The deck and Ling asked me to speak and the colloquium actually, I think for the very first colloquium talk of the fall. And in a lesson that I'll come back to at the end, you know, I, I like we often do thought, well, I don't want to do it now. So I'll just say could I do it later? And then of course then later shows up and later in this cases the week of finals. And it also happens to be the NSD IPC meeting. And I'm really happy that I had a good excuse for stepping out of that two-day meeting. But anyway, it's a it's a busy time of year. So and I think that maybe there were multiple visions for what this talk might be about. And so I chose a title that I thought was as generic as possible so that I could decide a little bit later what I want to end the talk to actually be about so old and new problems and networking that the ending tagline, including the computing kind, was it later edition? When I decided that I would talk about some things that are networking, that are not the computer kind. So, so the talk is basically divided up into some number of parts that hopefully we can. There we go. Yeah. So the talk is divided up into three parts. The first part, I'm actually going to talk about computer networking. I'm actually going to talk about kind of one problem, umbrella. That is both an old problem and a current problem and unfortunately I think probably also a future problem. Then I'll talk about my time, networking and the people sort of way. And maybe a little bit in the intellectual sort of way and the Computing Research Association. And then I'm going to end with a little bit of talking about how about choosing work for the future. So, you know, these talks, we usually like the speaker to say something that seems sort of, I don't know, more profound than all the other things they've said. So that will be the attempt, the attempt at that, at the end. And also I'm going to attempt to tie these things together under the rubric of networking. So the, the problem that's both old and current and unfortunately I think probably future that I want to talk about is the digital divide. And welcome to Merck has been here before Ahmed than that. I like say maybe haven't. So you haven't missed anything. So I'm going to talk about the digital divide as an old problem, a current problem, and, and a future problem. And in particular, I'm going to talk about mobile Internet access. So inequities and variations in, in mobile Internet access. And I wanted to show you this, this picture because I think that it really helps illustrate the point. So this is the state of New Mexico. New Mexico was really a lot like a square, a rectangle, except a little bit at the lower south, southwest. This is a picture showing a particular cellular ride. Or as soon as you want to talk about mobile access, you need to talk about what provider you're, you're, you're focusing on this. This is Verizon I next slide, I have a couple of charts for two other providers. And there happened to be two datasets illustrated here. One of them is provide one of them is the FCC's dataset. And so the FCC requires providers to say on an annual basis what geographic areas they provide cellular coverage for. So this is a requirement the FCC puts on cellular providers. The providers fill out this form. Fcc then gathers that data together and makes it available publicly as shapefiles. So geographic regions, there's a second data source here. And if you want to jump ahead, you can look here at what the meaning is of these different colors and how these datasets come together. But also say it in a second. To the second dataset comes from a location services company to which Georgia Tech has a very special connection. Kick Jones, who I believe worked with laying when he was a graduate student, is the chief advantage, a Chief Evangelist, technical evangelists or something at Skyhook. And he happened to visit, yeah. So I think lives in the area he happened to be visiting a couple of years ago and I met with him and I was talking about this work we were doing, looking at documenting and understanding mobile access, especially in rural areas, remote areas. And he said, Well, we have some data that might be relevant for that, is relevant for that. So it's been a very nice relationship. We have this data from sky hook on their view of where Verizon, in this case has coverage in New Mexico. And that, I guess the important thing to know about the Skyhook data. That it's, it's crowd-sourced. It comes from devices that people are using where they have signed on to use location services. And in exchange, their location is being shared and that's being aggregated into these Skyhook, the Skyhook data. So, so, so know that these are, these are actually fairly different in terms of where the data's coming from, what the provider says And and then Scott hex location and that location services. So then of course, since there are two data sources, there's four possibilities for any given geographic location. The two sources could say agree that the region has coverage, that Verizon provides coverage in the region. And so the blue here is everywhere where both the FCC and Skyhook agree that there's covered. They can also agree that there isn't coverage and that's the orange in the picture. And so if we just focused on the blue and the orange, we could sort of start to think about things like, well, how does this look right in terms of of 2019? And whether you can get coverage if you have Verizon as your provider across New Mexico? And the answer is, you know, you can get it in a fair amount of geographic blocks. But certainly, again, I'm nowhere near all this. There's quite a bit of orange here. And then also interestingly, if you want to ask like, okay, well, how well do we really even understand? Know what's happening? Then it's interesting to look at the disagreement colors, right? So the purple is where lighter, purple is where the FCC says, meaning the provider say we have coverage, guy who doesn't see that. And then the darker color is where the FCC says no coverage or pride or say no coverage. And that's guy excuse coverage. So and not surprisingly, these kinds of regions of disagreement, especially in this lighter purple one, they tend to surround. They tend to be on the boundaries between agreed, uncovered, integrate, covered, right. So so you know, every agreement, this