest issue was that we had really poor connectivity. So you can see all these red diamonds are ones that didn't connect at least at some point and we had only about two or three that did and of course part of this could be an engineering problem where we say okay okay, well, maybe we need to use LoRaWAN, maybe we need to use satellite, et cetera. But I think it also reflects an important socioeconomic problem, which is that these communities that are facing these issues of land use, mining, all of these things happening on their tribal lands are also facing issues around technical equality. They don't have the connectivity even for us to be able to try and collect these real-time data for them. Now, one of the things I've also been thinking about moving forward is how we might be able to redesign the hardware. And so we've gotten better. This is certainly better than what we first put out. But something that really struck me is that the word mccuc is an Ojibwe word that means container. And so one of our master students in the back corner, Madison, looked last semester at how we might actually make a device that looks like what the mccucke looked like for the Ojibwe culture. So these are beautiful birch bark containers and what we've found through some of our partners is that a lot of them still have the traditional knowledge to make these sorts of things. And because they make canoes out of birch bark, in theory these should be somewhat waterproof and floatable. And so we're really thinking about what that might look like and now working with a 3D modeler at the tribal college up north to see if we can make something that perhaps more closely mimics this sort of shape, idea, culture, etc. So sort of summing up this section with the co-design sensing hardware what we found is that you can integrate all these forms of local knowledge and expand the sphere of whom these experts are and then also get interesting perhaps more than human sensory experiences. So we think about, for example, light on rice beds, especially deep in the water. As one of our partners said, this is more like the experience of a rice bed. How can we think about sensors that enable long? So the second main part of the approach that I look at is deployment strategy. Where do we place sensors? Because that kind of determines what questions we can answer. And so for this I'll talk a bit about another project that's rooted in urban air pollution. And so I think many of us are familiar with the issue of urban air pollution. And one of the most harmful pollutants is called PM 2.5. So these are particles that could be sand, smoke, smog. And the 2.5 comes from them being two and a half micrometers in diameter, which is one thirtieth of the width of a human hair, if you can even mentally process that which, honestly, I can't. This is a huge issue. It causes deaths every year. It causes childhood asthma cases all around the world, particularly in Southeast Asia and Africa. A lot of urban residents are exposed to poor air quality. But even in the U.S., this isn't such an issue. And so, as I mentioned before, regulatory monitoring is not enough. So this project, which took place in Chicago. Here you can see the EPA stations around Chicago. And so what you can see is that most of them are actually not in Chicago where people live. They're more in the suburban areas. And another interesting thing is that they're on rooftops of buildings, some of them which are pretty tall. And so this is myself and a colleague deploying sensors at an EPA station and we had to take an elevator up five or six floors. And so really these aren't measuring what people are experiencing, which is a huge issue. Now for this project, I worked closely with the Urban Innovation Initiative at Microsoft Research to develop a sensor called Eclipse. And so these were lower cost air quality sensors that monitored PM 2.5, ozone, sulfur dioxide, nitrogen dioxide, and carbon monoxide. And we placed them around the city at bus shelters, working with a company called J.C. Deco, which owns them. And so the great thing about placing them at bus shelters is that they were only about eight feet off the ground. And in theory, we could also start to see something about all of the residents who rely on the bus system and what sort of air quality they're exposed to as they're waiting outside. Now, in determining where to actually place these sensors, we thought about a couple different criteria. And so one was that we wanted to be flexible so that we could answer different research questions. But we also wanted to make sure that we prioritize environmental justice. And so I don't have a map here, but if you're familiar with Chicago, just like Atlanta and many other U.S. cities, it is unfortunately quite segregated. And so most of the black and brown communities live on the west and the south of Chicago, whereas most of the rest of the city is quite heavily white. And so what we did was we worked with environmental justice communities from those areas of Chicago and allowed them to select where to place a certain number of sensors. And many of them wanted to place sensors near schools, parks, churches, places where the community often congregated outside. So we had 118 sensors that we placed and we used a mix of the environmental justice groups, some of our partner orgs like the Chicago Department of Public Health, choosing a few, and then stratified random sampling where we split the city up into a grid