Session Date/Time: 23 Jul 2026 07:00
[00:00:47] Curtis Heimerl: Alright. We're gonna get started. Welcome everyone to IETF one two six Gaia. The back clock seems to not be working or whatever, but it's actually 09:00. So we're gonna get started. So just as our normal introductions, we have the note well and the general participation stuff. It's Thursday. If you're a bad actor, you've probably been kicked out by now. So, but we still have to post this. I imagine everything is fine. Everyone sign in to the MeetEcho. There's stuff going on, online there, actively. Discussion and the like, that's gonna be a little out of bound, and then, sort of resources for the the general event. But the big thing of note is, of course, our agenda here, which is somewhat packed, actually. Normally, we shoot for, three to four speakers, but, lots of interest here in, in Vienna, and an ongoing RFC, which we would appreciate some feedback on. So I think, as discussion with the other speakers, we were going to start with Leandro, then Marcelo. Sorry. Leonardo's, then Marcelo, then, myself, Leandro, Steve. And everyone gets twenty minutes, roughly, which should be fifteen minutes of talk, five minutes of questions. I don't think we've aligned the, like, clock thing and meet echo to this, so you won't get any warning aside from me sort of side eyeing you, up at up at the front. So all set. Let's get going. Or I should say welcome for me. I'm Curtis Heimerl. I'm one of the co chairs along with Adisorn and Jane. Adisorn's online, and we'll be sort of helping out from there. So let's get Leonardo's's slides up. And then give this guy a slight control, and I think you should be good.
[00:03:07] Leonardo Maccari: Yeah. So thank you. So I'm gonna present this paper. Just a little bit of introduction about myself. I'm an associate at Cafossco University of Venice in Italy. And I've been working in the past on community networks, specifically in a big project, European project, also with Leandro here that was called Netcommons, which we were dealing and working on and with community networks. And this is a review of work done with from UPC in Barcelona and Gabriela Gemio that was a PhD student of us of both at the time, and now he's in Northeastern. So what what are the table of content introduction? What kind of data set we use? What is the way we design this wireless backhaul modeling the cost, and then results. So before that, actually, why? So what we wanted to do with this paper was to estimate the cost of a mesh based wireless distribution backhaul system based on realistic constraints. So the the question we were asking is we want to make it natural there where there is could be nothing, and we want to know how much this mesh based backhaul could cost compared with the alternatives in that area. And so now I'm here. This is a paper that was published one year ago. I'm here because I would like to collect feedback, understand if there is something that you think we did well or we didn't do well. What are the parameters we should change or use? And encourage and discourage us to go on with this work because it's it's time consuming. We all have to keep alive the the data, and it's not easy. There are way too many details to report here. They are in the paper, and we can discuss them and share them offline. So I'm just gonna give you an introduction and an overview. So the the ingredients here, a practical strategy to design a wireless backbone with point to point to point point to point or point to multipoint wireless links. We don't claim any optimality here. We just take a problem and we try to solve it in a practical way and associated cost model which we derive from public information and enough free award data so that we can do Monte Carlo simulations and try to understand if try to give some robustness to our results. So we'll start with this assumption here. Okay? You have a set view of locations that you want to connect, and, yeah, ideally, you know what are the
[00:05:32] Curtis Heimerl: line of sight link between those locations, and you know where there is
[00:05:36] Leonardo Maccari: a point of presence of an ISP, which you can use to achieve connectivity to the Internet. And you have a certain idea of what the the capacity you want to provide to every subscriber household, and then the problem here is how you design your network. So first, the dataset. This is this is coming from other works that we have been doing in the past. We take we took 10 municipalities in the center of Italy from which we have the the the land shape and the building shape. So in the end, we end up with a three d model of the whole place with buildings and their size and shape. This is just like an idea. This is a urban area, but it it's just to let you see what we have. And then for every building, what those areas there have really bad connectivity. They are generally underserved. The best thing that they had at the time was an XDSL and Starlink. And the idea is that once you you know where the buildings are, then you use GPUs to come to the line of sight. So what you're gonna have, it's this graph here that tells you where your subscribers could be and what are the wireless links that you can actually have in that in that specific place. And then what we do is we retrieve the population and the household numbers from the Italian Statistical Institute, which this is some average data of those areas. We set the subscriber ratio. So we say, okay. Let's assume that 25% or 50% or 100% of the people there are going to be part of the network. And then we can associate a certain probability p that in a certain building, have a subscriber. And that's where the way we perform simulation. So we can actually extract random choices of the subscribers so we can we don't have one single network, but we can try to generalize our results with some statistical confidence. In this case, we we do 30 runs. So that's how we get the data initially. Then how do we design the the backhaul? Well, we start from some assumption on the wireless nodes. This is a typical configuration that we have been working on and with community network for very long time. So you're outside, you have a poll with wireless point to point on point to multipoint devices. And then inside the house, you're gonna have the router that powers everything and does the management. So you're gonna have the routing protocol, and then you're gonna have an an access point at home. We
[00:08:17] Curtis Heimerl: have
[00:08:17] Leonardo Maccari: we we use two frequencies, five gigahertz 11 AC and seven gigahertz 11 AD, and we collected a lot of data from vendors. And we used state of the art path loss model to compute the link capacity as a function of of the distance. So ideally, we could compute this and understand what is the the capacity of a certain link given that we're using a certain device, which could be a ubiquity or microtich or whatever kind of device that we were surveying at the time. Then then what what do we do? So we have this more or less large area, and we want to bring connectivity there. The first thing we do is we split the area in clusters. So we use a clustering algorithm that takes this graph here. Well, the graph that I mentioned before. Again, this graph here, a subgraph of this on which we extract our subscribers. And once we have this visibility graph for the region that we want to cover, then we are going to cluster it because we don't want to have a big mesh with everybody inside it. And then inside the cluster, we are gonna choose one node as the gateway, and we are choosing the one that is more central in the topology of the of the cluster. And then we are going to build the minimum distribution tree rooted at the gateway to reach all the the nodes. And so in the end, we're gonna have one gateway, some leaf nodes, and some of what we call relays with the nodes with more than one link. We actually make the core network that is the network of relays and gateway to edge connected, so it's resistant to a failure of one edge. We have completed another paper that just meets with the robustness of the network, but so I'm not going to touch over this topic here. And and then we take the what we assign a capacity to the customers, so to the to