is covered, agreement, this is not covered and it's you often will see this sort of surrounding it. Yeah. Yes. Global actually, they're they're, they're they're global. Yeah. So, so right. So these disagreements tend to come in, in it, not unexpectedly in places where they're seeing, they're sort of these transitions. It's also the case that there's a lot more space where the FCC or providers are saying yes, we have coverage, but Skyhook or crowdsource data does not indicate that. So so, so multiple terms of takeaways here about a real difference in whether there's access or knots right around the, around the state. I haven't laid over overlaid onto the cities and probably most of you don't know New Mexico too well. But but, but our interest in New Mexico came from the fact that there are a number of tribal reservation lands in New Mexico. The Navajo Nation actually is right up in here and spreads into Arizona. And we were interested in knowing how access varied, not only on herbal, urban, rural axes, but also on tribal, non tribal axes and the sort of takeaway which you can't tell from here, but I'll just tell you is that rural is the most, rural and tribal are the most challenging. And tribal rural is, is the most challenging because it has all the issues of rural challenges of rural access and affordability and whether it's cost effective for providers to go there, combined with a number of complicated governance issues that arise from tribal governance interacting with FCC and, and, and US government governance. So just in case you thought maybe I had picked Verizon because it was I don't know, something. I picked it actually because it's sort of falls in between the middle of these three with respect to the, the kind of coverage picture. So the takeaway here is don't use AT and T. And T-Mobile maybe is a little bit better than, than Verizon. So in this research, so this is, these plots are coming from a research project that's a collaboration with Elizabeth building at UC Santa Barbara and a couple other folks who are in Arizona. And these are the sorts of maps we've been building to understand. First of all, what we thought we would try to understand where coverage exists. And then we realize that these disagreements are also a really interesting source of, I mean, they really reveal another research problem which is given in multiple datasets. When they sometimes disagree, like how, what, what can you do to try to leverage them and still end up producing better maps when you have this data disagreements? I wanted to to to connect. Yeah. Yeah. Yeah. Yes, there is a there is a good a good correlation with population density. Absolutely. Reasonable is yeah, It becomes a fairly complicated question and I mean, reasonable as us or a policy question, right? Yeah. So so it's clear what they are. I think they are optimizing profit almost certainly as a first-order consideration. So let me, let me show, let's keep going a little bit more. So I wanted to show something closer to home. So so we don't just think that well, errors, New Mexico's far away. I don't know what's in the Southwest. It's not not something these issues don't exist maybe on the populated, maybe it's population densities are low. You know, this may not exist on the East Coast. Well, so Pike County, Georgia. And so you know where it is? This is about an hour and a half drive. This is the server. So now this is a switch to fixed broadband. So that was mobile. Actually, I'm more interested in mobile, but let's talk about fixed for a minute. This is fixed broadband access. And this is the 2021 Georgia broadband deployment map so that the Department of Community Affairs in the state of Georgia undertook a pretty significant effort to produce maps of the fixed broadband across the, across the state. And this is, this is the results is quite recent data. And as you can see there in terms of population, you can't, you can't tell that because population isn't overlayed there, but it's 30 percent served, 70 percent unserved in Pike County. And Pike helped me is really close by, right? I mean, we have faculty who live in Peachtree City. If I blew this out even more, what you'd see is that there's this immediate drop off. There's this, you know, the section around Atlanta. And then, and then as you as you leave that, you see this very rapid drop-off in coverage. And the reason why I pick Pike County is because ADA and I and some of our students, one of whom is here. I have gotten involved in a project with your just smart georgia smart communities project to to work with Pike County and and Concord, which you can see in the picture there. And then another county and another city also down in the same region of Georgia. So that's why I wanted to show, to show these two. So, and even in the fixed broadband case, we also see data discrepancies. So the Georgia broadband deployment data map, they have a they have an interactive slider where you move the slider across the map and you go from seeing the the the data that the Georgia program pod project produced and the FCC provider data here on the fixed category. I suspect I haven't looked at fixed the discrepancies for fixed data very much. I suspect that this may actually illustrate affordability, that the difference here may, because the differences are so stark. And in contrast to mobile, it's frankly much easier to know whether you are offering whether you have a capable of providing service to a fixed location. Mobile location has a number of challenges that have to do with radio propagation that, that are very hard to, to, to, to be precise about. I think this is probably the difference between it being available and it being affordable, meaning people are actually taking it up. And so when it says the left-hand side being the affordable uptake and the right hand side being available. On the other hand, providers may also, there's probably also some issues here of provider, potentially provider exaggeration about where they were, where they offer service. So to data discrepancies, even in the in the fixed location, fixed providing coverage situation. So using these kind of two, so I uses these figures to illustrate a kind of background for a research agenda around the digital divide. And this is a picture taken from a hiking trip at codling canyon and it's impossible to see here, but the trail marker