of 200 meters by 200 meters assigned whether they were high traffic, high population, low traffic, low population, etc. And then split them up so we had 20 in each different type. And then we also had some that we placed at EPA stations for calibration. I don't go into depth here, but I'm happy to talk about that as well. So what does this actually look like when you deploy it this way? Well, we compared our network to the EPA and the Purple Air, which, reminder, are those consumer ones that people can buy for $200. And we looked to see, okay, how far are these sensors from the average Chicago resident, the average black Chicago resident, and the average Hispanic Chicago resident? And we found that our sensors were always closer. And you may just say, well, of course, you have more sensors, duh, which is true. But we also then looked at, okay, well, what about the sensors that are running for more than 75% of the days of the year. And we use that criteria because that's what the EPA requires for calibration. They say your sensors have to run for at least 75% of the year. And what we found were that our sensors were still closer and shockingly even some of the EPA monitors don't run for 75% or more of the year. Now furthermore we looked to see well what is the difference if we look at the ones that we chose somewhat randomly versus that our community partners chose. And what we found was that our community partner locations were much more likely to be near manufacturing zones, about three times more likely. They also had about three times more rail miles close to them. And so they were identifying these manufacturing sites, these toxic release inventory sites, and these rail stations, which our random sampling didn't capture. And so this is super interesting because one, in seeing the communities, we can see what's important to them. And then two, it also really touches on the fact that environmental justice could be very different based on the local communities that we're talking to. Now another interesting thing about our deployment strategy is that we were able to capture hyper-local events and widespread events. So for example, if you look at the top, you can see we first deployed just before 4th of July. And we saw a huge stake, as did the EPA, which is shown in red, around the fireworks. So TLDR fireworks are terrible, just don't do it, like keep your hands, also your lungs, but also because of things like barbecuing, grilling, etc. We also saw a spike around wildfires that were coming in from Canada at the time. But then we also saw that there were lots of events that one of our centers was picking up that the EPA wasn't. And so when we went to investigate some of these sensors, we found very hyperlocal sources of pollution. For example, one being this fried chicken spot. And so this sensor, we knew something was going on because we placed it across the street from another sensor as our sanity check. And this one was going up every day around 10 30 until about 9 p.m. And we found that it's because the vent, the exhaust for the fried chicken was right there. And so this is crazy because if you imagine waiting for the bus, maybe you're like oh that fried chicken smells so good but actually it's not really that great for you. A similar issue is that we found one near a bar that was like tiki churches outside at night and so similarly right you think oh great ambiance but in fact releasing all of these particles which are harmful for human health. Now some of the challenges we had several but I wanted to touch on some. So one of the big ones was again around connectivity and charging and this may be surprising. It was for me because we're in a city. We're in a massive city. Why wouldn't we be able to connect to cellular? But we found that 11 of the sites that we had originally planned for couldn't connect to cellular at all. And three of them were ones that community selected. And this is crazy because it shows again that sort of socioeconomic intermingling of all the different issues that people have, right? Air pollution, connectivity. Even some of the sites didn't have bus shelter, so a lot of the south were more sparse and places where people wanted to place sites that weren't enough bus shelters for us to be able to actually place a center. And we also had issues with solar charging, so we lost a lot of data, and some of this may seem obvious. If you're familiar with Chicago, that's downtown there, that yellow dot is real. Sorry, technical difficulties. That yellow dot is right next to a giant building, and so it makes sense that that one wouldn't be able to connect. But this one out here was actually like much more suburban and again in a low-income black and brown community and what we found was that it may have just been trees but it's really difficult to try and predict these issues because the open data for buildings and trees is not always there not always accurate and so yeah trying to figure it out is very difficult and this is the same for cellular connectivity. I'm okay. I'll clip it on again in a sec. Thanks. Now, another challenge that we had is that not everyone was happy, which I mean is life, but still it's important to consider when you're working with people. So for example, this is actually from one of our partners who worked for the Chicago Department of Public Health. And he was upset because we had totally ignored the area that he lived in. And he said, you know, I think this is a