the subscribers, and then we dimension the links, which means we choose the devices, which could be point to point, point to multipoint. And if necessary, we even double the link. We add links to have more capacity so that every subscriber has the guaranteed capacity that we want to provide. And then from the gateway here, we compute the shortest street path, like, on on the open street map map to go from the gateway to the point of presence of the of the open ISP that it's in the area, which we estimate heuristically in in some ways so that we know the amount of kilometers that we have to go to cover to go with fiber from the gateway to the point of presence. And from there, what we assume whether we can reach an ISP. And then so with this, we can design our backhaul, and we can actually enforce a specific capacity per per user per subscriber. So how do we model the cost? This is it's extremely complicated. You're gonna find all the sources of the data in the in the paper because what we are trying to what what we do is actually we we have the cost for all the physical devices. So wireless routers, wired router, fiber itself. We have the deployment costs of the wireless backhaul. So how much it cost to mount a specific node, how much it cost to mount the gateway, and we have the cost of the deployment of fiber to the point of presence. This is a real fiber that the model that we use. And then we have this is CapEx, and then we have OpEx. We have the wholesale capacity that we buy at the IXP. We have the transport to reach the I x p from the point of presence. We have maintenance that is based on the meantime to fail and meantime to repair of the of the devices, specifically the wireless devices that are hundreds. And we have energy cost, which we which we derive from the maximum energy energy consumption of the of the devices, which we scale down. All the data come from public sources, previous papers, or interactions with the Gifinet people. Because at the time, Euronce was UPC, Gabriela was a UPC, so we have them at hand. And so we could ask them, look. When you when something breaks, how much money do you spend to send somebody there and fix the the antenna? So based on this, we could actually do an estimation of the cost of the infrastructure. And while this is just a big formula to to say how much how we estimate the the cost per month per subscriber. And the important thing here is, okay, we have an OpEx, which we divide by the number of subscriber and by twelve months. The rest of the thing, it's amortized in fifteen or five years. So we assume that the fiber is going to be amortized in fifteen years. So the cost for deploying the fiber, we divide it by 180, and the cost for all the wireless infrastructure, it amortized in five years. And this because we found in the literature similar estimation for wireless infrastructure, we assume that the fiber is there and it's gonna be there. You can even resell it and use it for other things. So we are mortising in the longer period while the wireless things probably in five years, you're gonna have to change them anyway. So this saved some results. Again, very dense. What what we have here is the monthly cost. This is the penetration. So 25% of the population, 50%, and 100%. And this is the capacity that we give per per subscriber. Focus just on the on the square zero. If I have a a minute, I will also comment on the on the x. What you have here, this is the the cost of the network when you're using 11 a d. So 70 gigahertz, and that's the same when you're using 11 a c. And with 11 a c, we do some, like, basic connectivity. Okay? We we well, with a c and five gigahertz, we focus on cheap things. And with AD, we try to reach higher capacity. And this linearity here is mostly due to the cost of the wholesale capacity and transport. And this difference here is the cost of the devices. Again, updated to more one and a half year ago. And the cost with penetration, 25%, it's kind of reasonable between 30 and €60. That's the monthly cost. And, of course, if you increase the penetration, the cost decreases because because the the the dancer network repays itself more if you have more subscriber. Okay? So if you want to reach somebody that is far away, you're gonna have you're gonna need a new branch of the network going there and this is it it cost too much if you have to reach only one person. Then this is a comparison with what was available at the time. So the first okay. These are two profiles. This is with AC and this is with AD, and the difference, it's in here. This is the minimum capacity that we estimate for every subscriber. So in this case and this is the penetration that we estimate in well, that we use as a parameter in the specific column. So these are low capacities, so minimum the target here was five megabit per second. We can achieve a little more because we we overprovision. And this the target here was about, I think, 60, and we can achieve more because, again, we overprovision. So that's the minimum that our model says we give to every subscriber. At the same time, we only had DSL FDSL in the in the in those areas that guarantee you 7.5, guarantee you 7.5. That that's what they claim. And also start link that we didn't know what what's guaranteeing at the time. And this is the upfront cost. So the that's if you build the whole infrastructure and split the cost, so the CapEx between all the all the subscribers, this is what they're going to pay just to build infrastructure. And this is the upfront cost that you spend with DSL and SAT. This is normally it is you don't pay it upfront in in Italy unless you paid in, like, two years or three years with your bills, and that's what you at the time you used to spend with satellite. And that's instead the monthly cost assuming that you that the that upfront cost, it's amortized in in the period that I've mentioned before. So in the end, this is the amount of money that you're going to spend with connectivity with a, like, a low profile connectivity or with a high profile connectivity. And this is kind of comparable with what we have with what we used to have at the time with the DSL or with the or with satellite connection. The only thing is here we don't count is that there is no loan here. Okay? So you should probably add three or 5% this because if you don't put the money up front, then you get to have the money from from bank, and we didn't count it here. And then we have those connected households here, unconnected. This means this is the these are the original data from from the state. XDSL doesn't reach 18% of the households, and these are the numbers that we don't reach with our own design. But this is very it's an upper bound because we we don't optimize for that. In some cases, you need, like, a upper pole or a or a try list somewhere or a place or not place in a specific in a specific place that can go over a certain area, and we simply don't do it. We don't do any kind of this optimization. So that that's all on my on my side. We have a well documented model from planning and evaluating the cost of our match based wireless backhaul. It can be parameterized as needed. Of course, the code is is there, and the price ranges we computed are more or less comparable to what was available at the time in Italy as an alternative. And now I'm ready to take questions if you have them.
[00:18:42] Curtis Heimerl: Alright. Cool. Thank you. So the queue has started to fill up. Please use the queue if you have any questions. I'm up first because I jumped in pretty early. So just as a quick question on one of the earlier slides, the selection of the gateway seemed to happen before the cost of the fiber roll to the gateway. Makes more sense to me intuitively to say that you picked the gateway based on minimizing the fiber roll because that's probably the most expensive step. Is there a reason you did it that way, or did I misunderstand?
[00:19:15] Leonardo Maccari: The point is that then you're gonna have flatter links to reach. If if the the gateway is not in the center of the cluster, then you're gonna spend more money for having more capacity on the links because you have you're gonna have longer path. But that that's possible. I mean, we didn't try the alternative way.