say, it's very easy to get to here. The trail marker for the trail, the next spot on the trail is across the that, that river. So this is my illustration of the digital divide. It, it's easy to get to here. It's maybe hard to get to there. It's actually not that hard to get to there. Pretty decent ways you can step across the rocks. But so, so a research agenda that has to do with a digital divide, and they're actually just a number of problems that arise when you, if you, if you start to think about or focus on the digital divide in from as, as basic as mapping it. What is the situation? How do we, how do we document and, and, and document over time since these things change the digital divide, that raises issues of how to measure mobile conductivity efficiently, effectively. How do you go beyond conductivity? So, so right now, I've shown you only like do you have coverage or not, right? This very binary. I'm not not to say. Did you have enough coverage that you could be on a Zoom call or watch a video. So it's very binary and doesn't tell you much about performance. And then of course the, the, in some sense, the ultimate outcome methods to bridge the digital divide. So I'm gonna say a little bit about, about each of these. So first, let me talk about mapping and measuring and what makes it hard. So these are screenshots taken from an application that I'll talk about next. But just for now to know that the blue dots mean that there was conductivity. These are crowdsourced measurements we took ourselves. Blue dots are, are there is coverage, red dots are there's not coverage. The that picture I showed you with the crossing the creek to get to the other side of the data collection is is right around in here. And this on its own, illustrates spatial variability, which we will know, all we know will happen. This is perhaps a little bit surprising. The fine grain this, of the spatial variability it happening. So, you know, there's no scale here, but these are tens of feet apart if that probably mostly less than 10, 10 feet apart. But it happens to be in the canyon. So there's a lot of rock around. So this illustrates this spatial variability. It doesn't illustrate because I would need to show you maps over time, a multiscale temporal variability. But I can tell you that multi-scale temporal variability definitely exist. So whether depending on the technology, weather can be a factor in, in signal quality, seasons change potentially, for instance, when trees are what lose their leaves, cake has tree cover is a signif, can be a significant factor in the propagation of some frequencies. Though trees lose their leaves, hover, it gets better. Of course, on the temporal scale of the deployment of new infrastructure will definitely see changes. So this longitudinal, there's a significant longitudinal element to it. And most of you won't know what be able to guess what this is. But it also turns out that The Walking Dead have an effect on conductivity. We, when we were in Concord, Georgia, they were describing to us though, the municipal ISP that they have set up and said, Well, we set these links and working great. And then the Mayor, This was actually, I think to the mayor's house. The mayor call us up and said, You know, that was it was great. What happened? It's terrible. Turns out they were filming the walking dead in Concord and for TFL. And they brought these gigantic trucks, right? And they park them on the street. The trucks block the signal. So lots of forms of of temporal variability. Though it sounds funny, It's actually quite a significant challenge for that building a wireless ISP. These issues of line of sight and non-line of sight. They had they had high towers. But because of tree cover, they had to put in point some point to point links at ground level. Well, that's not on the ground, but anyway, ground level and yeah, that's where vehicles go. So super practical and issues that you might. No, not anticipate unless you Well, certainly I didn't anticipate the Walking Dead BM problem in conquered. Another challenge with this kind of work is lack of ground truth. So I can't show you a map and say here's the true coverage, then you could compare a method 1 method against the other. So ground-truth, that lack of ground truth is a challenge and evaluating solutions and crowd sourcing, which seems quite appealing, right? In the sense that the crowd is where it would be helpful to have the conductivity. So a little bit to your question Santos you about, you might say, Well, if there's no, If there are no people in a region, it seems not so important to have coverage there. Maybe if there's more people, it's more, it's more important. But crowdsourcing the coverage seems appealing as a way to get at that particular conditions. And the fact that someone wants to access the network at that point. But there's a crowdsourcing actually has a bunch of challenges, a bunch of problems. One of them being that they have to have a crowd, then people have to be doing the measurement. They have to be willing, they have to, you know, to me it require some action on their part. Which sky hope does that requires essentially no action. So that's part of why they've been able to gather so much data. But crowd sourcing requires a knowledgeable, willing crowd and crowd. So it's not, it's while it has appeal, it's not kind of a magic answer. So, so we want to investigate like incentivizing people to do crowdsource, data collection of coverage. And, and actually this is what was coupled with a pivot, a research project that was necessitated by COVID. So in February 2020 to students of mine, Beatrice, who's in the PhD program and then Michael Guang and graduated with a master's degree in spring. Had built up a starting point prototype to do measurement collection. And we went up to North Georgia to a wonderful place called the hike in. And in our kind of thinking here was, you know, like I said, Crowdsourcing is appealing, hard to measure. Locations are often sparsely populated. I guess COVID hadn't happened yet, but we were still. Yeah, So COVID hadn't happened yet. The pivot happened in the next month, which was the project taking place. The project site was Native American reservations in New Mexico. All of a sudden, there was no way we were going to travel to New Mexico. In fact, I had we had a planned a trip for the week of