misconception that everyone thinks the north side is rich and white, but I work for the city and my wife is a teacher. We're not rich. So of course this brings about the question, well, how can we then make a fair sensor network? And so that question is actually one that I tackled for part of my dissertation. I won't go into it now, but what I can tell you very briefly is that fair itself is a very loaded word and everyone has a completely different definition of what's fair. And so really, again, this requires local knowledge of knowing, well, what does fair mean for this specific area? So in sum for this section, co-designing the deployment strategy, we saw that we could get coverage for underserved groups, particularly in letting them choose where the devices go. And they could sort of self-identify what the important environmental justice areas were for them rather than us trying to decide as outsiders or letting an algorithm choose for them but then we also found that we might need this perhaps even more so than what we did for this perception of a fair design I'll clip on in a second any quick questions while I reclip yeah Michelle Yeah, that's a great question. So we, we didn't record any of those meetings. The question in case you didn't hear was if we collected any commentary on why community sites chose the places that they did. And so we did not record those meetings. And again, that sort of goes into the whole, you know, recording makes things feel weird. But what I can tell you is that a lot of it was really tied to like where they know people are or where they know or think specific sources of pollution are. So, for example, by the railway station, a really good example is there is an asphalt plant that one of the community groups really wanted to get sensors near because it was right across from a park where people would spend time outside. The interesting note of that is that that sensor that we placed outside the asphalt plant mysteriously got ripped off several times. No one knows why. Yes, but yeah, that's a great question. And I think, yeah, getting a sense. And I will say, too, though, that, like, of course, community groups are not a monolith. And so we had some groups who really cared about some issues and some groups who really cared about others. Like one group where their whole thing was they wanted more trees and another where they didn't want construction. And some of these groups also hated each other, which made it quite difficult. But is the reality, again, of all these people? No. Okay, cool. So I'll get into this third main section then, which is around data interfaces, right? So we collect all this data. We actually have to do something with it and show it to people and determine who can answer what sort of questions. So for the Eclipse Project, we worked very closely with city, community, and research stakeholders. So we did some interviews to determine what sort of things people were interested in with particulate matter data specifically. And what we found is that a lot of people wanted to look at things like policy, education, and empowerment. And so we tried to figure out how we could design interfaces to support those. Now, the first interface we made was kind of what you would expect. It was a map with, you know, showing where all the sensors were, colors, values, etc. and we thought we were being like super cute by adding QR codes at the bus shelters so that people would you know see the QR code and scan it and then be able to see the map and of course it turned out that no one ever scanned the QR codes and we also found that people were like I don't know this map doesn't really show me enough like I want to know how different my neighborhood is from the one that I think is much better for example where people said you know With scanning a QR code, not everyone has a smartphone. As we saw, there's not cell signal everywhere. A lot of people want to look things up in the library, and maybe they don't have internet at home. Now, as we were making the website, we also worked on a public API. And so this was a way for primarily researchers and programmers to be able to plug directly into our data and make their own visualizations. And what we found, though, is that, perhaps unsurprisingly, community groups and city employees, they don't really have the time and maybe the expertise to work directly with raw data. And so, you know, we were trying to figure out how we could create some sort of low and low cloud environment. And now with the advances we have in chat GPT, that's perhaps an option for people to be able to say this is the sort of graph that I want and work with the data. But at the time, we really couldn't figure out how we could connect this with the partners we were trying to work with. And another challenge that arose is how we actually get the data to the communities. And so this is something that our partners often ask about. How do we show the community members that we're even doing this and what data we're collecting? And so to this end, we made a sort of crazy, wacky interface called the EcoPod. pod. And so this was a horse trailer that we purchased and we retrofitted it with solar panels with, you know, little display areas. We drove it to Chicago and we brought it to events such as Earth Day at the Field Museum, which is a really big natural history museum there, a big recycling event in a park, et cetera. And so we went there, manned this station, talked to