[00:19:31] Curtis Heimerl: I suppose that's just another kind of optimization question that's that hasn't been done. Okay. Makes sense.
[00:19:37] Alistair Woodman: Alistair Woodman. So you're costing the levels you came out with would seem to indicate that the telcos are relatively rational economic actors in the sense that they they don't want to serve rural areas because it's actually very expensive. Is that your conclusion?
[00:19:59] Leonardo Maccari: I I don't know if that's my conclusion because in the end, they they got a lot of public money to serve those areas. So Right.
[00:20:06] Alistair Woodman: So if you so if you ignore the public money, then it they're they're rationally not doing that. Right? So the question is, are they pocketing the the that money?
[00:20:21] Leonardo Maccari: It could be, but I don't have an answer for this. I mean
[00:20:23] Alistair Woodman: because did you compare the price that the that they're charging people in the cities? So I presume that they're also only charging them sort of €660 in the cities for service. Right?
[00:20:34] Leonardo Maccari: I think that's the same price you get everywhere in Italy.
[00:20:37] Alistair Woodman: That that's the point. Right? So it's cost effective for them to do it in the cities because of density, and it's not cost effective for them to do it in the regional areas because of lack of density.
[00:20:48] Leonardo Maccari: And still, they they I mean, they miss a lot of people here.
[00:20:51] Alistair Woodman: Well well, right. Because they don't really want to do it. Right? So so you're saying that the subsidies so you could use what you've calculated to invert for the regulators what the actual subsidy should be in certain areas if they actually wanted to get better density coverage.
[00:21:08] Leonardo Maccari: Yeah. Okay.
[00:21:13] Curtis Heimerl: Any other questions? I have one.
[00:21:17] Leandro Navarro: Yeah. I know it's at Leandro. It's a tough question, but do you have an estimate of how the the numbers would look like as of today? I mean, how how is the evolution of the market and to what extent your conclusions apply or different will be different as of today or maybe next year?
[00:21:37] Leonardo Maccari: Well, the this I think this didn't change. I think we didn't have investments in public infrastructure in the last couple of years. So what was reached was reached and what was not reached, it's done. Probably, this the cost of the devices, it's it's much lower. And probably, the the gap there, it's it's reduced. So we we could save some more money. Actually, we were pretty pessimistic in many of our accounting, so we try to to keep the price highs as high so to have an upper bound. And if I have one one left minute, one of the reviewers say, like, what happens if you don't have the the fiber backhaul? Assume that you have a, like, satellite link for every cluster. So this is the price that the of the infrastructure without the the fiber and without the wholesale capacity. And at the time when we were looking for satellite links to for the backhaul, we only had the the price for Starlink, which has an option. It's like a community gateway or something like that. And the cost was enormous. Like, the the the upfront cost was more than the cost of the whole CapEx that we had. So that's what you have if you have a magic uplink, which well, you don't have, but that's it. We do have
[00:22:55] Curtis Heimerl: a little bit of a moment. So I I wanted to add on just like so I'm in a bunch of the wisp groups on Facebook. If you're interested in this stuff, the wisp groups on Facebook are delightfully churny, punchy people arguing about this, and they are just getting crushed by Starlink. And they complain. They sit around saying, hey. You know, like, we're we're we're better on price, basically, but people are still going to Starlink for reasons that they argue are irrational, but, potentially are still irrational. Maybe it's, it it's sometimes about, you know, downtime. Sometimes it's just, like, the big corporation is safer conceptually. How do you feel about the structure of these wireless ISPs given this rollout?
[00:23:46] Leonardo Maccari: I mean, that that we have, we have quite some experience with the community managed networks in Italy. And as we were commenting before, they had hard times when we started to have investments in broadband around. So many of those infrastructure didn't resist because the the provider was getting there even if it was serving bad. So they have the same problem. They said they have the same competition with Starlink. But by the way, Starlink, it's I don't think this price changed here, but but that that that's higher than the one the market offers in Italy. So but the the fact that the the network is community based, it's certainly hard to to keep to keep going. Yeah.
[00:24:33] Curtis Heimerl: I guess we have one more.
[00:24:35] Bill Baraza: Thank you. Thank you, Leonardo, for the presentation. My name is Bill Baraza from Kenya. I head an organization that's also doing a few projects around community centered connectivity initiatives. My question is around the I came in when you registered the presentation, but I just want to know in terms of the the infrastructure, the equipment, some of the suppliers we've been able to use for the radio side and the cost core network because you are right now trying to pilot commission network in Nairobi in far flung area from where we have very underserved communities. So I'd like to know what devices and manufacturers we are we have deployed in Italy to be able to get this costing on
[00:25:20] Leonardo Maccari: Okay. This was completely simulation, but we were using the data from Ubiquiti gears and microtic gears, and we were extracting the Ubiquiti has a planner that you can use to to plan the the links. We were extracting the data from the planner so that we could dimension the links. And so we we think it's pretty accurate for those two family of devices and MicroTic similar.
[00:25:49] Steve Song: Thank you.
[00:25:49] Curtis Heimerl: Alright. Thank you all. I might have messed it up. It was you. How'd you go? I think it's you and then me and then Leandro then Steve. Maybe Steve then Leandro.