Georgia Tech's spring break to go to New Mexico, which I canceled the week before. So we don't have any Native American reservations, but we do have, we do have trails. And so then we, we kind of realized that in fact, whether you have cellular conductivity actually has a significant connection to hiker safety. Some of you have heard me tell this story before. I was hiking with a group of graduate students, long before any of this work got going. And one of them got stung by a bee and started to have an allergic reaction. And I was the only like a grown-up. Okay. And it was a bunch of students plus me walking along. I didn't want anybody to pack, but there was no cell coverage. And I'm thinking I'm thinking like, what am I going to do? Like if, you know, her throat starts to swell up. I think my students get scared when I said this, but I was like, I can take a stick and I can like do an emergency tracheotomy. What do you do? Um, I got estaba with a stick in the throat to emergency tracheotomy. Since then I carry first-aid with Benadryl and a Swiss Army knife. So if any emergency surgery was needed, I would actually have a knife. But anyway, I had had this experience and you can these guys these guys went No, No. Now they should go with me because I am the best prepared faculty member. So that obviously had stayed with me. And so we started thinking about this issue of safety. When we first did this, we weren't thinking about safety at all, but then that story was in my head. So, so we built this app for Android and that works on Android and iOS, we call it hiker net. Michael did the vast majority of the coding. That was also very helpful in the design and in some code review. And it's a very simple start stop interface. And then when you stop, it shows you a hike and then it shows you a list of your hikes and it automatically uploads to our server. So this has given us a chance to do some data collection. And on trails, I mean, you don't have to be on a trail. You can start it anytime you want. You can stop at anytime you want. I did a weekend's worth of data collection because I wanted to. We then it's starting to experiment with automatic ways to extract the, the, the, the hikes or to extract the trips. As it turns out, even if you want to participate in this, it's really hard to remember to press start and stop. So we're trying to think about ways to both give people control over what data is shared for privacy reasons, while also not trying to eliminate some of the burden of remembering. So like even our own collaborators can't remember to turn the thing on and off. As amazing what you can't remember to do even if you want to. So so this is available if any of you are interested. And you go to high coronet that are knocked out, got tech.edu, you can or just Google hiker. And at Georgia Tech, you can find the iOS and the Android versions. And if you want to download them and use them, that would be awesome. Okay, so now how might something like hiker net or maybe even hiker net itself be of benefit to community. So hiker net was, in many ways, we've done a lot of thinking about it in a hiking contexts because we wanted to try to interest people in using it. And we wanted to also think about safety. So in in the July time period, So not a couple months ago, the FCC came out with a Request for Comments in the Federal Register. And this Request for Comments is, is driven by a need that Congress, a burden frankly, that Congress has placed on the FCC. Congress said, the FCC, you need to do a better job collecting mobile availability data, mobile broadband availability. Because as you saw, that is FCC data sets and the Skyhook data sets had a whole bunch of disagreements. So Congress said, this isn't good enough. You know, we need to be able to monitor our progress towards universal access. And the data isn't there for us even to know how we're doing. So. So, so this is a comment on technical requirements to implement a essentially a three-part process. And the bullets are over there, a challenge process to dispute what provider say their coverages, so to spell out how it is that you distribute and by how I mean, what measurements do you have to take over? What time period, how many do you need a process to verify what server provider say is their coverage data or for them to verify to say, look, we do have coverage and how does that happen? And then interestingly, this last one, a process to accept crowd sourced information from third parties. A process to have data that comes from say, something like hiker Annette, be fed into the process by which communities pan dispute coverage data. And you know that, that the, the reason to dispute the data is because when funds become available, those funds are directed to communities that are underserved. And if someone providers say, look, you know your community, we were there. You're just not signing up or you're not paying. And the community says, that's not true. Now, we want coverage, we we can't get it. They asked it the documentation of that as a portion of advocating for resources to come to the community. You know, that's that's that's part of what this is. This is about, you know, in, in Concord that when this, when COVID happened in the schools were online, those kids basically couldn't go to school because there were so many households that had no conductivity. Now they could go does the Zeppelin, which is the county seat. But that county is not tiny. I mean, that that really impractical for them. And even if they went to the county seat them, they would sit in the parking lot, the library or something like that, sitting in parking lots as a model. So so this is very interesting to us and this Federal Register, this request, this thing is probably 20 pages long and it's got all kinds of things in it that are kind of fascinating from a researcher standpoint. Like the measurements have to take place between 6AM AND gate for our time blocks. And then they have to have this kind of granularity. And there need to be at least this many of them and all these things that feel very arbitrary. Like where's the, where's the science about how you launch a dispute? And there's also a user. There's a user part to this too, which, which I think is interesting. How would the community organized themselves to, to, to, to counter, to gather this evidence. It's like a