people about the sensors. We had a screen that showed all of the data in real time for people to be able to interact with and we brought all the different versions of the hardware that we had for kids to play around with and they were super into that and we also had tables for all of our community partners to be able to also set up and interact with people and what we found was that people were really engaged with this and we would see a huge spike in website traffic during the event and then maybe a few days after but then it would trickle down again and so this points to a lot of questions of how we can keep that sort of engagement, and especially, I think, how we can have residents feel like they're inputting something. Because what we learned at these experiences was that a lot of people wanted to tell us their stories. They wanted to tell us what they knew, what they experienced, and we didn't really have that sort of input in the interfaces we made. Now, another challenge that arises in these sorts of interfaces is how we can integrate different sorts of cultures. And so last year and chatting with some of our partners up north for the Manuman project, one of them told me, well, people go out and harvest the rice in Labor Day weekend because that's when they think they're supposed to do it. But really, it's not about Labor Day. It's about the ricing moon. And so it would be really cool if instead of showing a traditional calendar, we could show some sort of moon calendar. And so for the interface we made for these devices, which we've called Noondawin, which is an Ojibwe word for as one is heard. We started exploring some culturally situated interfaces. And so we had a great master's student named Jonavi who was working on this last year. And what she did was she looked at all of these different Ojibwe art pieces. So beading, for example, paintings, some of the art that our partners publish in their quarterly magazines and on their websites, the moon, which is often shown as a turtle with all the different moon cycles and started thinking about, well, what would this even look like if we had a timeline that showed something like moon cycles rather than the Julian calendar? We also looked at different iconography and integrating the Ojibwe language as well. Another challenge we have is protecting Native Nations data. So tribes need to protect their data. It's often been exploited by governments. And even non-tribes, even all of these marginalized communities have good reason to want to protect their own data. And so this of course requires respecting the knowledge, the sovereignty. A big one in tribal communities is elders and tradition and then also our non-human relatives. So in a lot of the communities we work with up north, they really look at rice, animals, plants, etc. as relatives rather than as resources we would exploit. And then finally we have this challenge of turning data to action. So like how do we take data and make it into policies? What do we have control over? How do we know where to actually go and do something? So with Eclipse, we had some pretty good success through our API where local news would publish stories about it. And again, we would see traffic on the site. We would see people locally try and do things. And even the Chicago Department of Public Health went to investigate one site that we found had high pollution and change some of the traffic patterns around there. For the Wild Rice Project, what we've been looking into is... we can do something called agenda alerts. And so this is based on the Public Meeting Act, which is a U.S. thing, United States thing, that says that local governments who have public meetings, so city council, town council, et cetera, they have to tell you beforehand that the meeting is happening and what's going to be discussed. And so based on that, could we then alert people to say, hey, they're going to have a meeting talking about this lake you care about or this project you care about or your town for example so i've been working on that with one of my collaborators and there's a lot of challenges right it sounds like everything it sounds easy but it turns out that these agendas a lot of them are kind of just like useless text um and so this is an example of one and you can see that it's like oh you know comments whatever and then there's just a lot of numbers and addresses and you know It provides challenges, and a lot of people now would say, well, throw an LLM at it. Well, there's not really enough text to do anything with it, maybe if you read every agenda ever. And so we're working on this now, trying to look at named entities, which is towns, people, et cetera, and then also looking to see if maybe the meeting minutes that get published after the agendas can be used to learn more about it. But again, open challenges around action. One of the biggest ones is the livelihood. A lot of the people in these communities are working for the industries that are polluting. And so their livelihood depends on it, but their actual lives are being harmed by it. There's also politics, of course. Once you finally get inroads with one mayor to put up censors, well, then they get voted out and now you have the next mayor to worry about. And a big one, of course, is accuracy. Low-cost censor data does not equal proof. So I'm going to skip ahead to make sure we have enough time for questions. But I really want to talk about this last idea of what I'm calling