[00:26:31] Marcelo Santos: Hi. I am Marcelo Santos from Brazil. I'm a professor at Federal Institute of. I will present today the CN-RMM. I know it's a terrible name. We can prove it. It's a community network readiness and maturity model. So let me know who knows what is a community network. I believe in most of the people here, but please raise your hand if you know what's a community network here. Okay? Most of the people. So I will be brief about community networks here, but we needed to have in mind the purpose of a community network. Of course, the first thing that came come up in the mind is connect and connect to the people. But sometimes because it's expensive and for the community, and we needed to have an another option cheaper. Well, so from my experience in Brazil, in and other kind of communities, we don't need heroes. Generally, we think if we go to the community, connect a lot of devices with fiber, Starlink doesn't matter, it will solve the problem. It will not. Because after six months, one year, the community network will stop because we don't have a governance. So it's import it's more important if we have if we think about a continual network working for a long time, it's more important to have a methodology to give a governance very. So it's not about put a lot of devices with gas leds blinking, but we need to have a sustainability, and we have a gap. We have a lot of NGOs and universities, and I'm from university, creating community networks. But after a few months, the network stopped. I mean, we have a lot of gaps here. Like, for instance, one volunteer leave the leave the community, and he has the whole knowledge about passwords and how the network works. And he it's one one of the problems or one stop, and they don't have enough money to replace this equipment. So community networks don't do not need only deployment. They need continued autonomy and learning. We needed to transfer the knowledge to the community. It's the big challenge. So what we propose here, it's a readiness and maturity framework. We have a lot of manual toolkits from IASOC, AAPC, a lot of antics with fantastic materials, talk about community networks, how to implement, how to choose the better device, Mikrotik, doesn't matter. It's a link. But we don't have a material talk about the life of cycle guidance to have this network working for a long time. So we propose a framework architecture based on three pillars. The life cycle stage, five stages, and we start with community readiness, planning, deployment, operations, and evolutions. I think it's the main contribution of this work. Based on in in some dimensions, social, governance, technical, financial, local capacity, partnership, and regulation. And with some immature mature sorry. Maturity steps, like initial, defined, established, and systemized it. This slide shows the main contribution. So we define what we needed to do, not exactly how. For instance, in the first step, we needed to to have a mobilization of the community. We needed to talk with the leadership, And it's fantastic if if you think about it. It's not enough go to the community and put the equipment device, connect fiber, and so on. If they don't have the intention to maintain this network because the community networks, I'm not the owner. We don't have a provider. I will give a short example about that. I went last year to a community net to a rural zone community, and I spent forty minutes talking about community networks with the people there. And they answer me, professor, we don't have union here. We don't have we want have this mobilization here to build a community network. We want only a cable of Internet, but we don't have we we would like we wouldn't like to have more effort because we don't have time. We needed to put water in my culture here, and we needed to listen to the community. But in my purpose here, we have these k processes. We have stages, stage one, two, three, four, five, and we have this escape process. So in the first stage, we needed to mobilize the community, create a diagnosis. After that, we can, in this stage zero, realize that community networks don't have doesn't have a fit with this community. And it's a good result. In their stage one, we plan the design of the network. And the stage two, we think about the deployment, inventory of the devices, the documentation installation. If someone leave the community, just read the material created, we need to have financial management in the stage three monitoring, maintenance, and documentation to guide the community. And what what is in what is interesting here? It's because we say what they needed to do, not how. And they give example better about that. We say, okay. In some stage, you need to have a financial plan. And we have a toolkit from I look fantastic. And when we open the toolkit, we have Excel with a lot of fields to to fulfill. I can't say for agriculture use this Excel. It's very hard for him. So we need to we say here, you need to have a financial plan, but how the community decide. It could have a notebook with a pen available in some place for everybody look. They needed to have this plan, but we don't find how how we implement that. So we say what they needed to do, not how exactly. And if we are they needed to have also evidence. We have a framework. We have steps. We have stages. We have key processes, and they they needed to have evidence, for instance. Okay. They will plan their financial during the time. And what's the evidence? Okay. The notebook with a photo. We
[00:35:13] Leandro Navarro: needed
[00:35:13] Marcelo Santos: to mobilize the community, the leadership, and the community, and so on in the beginning of the the network in the stage zero. And what's the evidence that they did that?
[00:35:23] Rumesia: A photo.
[00:35:26] Marcelo Santos: Paper with the signatures from everybody. And here, we have examples of evidence like a meeting, minutes, photos, financial records because we can use some entity to certify this community. Maybe I was looking, maybe another entity. If this community implement this framework, we can have a certification for this community. So what c and r m m is not? It's not exactly only a GT 3.0. It's not the the final purpose here. It's not a bureaucracy. It it is a little bit, but the community can decide how implement that. Not all the stages, not everything that we we propose here. It's not a race for a maturity score. It's not first all deployment receipt. This work, it's a facilitation for implement community networks. And look this figure here. In the beginning, we have a peep a person with knowledge, of course, high knowledge to transfer knowledge to the community. But after stages zero, one, two, three, this involvement of dispersal or entity is minimal. We we have here the community autonomy in the end of the process, a self sustainability for the community. So this proposal is not about only bureaucracy. It's about helping community networks stay alive, autonomous, and meaningful over the time. So thank you.
[00:37:46] Curtis Heimerl: Alright, everybody. The queue is open. Messed up using my moderator privilege for a moment. So okay.
[00:37:54] Leandro Navarro: Hi, Leandro Navarro. I thank you very much for for the effort to to structure the the how communities well, which are the ingredients that community and the networks use, have, whatever. And you you mentioned about the what and the how or or something like that. I I think there's a, I mean, a tricky point in which we have to find out to what extent we are able to provide something useful for, well, for the for everyone, let's say, communities or observers or Mhmm. Yes. Whatever, in in a way that it doesn't it it describes, but it doesn't prescribe things. Because if you compare, for instance, with software development, you know, there's a typical discussion about the waterfall model or the agile methods among other methods. So it's it's a tricky point to to to be useful, but not to be prescriptive. Mhmm. I've seen different community networks evolve in in very different ways. You can even see the opposite. Let's say, typically in rural areas, sometimes when you go and and discuss with with them, about, creating a community network, they say, okay. That's what we do with water, with forests. So we know how to do. We have a governance. We have everything. Oh, tell us how to apply our practice practices for networks. And they have a different start from from a champion that starts by himself and tries to create a community around. I like very much, one, researcher, Elinor Ostrom, a Nobel Prize of Economy, who who said that what works in practice should work in theory. So it's, I think there's a challenge in in trying to see how communities resolve their problems and organize themselves in a kind of strange manner, but works. Mhmm. And then try to observe and extract some patterns, which can help us to collect them and offer offer them processed to to to the community. So that that mean, there's a lot of potential here, a lot of knowledge that can be kind of gathered and and and offered properly to the to the community as a as a document from from the group. Thank you.
[00:40:14] Marcelo Santos: Thank you.
[00:40:15] Rumesha: Thank
[00:40:15] Marcelo Santos: you. I think what what we can discuss and improve is these steps and the stages. I think it's the main contribution like you described before. It's a clear step that the whole community network should follow. But with what key processes? So here we have trends trend seven, and we can discourage you what they should to do in each stage. I think they are good discussion and improvement of the work.