citizen science problem. And then there's all these requirements for how often. And so a platform that would let a community organize itself to do data measurement that then meets the requirements of this challenge process. That would be, that would be cool and that would be useful because communities don't really have the, generally they don't have the time. And they have some motivation. But, but linking between community capability and what's being formally required would be a really valuable kind of a piece in this process. Okay. The next two things I'm going to say, well be quite short because I'll transition to the second part. Oh, my goodness. Okay. Doesn't work backwards. Another one. Okay. Thank you for those comments. Yeah. Yeah. Right. I guess you can, depending on what which of the states you start out in off versus when you forget, you can either go Not at all or way too long. Okay. So and at 1 I, I don't I think I did this on purpose, but man, it was fascinating. I watched my phone. I blew up the map on a day when I had had it on just lives only in my house. And you could tell where the kitchen was and you could tell where the bathroom was. You could tell where the office was because of the density density to this point. So very serious privacy considerations. And we're thinking about those very much in the sort of trying to make it more automated, yet still balanced the right amount of control for privacy purposes. Okay, So coverage is a bare minimum that measure the current definition of mobile broadband is 25 megabits per second down and three megabits per second up. We all have had this experience. You can have good number of me and a single signal strength can be fine, but performance can be terrible. And if you want to measure something beyond conductivity coverage, like what's the bandwidth, then you need multiple measurements. And so the measurement task itself becomes more challenging. And if you further want it to map that to something like the quality of the user's experience. You have, you have. Another, another technical challenge. So we, we have some work where we measured quality of service and quality of experience and then used ML models to try to map between the two. If anybody's interested, I can give you that reference later. Okay. I'm going to say one thing about an effort to bridge the digital divide. This this is not a, this is certainly by no means a full bridge. Maybe this is like a set of stepping stones through the water. Maybe. So there's a lot of things happening at the link layer and link layer technologies. And one of them is technologies meant to allow you to send relatively small amounts of data from Internet of Things devices. So these are out in the field in places where there's very little power. So thinking outdoor settings and they generate, they take measurements and you want to send those, get those back to a central location. And so Laura and SigFox and other tech link layer protocols are making that possible. Very low data rate, very low power. Until we have been, we've asked to this kind of architecture, network architecture question. What might a ubiquitous low data rate network do for disconnected internet users? So if you think about the world as having an Internet which has a particular geographic reach, and then a low data rate network, which is everywhere. How might you use those in conjunction to improve things for people who are in the disconnected portion. And so this is with a project is called Lorax for Laura extends to the Internet. And this is just a picture of a prototype device that students built. Thank you for your assistance in the, in this project. And we did a bunch of measurements. Basically using messaging over IoT for things like. So it's one of our use cases was. You have an online store. You want to be able to do things like update the number of an item, right? You're, you're a craftsperson. You're creating new items. You want to keep your store up to date. You don't want to have to drive to the parking lot or drive to the library to say, okay, I've 10 more of this item at such a small, small transaction that if you could use Laura to a proxy and then the proxy could interface with Etsy is an example of one of these storefronts. Then you could keep your store up to date. You don't need a high bandwidth connection for that. So that's Lorax. All right. Just one picture of the sort of reality. And I, I actually mentioned this when I was talking about the problems with the walking dead. So Concord, George's water tower. So water tower space is like the most valuable space if you're trying to improve your wireless network in a location. So because you need, you need height, you need power, and you need permission. So there are multiple things actually on this Concord water tower, including a ubiquity access point. This is Wi-Fi 6 DOT 3 high points, multiple ground-level points. They're experimenting with pricing, okay, So pricing matters here. They're charging $45 a month for city residents and 55. If you're outside the city, they need a 100 subscribers to break-even and they currently have about 50. So this is this is a reality of of Concord trying to extend its network. Do they guarantee the bandwidth? I don't know what service I don't know what service level. Yeah. I'm not sure what they what they include in their contract. Yeah. Thank the envelope. The ventricle. How? I don't remember. A couple thousand. It's a couple of thousand. Yes. Yes. So one of the problems is that people are locked in have locked into contracts, and they're finding that that's an issue? I think. So. They mentioned this in particular the lock-in as one of the, one of the challenges, actually one of their best sales. They're both methods for convincing people to do this is when their grandchildren, the grandchildren come and want to play video games. And then the grandparents are like, Oh yeah, we do need better internet conductivity. Alright. So this is the end of the section on the digital divide. And this was, I knew this was going to be the longest, so don't panic that you're going to be here for another hour and a half. So this is the seven layer stack, layer 8. Financial, political, social considerations. So the digital divides an equity issue. It's a complicated social, social, economic, political, and technical issue, which is definitely outside of most