redefining local knowledge. And so we throw this term around a lot in HCI. And one of the things I've really been thinking about is like how we can incorporate different kinds of knowledge. So one of them is indigenous knowledge, which is basically like ancestral knowledge that's been passed down through generations about land. And usually it's oral tradition, songs, stories, et cetera. And so the question is, well, how might we use that? So I've been working on collecting indigenous knowledge. A lot of it is through books, for example, online through songs. I had a student help do some interviews about wild rice. And for example, this might look like a traditional wild rice story where Nanabuju, who's often the character in this story, can't find food. He's getting annoyed, follows the ducks, and they take him to a lake full of rice. And he sees all the ducks and geese eating it. and now he knows, oh great, there's rice here and the ducks and geese eat it. So could we use that then to incorporate something like a microphone into our sensing systems to say, hey, we hear a lot of ducks and geese munching on something. Maybe it's time to go harvest the rice, for example, when we combine that with all of the other information we have. Now the last quick thing I want to talk about for local knowledge is this idea of experience. So Cleo and I have been working on a very strong, depressing project around carceral facilities. And these are another huge environmental justice issue where people are dying because of heat, because of pests, because of illness, because they're not being treated in a humane way. And this is a particularly huge issue in Georgia where we have a large carceral population and where we have so many facilities that get super hot in the summer and do not have any kind of air conditioning so we interviewed 19 formerly incarcerated people as you can see the majority are black um which is probably not surprising black low income uh etc and we asked them questions around what their experiences had been and then if we were to have things like environmental sensors or health sensors what they would be comfortable with us with the data that's collected. And an interesting finding about the health wearables is that a lot of them actually wanted their location tracked, which was surprising for us because most of us would say, I don't know, location feels very personal. And they cared about that much more than something like sleep tracking, which many of us like to have as a monitor for our health. And what was interesting was that for location, for example, someone said, well, this is so important. It's like the key to safety because you can be forced somewhere you can be drugged somewhere you could be you know beaten up somewhere somewhere you're not supposed to be and it's essential for your life and well-being for officers and administration to know where you are and then sleep which seems innocuous to us well actually maybe you get in trouble because you didn't go to sleep the right time or you didn't wake up at the right time and so it really just highlights the importance of learning from this lived experience and what that form of local knowledge is. I have a lot of collaborators. So many of them are in here, which is amazing, so thank you. And yeah, so please, you know, ask questions, comments. I'll leave up the contributions. I already know what my lab will be called at MIT. If you hate your advisor here, let me know. If you have any promising students and we are thinking of a postdoc and any of this speaks to you yeah please reach out thank you all so much a question now that I ran through the end yeah Brittany thank you for the talk um this is just a simple question but for I think it was the project was there a reason why specifically you all chose to create a website like was there any data that kind of led to that to say this would be a useful technology for people. I know you said that like people in a scanning circle. Yeah, yeah, that's a great question. So the question was like, for the air quality project, why a website? So there's a lot of ways to answer that. I guess the most simple is like, that's just what people do, which I know is a bad answer for why to do something. But spatio-temporal data is difficult to work with. And how you visualize both space and time together is something I think about a lot, and I don't think anyone's come up with a great solution for it. It was all, you know, part of it too is back to that issue of pacing. Like we had a date that we agreed on deploying these, and so part of it was like, oh my God, what's the fastest thing that we can get out of this? But it's interesting because like even for our wild rice project, for example, what we have come up with is a dashboard. The dashboard is just the common view, and it's it's one of those like unfortunate interface things where it kind of ***** but no one has figured out a great alternative for it yet and yeah I know that's probably not a satisfying answer. I was just curious if like because I know you talked a lot about having the community like yeah I didn't know if it was something that came up. Yeah so one thing we had thought about which especially working with the bus shelters um they have like you know sort of spaces for advertisements and stuff and so we had wondered you know what if we could put like some sort of screen that's showing you or even a small led that's showing you things in real time that way