[00:40:52] Leonardo Maccari: Hi. Leonardo from of the University of Venice. Just to add that to to what Leandro said, in the project that we had at the time, netcommons.eu, there there is a significant amount of work done on also on this subject. So there there is a lot of overlapping. There are descriptions of how on the trajectory of many community networks that have been going through all those steps. And there is also a manual, like, a how to deal with most of the things that you mentioned because they are effectively. They happen all the times. And we even had, like, a a card game that you could use to discuss with other people in the community. So to raise those points without making it look like, you know, a lot of bureaucracy and stuff that you have to do on top of the time that you still don't have to go on the roof and mount the the wireless device. So check it out.
[00:41:44] Marcelo Santos: Okay. Fantastic. Like you described it before, it's important to, for instance, to meet the cost, of course. And it's we say it's necessary, but we don't say how you can use, for instance, your methodology. But we we will we we won't put specifically how to calculate the cost. We need to say it's necessary how the community decide with the entity or deployments, community network. So here, it's a framework. It's a step by step. And okay. Thank you for that.
[00:42:21] Leonardo Maccari: But but based on our experience, I mean, it's double edged sword. On the sometimes you have people that just want to know how and what they have to do now. Sometimes you have other people that don't don't have that mindset. Okay? You cannot just prescribe them that you have to do this. Yeah. But you have to explain them why it is important to do so. So it it's very local specific.
[00:42:42] Marcelo Santos: Yes. I completely agree with you. But from my experience also, if you don't have this transfer of knowledge, the community network tends to die.
[00:42:52] Leonardo Maccari: Yeah. Agree.
[00:43:00] Eduardo: Thank you very much, Marcelo, for representation. It's a bit of a two part question. You had a list there when you mentioned that you sort of, like, give a stamp based on the information that you gather from the committee network to follow-up how they are evolving in through this the stages. My question is, does the give a a certificate in any way, shape, or form? And if you do or if you don't, but what do you think about having such a stamp of improvement from a a university or someone saying this CN or this committee that will have has evolved to a certain stage and how that might help getting them more financing, particularly for stage four. Do you think that's something that is useful or
[00:43:53] Marcelo Santos: or not? Yes. I think it it's very useful. It's a initial work. So it's here, it's only the first presentation about this work, to be honest. So let me try to explain a little bit better. It's totally totally useful to have a certification, a entity that give us certification and so on. We don't have something like that yet, but and we don't have a budget to do that and to give the certification. But it's extremely necessary to have a main entity responsible for that. And I go beyond that. It's important to have a entity, for instance, like Isaac. And it's important to have, older entities trained with this framework, like a NGO. Like, we can have in the top, Isaac. We can have a NGO trained by I look based on this framework, and they can have after that, they can implement help or give some support to implement community networks in every place. So in summary, it's important to have these certifications for the community, but it's important to have a certification only also for some entities. And, generally, it work better when you have something in the top organ organizing everything.
[00:45:25] Curtis Heimerl: And just a brief reminder maybe is to if you can join the Meet echo queue, it helps the notetaker so they can get your name more easily than chasing your dev. Okay. So I think I think we'll start. I'm I'm in the queue next, but, basically, I've been chewing on what Leandro said a little bit in my head, and I'm talking next anyway. So I'll just, do that. So thank you very much.
[00:45:45] Marcelo Santos: Thank you.
[00:45:51] Curtis Heimerl: Take back slides. Maybe also just a reminder to sign in in general. Yeah. In this case, it wasn't, I couldn't find when they're not signed in. Yeah. Please sign in. Hi, everyone. Yeah. And that was a really good question, Leandro. I'm just thinking through, like, how technologies map to social structures, and how some are better at certain social structures and other ones. I think a big thing we did in Indonesia was not try to change the existing structure of whatever, and we just went with what our partner had structured, which was like a normal for profit entity, which we got to push back on. But, I mean, that network lived for over ten years. Interesting. Okay. So, hi, everyone. I'm Curtis Heimerl. I'm one of the co chairs of Gaia, but, come here with a talk this time. This is actually a presentation from HotNets 2024. It was a paper that we published there focused on broadly human methods in networking, of which I'm one of a small set of researchers that really live in that space. So this is a talk for the researchers in the room and specifically the networking researchers with the hope of sort of expanding your mind a little bit to other methods and potentially providing a nice robust space for debating those methods. Because often, when I present this kind of stuff to deeply technical researchers, I get a lot of pushback on maybe the scientific validity of them, the value of them, and these kind of things. But I'm one of a handful of people who regularly publish at human computer interaction venues as well as networking venues. And these are kind of hardworn battles over a long time that, hopefully, what comes out of this is I didn't start my timer. Sorry. I did start my timer. The the hopefully, the thing that comes out of this is a little bit more in your quiver as researchers to bring. And the one I'm really gonna harp on is participatory action and action research in general as a method. So a big piece of this is to say that an enormous amount of human centered work happens in networking research. And we're not even talking like we're in Gaia, and you can see there's a strong community networks theme going on in the room. And so as researchers, as networking researchers, this is a group that I think does more sort of direct human work than most. But even like your data center researcher working at, you know, at Google actually has, you know, hundreds of hours of meetings with various stakeholders trying to understand the problem and the ecosystem. And then they write up this paper where it's like this number is better than this number and so the system is good. And all of that human side of things just kind of gets removed. With the intuition being that that human piece of stuff is like not scientific. It's not valid. It's not interesting. And so the scope of this paper and again, if you just go you can go to my website or or hot nets and see the see this paper there, is trying to expand networking to take into the the those pieces, not just because that research happened, but also because that's critical to understanding the research contribution that happened, and understanding the value of it. And so all of that work that happens, I think the IETF is one of those places where a lot of it actually does happen, is important to bring into your research, not just again to broaden where you can publish, but also for correctness. Like understanding why something happens and why decisions are made is just as critical as which number is higher than than which other number. And so the argument here is, of course, that these insights should no longer be treated as this kind of background noise, this unscientific thing that happens, but instead as as central pieces of what networking research is. Now there is a long history of human methods and huge conferences. CHI as a venue is the big one in HCI, and that is larger than SIGCOM and NSDI put together. It's an enormous, very silly conference for the record if you've never been. It's one of those ones with, like, 20 parallel tracks. And if you get bored, you can just wander and find the, like, weirdest session, my favorite of which was foot based interfaces. There's a session of just five papers on that. They also have a really interesting subset of people who do human or sorry, animal computer interaction, which is like apps for your dogs. So you can always find something interesting at Kai. I highly suggest you go. You know, this is no value judgment positive or negative on the work, but it's fun to sit in on something as a lay person I can do relatively easily. But through these human methods, we kinda picked out three that we thought the