computer scientists comfort zone. More pressing during COVID infrastructure bill offers promise, but actually getting resources in the infrastructure bill, I mean, getting any government resources to the right place to do the right thing is challenging. And for instance, in Concord, they have tried to go after some funds, but matching is required. They have a $60 thousand a year budget. That's the the city budget. Everything the city needs to do, including higher someone higher city personnel is $60 thousand. They can't they can't meet a $100 thousand matching requirement. So whether the scale and maybe the answer is concourse, not the right scale to receive such a matching grant. But, but then, you know what what what is the right scale and maybe it's the county, but, but a number of challenges with getting funds to the right place and and to the right use. Okay, now I'm going to talk, this is going to be completely different, but I'm not going to try to relate them. This is the part that came for my time at the Computing Research Association. So that Computing Research Association is a non-profit, that is a membership organization. Departments join and they're in industry can also join. So at Georgia Tech, I see. And CSR members and CSE probably is too, although I actually am not a 100 percent sure. It's been around for quite a while. Peter Freeman, I think actually was part of the in the early days of getting it started. And two days ago showin barks, I'll interviewed for the IC chair job. And had this very nice slide showing the progression of her involvement and SIG chi. So progression of her involvement in one organization and service roles over time and I thought it was really nice. I didn't have time to pull out and, you know, there are other things on here. So i is a simplified version. I was elected to the board for the first time in 2011, its three year terms with a limit of three. If you do the math, you'll see that I exceeded the limit by one year because if you're elected chair, then that's a two-year term and they let you stay. So I did a variety of things. I would say. One that was relatively substantial was to co-chair the annual Snowbird conference. Back then, did it two years after that? They tried to get me to do it a second time and I was like, No, no, I'm not doing that. And then I was like the chair and that as a gift at the end, they surprised me. Dan help them. They surprised me with this really amazing sewing machine. And it was in the board meeting, the last board meeting, and this was during COVID. So I'm up in my office, I'm in the board, me and dancers knocking on the door. Now, if you are working at home with other people, you know, you do not knock on the door while people are in me. And so he opens, I'm like I'm animating. And he's like, I know. And he brings in and orchestrated the whole thing, brought in the machine. If I if I had known that was the end result, I, I would even more happily. Then part of it was it was network enabled Exactly. So there are three things that I wanted to kind of highlight that I really feel good about from the time that I was bored chair. The for and this is going to all tied together eventually. So so that a CR I had never had a strategic plan despite, I guess about a 30 year history, which might tell you something because it's actually a pretty successful organization. So, you know, maybe we don't really need strategic plans so much. But some really good things that came out. Actually, one of the reasons why we embarked on this effort was because we struggled to figure out how to involve industry more centrally in the conversations and the CRA was having and we kept running up against, well, how about we do x, which cost this much money? Or how about we do y, which costs this much money? And it was always enough money that we couldn't quite get there to commit to it. And so running through, running through a process of deciding priorities, then helped make it possible to say Yes, this is right thing. This is a good thing to do. And that is the co-chair of the CRA industry group, which is which is relatively new. It also elevated some other things and I guess maybe the one here that does connect in somewhat to what I'm talking about today is responsible computing. I also wanted to highlight some of the other Georgia Tech involvement. So Ayana Howard lead a group that refined the mission statement. I mentioned the vec already. A second thing is the Computing Innovation Fellows Program, and I'll talk about that in the next slide. And then the third is, which doesn't seem like a, I don't know which I think was actually, is actually a pretty big deal for the organization, but we'll only resonate with and will resonate with us probably smaller number of you. But there had been this is Andy burnout and he had been the executive director of CRA for more than 20 years. And he did a fantastic job. But 20 years is a really long time. And the best part about this is the Andy has a smile on his face. So multiple times there had been discussions like, we need to think about transition. We need to think about what comes next. And I, I feel really good. And, you know, some of this was the time just then was right. But I also think part of it is how it how it went, how it happened, that we have transitioned and Sue and they will announce the new executive director. And I'm very excited about, about who that's going to be. So, so and a smile here is the positive part of that transition. So the strategic plan kind of talks about, about CRA, continuing to excel on a set of dimensions that I think are really valuable actually to label, right? To, yeah, to, to talk about what it is that CRA does well and has continued to do well because sometimes the work relatively invisible and actually this is a, it's a weakness of CRA that it doesn't make it more visible and more possible for people to have input and influence into what CRA takes on because the organization can do a lot and has done a lot. So content, best-practice memos, those have been, some of them have been very influential over time. Second one is a kind of CCC Computing Community Consortium nod about novel approaches being our respected voice. So in my time as chair, I signed, I signed for for CRA on several Communicates that we're basically saying, here's the computer science community position on x