you don't have to connect to a website and you can see it while you're there um i mean that's an area of research i would love to explore and yeah i guess i'm allowed to say what i want now like full disclosure msr shut that project down abruptly and it was very upsetting and we were in the midst of working with new york city like you know planning oh what amazing times square installation what we do and oh my god we're gonna be famous um but like obviously that didn't happen um but i think you know something like that could be really powerful where it's like maybe you don't why even have to pull out your phone we don't need to do that right why not just be able to show people so that everyone can see it and maybe you could even spark a conversation at the bus shelter rather than have people just staring at their tiktok or whatever yes you mentioned putting up QR codes at the bus stops. Can you describe what those were? Like, was it just a QR code on a sheet of paper? Yeah, yeah. What did that artifact look like? Yeah, that's a very great question. So we had these little stickers, because you have to, like, put it on the pole, which is rounded. So we made these stickers, and they said, they had, like, a little picture of the sensor. Like, hey, this sensor is measuring air quality. Scan here to view it. Did you print out maps at all? Like, we're just going to take, you know, every week we're going to go out and stick up a new map of you know that's a great idea yeah especially in that that big space that would have been a great idea yeah where maybe it's some sort of right it's like somewhat real time but um yeah not having to be a screen that could get damaged or something i love that no we didn't think about it but i might steal it for the future okay amy in terms of local knowledge i think you spoke a lot to like the value of taking like local knowledges and bring that into research into policy like giving them that power of precision in a way to be more politically powerful i'm also curious like what like local community members learned themselves from doing these sort of about more than i mean yes obviously the precision to have that political power in a way but like what yeah what they learn about these concepts and the way in which integrating more quote-unquote local traditions of knowledge helps that process yeah that's a really great question um i guess so far that so that particular part that i think you're talking about around like integrating the indigenous knowledge for example like that's still pretty early on i think in terms of more broadly though like what communities might be learning one of my hopes having formerly been a teacher is that there can be lots of educational opportunities and in an ideal world like it's not sustainable for researchers to develop these and keep them running we need to make them then community run and owned and so i think in an ideal world perhaps community members who are interested are learning the skills to do the data analysis even to make the sensors for example right so that way it could be like hey let's help you prototype and get through the initial process but now this can be a community run project and a community owned project in addition to the data part of it Yeah. Hi. Actually, I want to continue asking a question on the QR code. Was there also a concern of security and privacy? That's why people did not scan? I think it was during COVID, so I think there was QR code fatigue. And I think most people, I wouldn't walk up and just scan a random QR code. Right? Like, everyone remember that Super Bowl ad where they just had the QR code back? I wasn't going to scan that, you know? So I think that's part. And also, probably people just didn't even see it. Like, you know, if you actually sat and just watch people walk around a city, what you'll see is that everyone's in their phone anyway. So yeah. Yes. Yeah, and then I'll go with that side. Sorry, yeah. About co-designing with the communities, I was wondering how much the visual, taking their cultural traditions and incorporating that into the visual aspects of things, of the tools, how much that trusted trust with those communities and also if you think increased trust for those communities and better results? Yeah, yeah, great question. So the question was if the co-designed with especially community visuals knowledge might increase trust. So we did do a user study with this dashboard where we had some of the iconography and color schemes, et cetera. We've also done some small user studies about language. So the language part is a completely different beast because so many people, unfortunately, no longer know the language because of colonization, assimilation, etc. With the cultural elements, we didn't specifically ask about trust, but people did remark that they liked the things that we were doing. And I think it spoke to some of the colors or some of the familiarity that they have with things. One thing that we're working on now is hiring an Ojibwe designer to help put even more of this and have it fit in more with some of the things they have. So, yeah, that's a great question, and I think that's something we could explore in the future. I'm going to go this side, and then I'll come back to you. Yes. yeah yeah that is a great question um that so yeah i mean it's funny if you talk to like um i don't know how to if you talk to the more senior researchers and sensors if you ever go to a conference they love