networking community in particular should make more use of. These are participatory and action research, ethnographic methods, positionality. There are many many other ones. Surveys are interestingly one that actually do get used in networks, but get used very poorly. We left that out of, out of this particular paper because that's like a whole other side paper of me yelling about statistics and survey methodology. But woe to the person who gets me assigned as their reviewer, at SIGCOM when they have a survey in their paper because it's often, not very methodologically strong. But anyway, I'll step into, these three, and we're really gonna focus on action research. So I think it's actually just what we all do and don't have a word for. So action research. Participatory action research, known as PAR, strives to have active partner participation at all levels of the research. The basic idea here is to work with partners first, like pick out partners, and then work from project ideation to evaluation to empower the partners to actually solve their own problems. The idea is you'd go in without a problem in mind necessarily. You have a toolbox. You're a networking researcher. So you're not gonna go in and, like, not work on networks. You're gonna say, hey. I'm a networking researcher. We'll find something. But you don't come in with, like, I'm going to optimize this particular thing. Please help me with that. Instead, you go in and say, what sucks? Right? And then the person's like, well, this sucks. And they're like, okay. We'll work on that together. And that's really the heart of participatory action research as a methodology. Now what's interesting is it does go all the way through to evaluation. You have to evaluate a way that, like, your partner cares about, and your partner understands the output of it. And this can be a little painful because often the partner, what they care about is maybe not what necessarily the research community cares about. So we work with partners through and through. You iteratively develop and evaluate these interventions that you decide on based on the goals, that the partner has. When we say social change here, that's a a fun side discussion on the sort of history of these methods. They come from activist communities. And so fundamentally, by design, action research is supposed to work on social change. Now that is obviously a very, broad definition. And so as long as you're working with a partner that's focused on some kind of impact thing, you're doing that. And then you actually ideally even include partners in the authoring of the work. This can sometimes get a little bit wonky as well because they don't really care or know what SIGCOM is. But can that's really the ideal when you look at someone's paper to see the actual partners on it. Our big critical argument we make in the paper is really that, like, this is what we all did. This is the Internet. This is even the IETF. Right? The whole point of this is to get the interested stakeholders in a room together to understand what their problems are, work with them on those problems, and output stuff. Now the interesting thing with the Internet, of course, people probably have a working understanding of the history of the Internet in the room. Probably some people have a better understanding than I have of the history of the Internet in the room. Right? But circa, you know, nineteen sixties, DARPA is like, hey. We need to build a new network that can survive a nuclear missile attack. And so they engage with some researchers, and those researchers back and forth. Now you might say, does DARPA is is DARPA a social change organization? I would say yes, for better or for worse. But, right, they weren't making these choices just for financial consideration. The point wasn't let's make the most efficient network that we can charge for. The people in France building the GSM network were doing that, and those network came up at the same time. And so basically, everything that happens here where you, again, go to a partner, work with that partner, they identify the problem, you work on them with the problem, you output stuff, is action research. And I think it's really fundamental to actually the history and and practice of computer networking and computer networking research. And what's interesting to me is it's just something that we don't talk about. All you do is come out at the end of the day and say, well, we've made this particular algorithm or this particular system faster or slower, but not really what's the problem that the organization we're working with is trying to solve. Now if you go to a big academic conference, the answer is almost universally at this moment, we are trying to get more money out of our infrastructure. But that's still I think a fine answer to understand in the forefront. And so we had this paper at SIGCOM twenty twenty four on our architecture for the Seattle Community Network, which is a cellular network that we run there, in a community style fashion. And in that paper, I think we we specifically caught that we're doing action research with our partners. This manifested for us in that we built a distributed core network architecture. The critique of it we got from the reviewers is why don't you do this in the cloud? Why are you building this distributed architecture that's more fault resilient? If you just put everything in the cloud, there won't be a need for that. And the answer is our partners don't wanna put things in the cloud. They don't trust the cloud. And, like, that political angle and that sort of, goal of social change from our partners really drove the design of the work, in a way that, it was really nice to be able to forefront in that particular paper. So on from here, but, again, I think par is the most interesting one. There's also just a broad set of ethnographic methods and, where's my clock? Okay. I'm with Ethnographic methods have a really fun history as well. So ethnography, if people know, is this like branch of anthropology, where the big idea is to go and actually embed yourself in a community for a long time and then really write about your experiences and perspectives on that. Sometimes, traditional ethnography work as you go and you hang out, and and the big thing is to, like, do the work with people. So if you're doing an ethnography of, a particular workplace, you should work there. It's not just observational. That's sort of traditional anthropology as someone sitting and, like, writing about what they see. Ethnography, you should go and you, do the thing with everyone. Right? And this goes on to a space called autoethnography where you write about yourself doing that work versus ethnography, you just participate and then write around the the people, around you. And the big goal of this, is, I think the interesting distinction from what people consider traditional science and this is that you're no longer going for generalizability in any meaningful sense. The goal of ethnography is not to develop knowledge that replicates across multiple areas, but instead really generate knowledge that's very specific to an environment. With the ideas that there's so many different environments, that specific knowledge will be really interesting. And sometimes that can be attention when you're publishing work because, the question will be like, how do I make use of this? The answer being you don't, but is it interesting? And hopefully, the answer is yes. So to do this, what you really do is situate yourself, inside of these relevant communities. This is the tribal broadband boot camp where my student Esther has been teaching for god, it's probably five years now. Teaching wireless, at these particular environments. So she goes, she's never written a paper on it. She just kind of does it because she really enjoys it. I guess they pay her. That's actually fair. But, yeah, it it is, I think, that broad piece of ethnography research where she's there to participate, engage, and and be a part of the process and not just be there observing the process. And then you have to document everything, which is my pain point. I actually quite I'm quite bad at this, as a method because I'm not that kind of person who takes notes. But, taking notes every day of everything that goes on, your thoughts, your experiences, so you can go back to them. And you use a set of tools, a set of methodologies to be able to, glean information and distill information from your big set of notes, to come out of that. And that's coding, which is this widely overloaded term in the world. But you do this sort of ethnographic coding to get data out the other side. And that's that formal understanding there. We had this great example in the paper. Again, if you want the deeper version, go take a look. One of our authors doing this in IXPs. So she went and she participated in IXPs down in the global South, I believe, in Central America, if I remember