frequently on things at that time that the Trump administration was doing that we thought were very damaging to the research ecosystem in the US. And then a champion for diverse, welcoming and equitable, equitable research community. But then two additional work, establishing itself as a leader in promoting responsible computing. And then a catalyst for bringing together the ecosystem of government labs, academic research, locate research and industry research. So acknowledging the ways that, that ecosystem needs to work together. Okay, That's Computing Innovation Fellows Program. I'm proud of this because at the time when COVID was first happening and everything was crazy, a few of us, enough of us had the foresight to sit down and say, This is really bad. You know, what could, actually, what could see or I do? What, what role do we have? And to realize that the academic job market completely tanked. And that this was the middle. You're by analogy to the two thousand, eight thousand nine economic crisis. And so we took a proposal that had been written at that time to fund post-docs, kind of dusted it off, updated it in eight weeks. Conceived of this, wrote the proposal, hadn't reviewed, and got the funds. And then in the next eight weeks did a marathon reviewing at least in the first round of 550 applications from students for postdoc positions. And, you know, These are the biggest numbers on my CV in my, in my funding category. So yeah, 16 million and then 20 million in 2021. And actually, it was a really it was really gratifying to work with NSF on this. They were very interested in having this happen to, you know, to, to keep the pipeline in, in, in good shape them we had fellows from the previous time who basically said, I would have left academia. I just flat out I needed a job. I would've taken that industry job and I really don't think I would. I'll come back the other way. So that feels very good. So reflecting on CRA, I think Georgia Tech and see or I have had a very long and very positive interaction with one another. I actually, that the list of everybody I could think of who this then Georgia Tech faculty and on the CRA, I included the CCC. So Beth Well, I concluded CCC leadership. So Beth is on there. I may have forgotten some folks, but I really think that it's it's provided for me, Anna, and I know that that could speak to the same, a unique opportunity to talk to the top people in computing across different institutions. In the US, mostly are all a little bit Canada and across computing disciplines, right? Like we don't frequently, we go to our conferences and we can talk to top people from all over in our, in our own sub, in our own research field. I can talk to all the top networking people than just can talk to all the top theory people. But this cross, cross areas of computing is really quite unusual. And I think that because Georgia Tech has been in many, many of these conversations and in leadership. So Jim fully was also that the chair the board. Annie probably belongs on this list. I may have left her up. That has given us a chance to influence both be influenced and to influence things. And, you know, the number of meetings where a conversation comes up. And then there's a chance to say, well, we tried this at Georgia Tech. I mean, obviously we try not to be obnoxious, but, but, you know, or people will say like, what can you tell us about the OMS CS program only? Roller as and then we tell him what's great about it. So this chance to be in the room at the table with a bunch of people who are in leadership positions, I think has actually meant a lot to, to Georgia Tech. And there is this chance to influence the field. In some ways you could think of it like going to NSF except you don't have to go to NSF. I mean, just in, in, in a way that there's a chance to influence at a, at a level that is very different from any kind of individual or even big picture research effort. All right, the last thing I want to say in the remaining two minutes is about choosing work for the future. For that, I think choosing work for the future means both choosing work to do in the future. But it also means like choosing work based on what the future is going to be. And so I just wanted to share these and grab these different images. And unfortunately, most of these don't illustrate exactly what I wanted. So I was looking, I wanted to make a point about global interconnectedness and global interdependence and the ways that so many things that are happening right now are that are happening globally and with global influences. I was looking for diagrams are basically they had arrow is kind of showing, showing influence are showing propagation that for some reason those don't. Either on my Google skills are not good enough or they're not, they're not showing up. There was a map of COVID that at 1 that showed spring break locations and then students returning to their universities after spring break, I tried for 15 minutes to find that, couldn't find it. So this is the only, this is the one that's COVID. This is actually the last 14 days, but it's not showing quite the dependencies. This is about climate change. These are undersea cables, so the Internet basically looks a lot like some of these interconnection diagrams. And then the upper left is is the flight, airline flight routes. Actually, I think that's fairly old, but probably also is a pretty good illustration of COVID propagation. And so, you know, this is the world we live in where, where there is so much interconnectedness. The internet's actually always been about interconnectedness and dependence. And I think at least for me, kind of thinking about what broader impacts. That's that section in the NSF proposal that we write down the last afternoon while we're waiting for the routing sheet to go around, right? So, so thinking about real, like, do you believe what you're getting and real in broader impacts? And are you taking that more seriously and thinking, thinking about the heart about that and not not I mean, it's fine if we we, we, we proposals have to get written, but we should not be fooling ourselves about what it means to have broader impacts. I'll skip that one. Yesterday. I lead a panel and cash of who some of you will know from networking. He's doing a lot of work now in energy. He was at Waterloo but he's move recently. Cambridge