to talk about this thing called sensor dust which is like you know the vision that they had 20 30 years ago where sensors would be so tiny they would just be floating around in the air all around us which is absurd who wants to be breathing that in but whatever and so if you ask a censor person they would say yes but if you ask me like of course not right like what city is going to have budget to have a censor on every block even and so this is where we the question of like how many do we need and where do we need them is something I looked at in my dissertation and the answer is who knows it's almost certainly highly specific on the city itself or the place you're looking in and so like what I looked at in my dissertation was okay well one we want some idea of like representative data so if I'm measuring over here is that good enough for me to know what's happening five blocks away for example and the answer is it probably depends on the layout of a city and so there's different types of street layouts like if you think about you know there's some streets that are really narrow that have tall buildings versus streets that are really wide and have parks and so the way that like pollution moves in those can be very different for example um there's also like the very practical question of well where are people right if we're measuring for people does it really matter to measure something if people are never over there for example um so usage is a big one and then of course this whole idea of like you know we talked about connectivity and charging issues like if we actually knew where we would have these resources, that could also play into it. And so, you know, I think there's, I know that's not a solid answer for you, but it really is a nuanced thing and it probably does matter on budget. And I think there is such thing as overpay, right? Like we also have to think about the, like collecting even more data, adding to more data centers, adding to more electronic waste. Like that's not the solution necessarily, but having some sort of computational model or at least like confidence that, hey, something that's happening over there, I feel pretty be confident that it's, I know what's happening to me. I think that would be the goal. Yes? First of all, you're a great speaker. Oh, thank you. And I like the fact that you need the term environmental justice people are because of political pressure. Yeah, I'll end up on a blacklist, whatever. You touched upon it towards the end of your talk was, have you done more studying about translating? You can have all the great data in the world, blah, blah, blah. lot, but translating that into policy, so what? Yes. Yeah, so with the Wild Rice Project, we're working with, we have a governance team, and so we're working with them to try and determine that so what, and what it involves, what their work is, is a ton of reading, like, legal policies, tribal agreements, you know, as far back as the 1800s, for example, and then sort of determining, like, okay, these are all the different intersecting things and where we might be able to have a play. I had a great meeting this morning with Jackie from the South River Water Alliance. Nope, something like that. I had a great meeting this morning where we were sort of talking about this and it sounds, basically it takes a lot of work, it takes a lot of reading and it takes, as far as I learned from that conversation, the wings to sue people or at least align with people who are going to sue people. And so it seems that, you know, and at least in working with tribes, that certainly is the direction a lot of them are planning to go. Like one of the tribes we work with, they're somewhat famous now. It's called Bad River. They have a documentary about them fighting against the building of this pipeline. And so, like, they are, that's their goal is to sue and say, hey, what you guys are doing is violating our sovereignty and the agreements that we have. But, yeah, it's so interesting because, like, even in the meeting this morning and even in working with the governance folks, like, and trying to think, well, how can we help, right? Like, could we make tools that read through these documents? Sure, but could we trust them and know that we're not missing something important or how can we help make these inferences? So, yeah, it's something I still wonder about. And I know, like, Haisley, for example, is working on a project around data centers and we're thinking, well, maybe there's something about making tools for policy makers to see and be like wait this is what's happening like that's crazy because we know they don't read all the bills and stuff anyway too so maybe part of it is figuring out how we surface things better to them um so i guess the answer is yes but not enough uh that's that's the next big step yeah in the back i have a first of all thank you for your talk um i have a more practice oriented question uh i i'm currently trying to do like more co-design in my research and uh i wanted to ask like when When you do co-design with community members, what do those engagements actually look like in practice? Like, are you doing workshops, are you doing interviews, et cetera? And after or during those engagements, how do you decide when to defer to community members for design or technical decisions? Yeah, that's a great question. So what the co-design actually looks like, it's really tough. So I've done very few of the, like, typical, I don't