right, watching how BGP peering happened in that particular environment. You know, you go to a sitcom, maybe even here, and you talk about BGP and these technical arguments for what gets supported. And then you go to where, like, a Nanog or these other sort of industry environments where BGP peering actually happens. You know, and it's like, let's go get a drink together and talk about the peering. It's not like a protocol discussion. It's often a really informal personal business relationship discussion, with various stakeholders. And being able to document that, and the role of the protocol in that process, was a key thing that she was able to publish that, was really cool. Lastly, and I think most interest or not most interesting, most most, controversially is this concept of positionality. And positionality is something that comes from, again, these other spaces. And the big idea here is as a researcher for you to kind of say what perspective you're bringing to a a piece of research. So what you'll do is you'll question your methods, your results, everything where your personal perspective influences that. Now traditionally, what this is is an understanding that, like, if I as, you know, a a white guy from a large public university in The US goes to Indonesia to do my work, I'm going to be viewed in certain ways. In Indonesia, for example, I was viewed as a missionary. I am not a particularly religious person, but that's what the white people there were. And so I would just immediately put in that bucket. Right? And that's relevant to the research, like where I go, what networks I get to deploy, what interactions I have. They're all affected by that reality. Now you might ask if you're publishing something about data centers, maybe that preserve positionality is not that relevant. And I think everyone kind of reflect on how their positionality does affect the research. But I think if you're a researcher at Google or if you're a researcher at a university, that does affect that research. You know, you're gonna have different doors open. You're gonna have different perspectives on you. So it doesn't need to be this kind of identity style of positionality. There's a whole other genre of this. My favorite version right now if you look at the operational systems tracks at these conferences. These are places where industry folks can come in and present kind of industry perspectives on research, like stuff that they reflect on on their own kind of work in industry. And the media question is like, what's the positionality of one of those researchers? And we went back and we looked at a bunch of the papers, and invariably, the reason people work on that work is because they're paid. They're just like, that's their job. Right? That's the position now. A 100% of operational systems tracks papers at these conferences are that, and it's just subtext at this moment. Right? You just look, this is a researcher from Google working on this Google system. They're doing it for that reason. Right? And so we're right now trying to draft up an operational systems track paper for our Seattle community network to sort of have a different, positionality come to that track that I don't think is in present in the networking research, at least, in in the papers I've I've looked at recently. And so, this is sort of that reflecting on your role in the research. What of your identity? What of your perspective? What of the way you present yourself is relevant to this research? Put it in the paper so that people can recognize that that piece of the contribution. And then, we had a little bit of a thing on on how the different communities view research communities engage with that. Here's our example from the Seattle Community Network, which is again that, like, we are university researchers, activists, obviously somewhat radical. And that really, I think, pushed that SIGCOM paper that we had. We were able to put that in there and say, hey. Like, look. We're building it because we're radicals, not because we're trying to build a more efficient core network. That's explicitly not our goal. So the general recommendations if you're gonna start to kinda go in this way with your your research work, is to document partnerships, detail everything. Just write stuff down. It's kinda weird to document your one on one conversations with people, but it's really important to be able to reflect on. This is data in the system, and and and then actually reflect on it from there. And so here was our call out for the network research community, which was to, like, actually allow this in the CFPs. We haven't gotten there yet, but I'm gonna, you know, keep fighting my battles. So that's it. This was a big collaboration from a whole number of researchers. We actually had an NSF grant to do human centered networking in a workshop, which we held about a year and a half ago. And this is all kind of our dissemination steps for that particular thing. We'll see what happens from now. The NSF is a little bit chaotic as you may or may not have heard, but they, at least the program manager seems to still like us. So, I will end there, and with a small amount for questions. Christopher in the queue.
[01:04:00] Chris Locke: Hi, Curtis. Great presentation. I'm Chris Locke from Internet Society Foundation. Linking this back to the previous presentation, I think one of the absolutely fundamental values of research approaches like this is actually in helping inform long term sustainability of the community network programs. If you understand the human level of usage and the economic and social aspects that people bring to the usage of their system, it fundamentally helps you building community networks that absolutely are eventually long term sustainable. I did a previous role, funded research that examined Facebook zero's program in India where we discovered that actually it hadn't brought anyone new to the Internet, but was being used by an already partially wealthy middle class to reduce their overall mobile spend on a monthly basis. So doing that, I think, really links for and and one of the things on top of the recommendations you make, I would strongly ask everyone working on any form of community networking, is having that deep understanding of the usage patterns and the activities that people bring to the way they use the networks in social structures and familial structures as a way of informing what sustainability looks like.
[01:05:15] Curtis Heimerl: Yeah. Thank you for that. Think that's absolutely correct. And, you know, I I one big piece I wanna reinforce is that I think a lot of this is about correctness. Like, without this information, like, you're missing fundamental elements of the research. So if you write that community networks paper without these elements, you you like, it it it's harder for people to replicate. It's harder for people to understand. And so having that data is really important, for telling the whole story so that other people can take it and run with it, which is the goal of science or a goal, I should say. We have a Sophie?
[01:05:46] Sophie: Thank you very much for the presentation. Really fascinating. I did research at OSAX piece, but at the kick, so at a larger one. And I'm really interested in curious what do you think should there be more research on not just community networks, but also to genuinely also when it comes to positionality, reverse the gaze in a way and also look at very corporate for profit models?
[01:06:09] Curtis Heimerl: Yeah. I mean, this was a presentation for, like, the entire networking community, so I completely agree. I think there's you know, even, like, IXPs are such a fascinating environment. I've only, like, bounced off. And my favorite version, I was in Indonesia, and one of my friends connected me to the big I x p in Jakarta, which handled 80% of Indonesia's traffic, which is an enormous country. Right? And it's just a mom and pop, and they live upstairs. Like, it was just mind boggling to me, that whole ecosystem. I understand it's corporatizing in ways that maybe makes things a little bit less exciting, simpler, you know, better by someone's definition and worse by another. But I highly suggest you look at Fernando Rosa. She's a professor at Virginia Tech who did that that iXp work as an example of someone chasing that down. It sounds like your work is super interesting as well in that space. So I completely agree. Like, this isn't just about how to do community networks better. I think the community networks community of researchers, has taken a lot of these methods to heart, over a long time. But I think even into those corporate levels and even like like, the same things are happening and the same questions are there, and the research is missing this piece that's really critical to understanding what's going on at at even highly corporate IXPs. They're still going to the nanogs and and and making weird deals. So thank you for that.