said, if you want, I don't want to work on, just look at the front page of the newspaper. Like you don't have to. This is I'm quoting it. You don't have to read a bunch of papers. Just look at the front page and you will see what the world is struggling with. And, you know, if you could make some progress on something that you see there, that would be that would be big. And then to hearken back to my layer eight, you know, not working in isolation of the contexts, the social, political, economic context of that we are in. Even if that's kind of baby steps because it's not it it is intimidating and uncomfortable because you don't know what you're doing. When I first started doing ICT D work, Gaetan know Borello from Washington, who was a pioneer in the area and really moved from mainstream computer science told me, the water's only cold when you first jump in. And I have remembered for, for, I think that's a really great thing to remember if you're thinking about doing something that would stretch you. Okay. Run two minutes over, so I'm going to stop and I guess we could thank you. We could take a few questions and then we can also chat over cookies. Are there anymore questions in the chat? No. Okay. Anybody want to comment or ask a question but yeah. Workflow. Yeah. I mean, I think Edge computing potentially has a play, has a role to play, right? So technologies that are kind of moving things closer to keep moving, moving resources closer to people. I think things like municipal networks and community based networks. Which then you'd say, well, what's the technology piece there? I mean, there are serious questions about appropriateness of technology and, you know. Hi, For instance, helping communities understand or make choices between technologies. When satellite is available. That's going to offer really interesting alternative link layer if you want to say it. But alternative network with a bunch of very different properties than what we're used to in our terrestrial networks. So, so that's, that's some of the range of technologies. You know, what thing? Technologies that produce new price points are new performance and performance and price point trade-offs I think are quite, are quite interesting. I also think, you know, not well, low power. This was when I was had in mind about Moore's Law. I mean, maybe what it could mean is like, let's just stop, Let's just stop trying to double the number of transistors every year because, you know, look at the power consumption and, and look at the planet. And so, yeah, Good question. The networking of the film. One thing that's interesting enough, the list. Couple of decades ago we need a specialized equipment to areas that now the equipment at all then we gave example though, Edwards and everything that everyone uses daily basis. But my question is, suddenly think that works well with networking research at the back. You feel that we should be doing more in terms of building our own specialized lab or should we be just I mean, I think it depends on what kind of problems you're looking at. I think for some problems, you actually want to use infrastructure that's, that has the wide area as part of it. For other thing, I mean, you know, building IoT, test beds, things that we know where, where you can get a lot out of doing smaller scale, right? Kinds of experimentation that, those things I think are really valuable. We've gotten a lot of mileage out of partnership with OIT. So treating Georgia Tech and Georgia Tech's link in and out of the Internet as part of our I'm available resources testbed, if you want to call it that. That's been huge and we've really benefited from that. So I don't think there's any reason to I mean, I think the white areas, especially right eye, we're not, you know, there's you can emulate it but it's not the same. So for that, I think we want it for a lot of that work. We want to use, you know, be able to use things that are of the scale that it wouldn't make any sense for us to try to replicate. Yeah. Alexa, you oh, yeah. I actually have no idea. I guess. If Tom Conti were here, I would even give him a hard time or like, we all just we've heard it forever. I I mean, I like I believe and I believe there's a scaling mall that's going to be hit. I just I chose it in part because I think that I want it to be a little provocative like, alright, certainly, when we talk about going beyond Moore's law, it seems like that the complete assumption there is, we need to have more computing, right? And so it's about, it's about poor computing and what if it wasn't? What if beyond Moore's Law was like less computing. So I was just trying to be provocative. I actually don't know. Yeah. Not more that it wasn't technically deeper. And Yakuza was just, you know. Yeah. Yeah. I mean, Ahmed would be a great person to talk about that at some point because his work, he's, he is interested in high-speed and, and, and computation demands for protocol, for protocols where you're trying to scale very, it's a very large number flow. It is interesting. And it's also centrally grew organically. People who exactly what happened. What's beyond that as well? I think say Here I am. I don't know. And I mean, I think that I think satellite satellite is getting a lot of attention. That will be a different client. There will be a different kind of network from a performance, performance and cover footprint foot print point of view. I think there's a there's a lot happening in the physical layer, physical to link layer space. If you look at, you pick up any of our top networking conference proceedings now you're going to see, you know, work along the lines of what Fadell, a dog, a dB from MIT, I think maybe even it was you. It might have even been used. Or **** Lipton who said, when we were talking about him as a faculty, kinda like he can see through walls, right? This was actually Dane our Randall said, I don't want the office next day or so. So a huge amount of, I think, creativity and signal processing prowess happening at the, at the physical layer. And so of course that's in some sense, I don't know if you call that sort of underlying or at either at the edges. A lot of it's meant to be at the edges of the Internet. I find that fascinating work. So maybe I'll answer you. It's like Layer 8 and layer 1. We've done a lot. We spent a lot of time in layers two through four. Wow.