know if it's typical, but, like, the often used HCI, like, co-design things that people would say, like the card sorting or the, you know, workshop. I'm familiar with them. I encourage some of my students to look at them. But for a lot of the communities we've worked with, part of it is that, one, they're more our collaborators than communities, if that makes sense. And so, like, a lot of them, for example, our tribal partners, we include them as co-authors on our paper. We're not running sessions where we are compensating them for an hour of their time. we're meeting with them every week, calling them on a lot of questions. They're calling us when they have a question there in our Slack. And so, I think part of, a lot of it has to do with what sort of communities you're working with and to what extent you want to do a co-design. For like, starting out, you know, a small project, like yes, I think like a workshop, a part of some sort of validated activity makes a lot of sense. for the incarceration project, like Theo and I did semi-structured interviews that we put together a real guide for it. And we compensated people because that was a, hey, we're meeting with you long time, and we're going to compensate you for your pain. And for the more long-term partnerships, what has worked for me is more of like, hey, we are collaborators. And so, again, you end up with weird things around IRB and stuff. So, yeah, another thing that I've learned, though, too, is, and there's a great paper on this, Justice Design Oriented Learning. And so, that paper highlights something that we found in our work with tribes, which is that, like, a lot of the ways that we in HCI do these sorts of design activities don't necessarily equate to how communities work. And so a good example of tribes is that a lot of them prefer to do communal feedback. And so, you know, we've talked about, like, hey, let's bring the dashboard and, like, do a little interview. They're like, actually, no, why don't we have the, you know, whole town, whoever wants to come, come sit, and you present in front of everyone and get their feedback. They're like, it's the talking circles. They're like, this would be me talking at you. Like, no, let's sit in a circle and all talk together. And so I think that's another good thing, too, is, like, once you identify the community, trying to determine how it is that they prefer to participate. And then in terms of, like, defined expertise, I mean, I think we always have to defer their expertise. Carl has great work on this, too, Carl DeSalvo, where it's like, you know, maybe the output isn't, quote-unquote, novel research, but what drives you, right? What motivates you? I think that's the big question. Of course, technical expertise, you know, if they're like, hey, build a spaceship, you know, I don't think I can do that. But most of the time, requests are, you know, reasonable. And I think then it can be a fun challenge to sort of determine, you know, oh, well, based on what you said, here's what I think would work, and, yeah, find some happy ****. I've done homework on this, but I did notice that you were going to be a middle school teacher at one point. Yeah. That is the Lord's work, and I want you to believe in God that he did not do. But I saw this, a lot of this stuff would be very applicable to a middle school-type curriculum, and having, especially with the, I was thinking about the Rice Project, where you could have the kids go out in canoes and play stuff, and it would be a project that would build community, and there's a, there, you did, you know what it's like, those kids love that kind of ****, that age just, it's very exciting about life, and it would be a great way, It's also a great way for the elders in the community, I'm not talking about elders in particular, to be, to integrate with their young people. Yeah, absolutely. There's a lot of that. A lot of this could be, I'm looking at it, I have a background in wireless communication. A lot of this could be dumbed down a lot. You know, you don't need as much worry about this and that and the other. but there's a lot of it that can be very basic in terms of sensors. Sensors can be a lot cheaper than what people can do. But I'm really intrigued by this. I'll talk to you all. Yeah, thank you. Yeah, no, and that's, you know, there's never enough time to say everything you want to do, but that is why, you know, the education is one of the E's in the future of that name because that's so important to me. Like, I love, I don't love kids. My God, they're terrible. But I loved being a teacher and figuring out how to make things fun for them. And, yeah, like, even, you know, what Josiah and I have a couple times now run a workshop up north where we use micro bits, which are these very low-cost sensors made by Microsoft, and you can use block coding to program them. And I've figured out how to make those work with some water-quality sensors to get them out. And, yeah, that's something I'm super interested in, like, even how could we make, like, some sort of low-cost toolkit? Could we get kids to be really thinking of the enclosures? And then also, too, you know, I think in terms of touching on the expertise, like, these are also people who live in the communities, who know what to look for, who know what they want to measure, where they want to measure things. And so tapping into that form of expertise as well, I think, is super useful. Yeah, thank you. Yeah, thanks.