[01:07:26] Marcelo Santos: We have Renzo. Yes. Renzo Navas, IMT Atlantic. Thank you. Great presentation. I have a question. I'm I'm a research. I'm I want to engage on these kinds of things, doing things I I value and and that helps the community. My question is how you find the financing to to do this? Where do you go look to your partners? Personally, I will start in my local community and go see people and public actors maybe. But, of course, maybe it's not enough. How how what's your experience on getting founded and getting people on board or entities?
[01:08:02] Curtis Heimerl: Yeah. This I mean, this is why we're starting at this, like, researcher level as I think we need to reshift what the research community appreciates because a lot of the funders, at least for science foundations and the like, kind of follow that. At the end of the day, I think this particular argument is to say you should do this as part of good networking research. So whatever funding exists for your existing networking, you should just bring these perspectives and methods and be able to acknowledge if you're doing action research, who you're doing action research for, how you pick your problems, what your perspectives that you bring change things, that's good networking research. And hopefully, if we can move the community a little bit, the funders will follow. That's not to say that, like, the general community networking space also has our funding challenges for this kind of stuff. And solving that is, I would say, potentially out of scope for this talk, but, like, true close to my heart, if if it's not clear. Thank you. I think I should stop with Rumesha.
[01:08:59] Rumesha: Hi, Rumesha, PhD student at Stanford. I really enjoyed your talk. Like, I think this is something that the research community really needs to talk more about. During my undergrad, which I did in Pakistan, I remember there was a visiting PhD student from Paderborn who was like well, wanted to learn more about infrastructure in developing regions. And I just remember thinking, why don't people do this more often? So I guess that's my question to you. Because, like, people in at these home institutions, especially in developing regions, are actually really excited to share how their world of the Internet works, but we don't really see that often. What do you think is the barrier over there?
[01:09:39] Curtis Heimerl: The this you know, the research community is kinda myopically self centered in this, like, really problematic way. I I'm sure you've experienced the, like it's a big discussion at Kai all the time of, like, if you do some research, like, you're doing some study on people, you can write the study. But if you do the study on people in Pakistan, you have to put in Pakistan in the title of the paper. Right. Like, same research if it's not in Pakistan, you can just say, like, it's happening. If it's in Europe, it's just happening. But if it's in Pakistan, it's happening in Pakistan. Otherwise, the reviewers will be like, why is this relevant to me? Right. Like, how does this generalize out of Pakistan? But no one's asking how does this generalize out of Germany. Yeah. And it's just this, like, centering of of methods and and understanding both, I think, for HCI in the global north, but in networking, largely in companies. And so when I write a paper, I have to sort of justify that that why we're doing something different than just, like, putting things in the cloud. Yeah. And I think this is part of that effort. I think that Kai's done a really good job at this point. Like, you don't have to do that anymore. You still get some pushback. I'm an AC there, and I have to tell the reviewers, like, no. It's fine. But, networking, I I think we just all have and and I think networking is actually, like, really thirsty for non data center work.
[01:10:51] Sophie: Yeah. Yeah.
[01:10:52] Curtis Heimerl: Everyone is so tired of data centers, but, like, that's where all the money is. Like, that's where Google funds, all this kind of stuff. So I'm hoping it's an opportunity right now to be able to open that door a little bit, and it sounds like you can be part of that process as well. Yeah. Yeah.
[01:11:06] Rumesha: I'd love to talk more about that. Okay. Thank you.
[01:11:08] Curtis Heimerl: Alright. Thanks, everyone. I guess we have Steve next. My thought is the RFC is probably the most open ended, so we'll just run that to the end. You don't have slides. Right? So you don't are you gonna talk, or do you wanna press or you're still gonna have, like, a a thing?
[01:11:28] Steve Song: I'm gonna share my screen.
[01:11:29] Curtis Heimerl: Okay. So then I'm going screen to
[01:11:32] Steve Song: share request.
[01:11:33] Curtis Heimerl: Yes. I didn't get it because I was standing up. I will find you.
[01:11:48] Ben Roberts: So while while I'm being found
[01:11:51] Curtis Heimerl: Hey, Mara.
[01:11:53] Steve Song: What I want to convince you of today or to tempt you with today is to become a mapper of fiber optic network infrastructure. So the reason I and a number of us are here this week is for a governance meeting of the open fiber data standard. And if you haven't heard of that, it's very simple. It's a structured language for describing terrestrial fiber optic networks. There's lots of information available about undersea cables. You can find maps of undersea cables at telegeography and elsewhere.
[01:12:30] Curtis Heimerl: You should have screen now. Although I think my slides are still up safety killers, but you should have screen control.
[01:12:34] Steve Song: Okay. So if I I'm not seeing anything.
[01:12:44] Marcelo Santos: Okay. Let's just
[01:12:47] Steve Song: You know what? I'm just gonna talk. I think and if the screen screen appears, that's great.
[01:12:55] Curtis Heimerl: I will play with it. So
[01:12:59] Steve Song: and linking this back, terrestrial infrastructure back to the discussion of community networks, You know, we often talk about the the the last mile of community networks and the sustainability of the infrastructure, but one of the key issues of sustainability in community network infrastructure is actually backhaul. As demand for broadband increases, backhaul becomes one of the most significant factors in the sustainability of any network. And the key to to solving that that affordability is actually getting to the deepwater ports of the Internet, which is fiber optic infrastructure. We can have a side conversation about Starlink and its impact and, retail versus wholesale networks, but fiber, is the kind of gold standard for the Internet. And yet, when it comes to terrestrial networks, it's really hard to find out where it is. I live in Nova Scotia on the East Coast Of Canada in a very rural area. In my community, we formed a cooperative. We got a, nondominant operator license from the Canadian regulator, and we approached the incumbents, Bell Canada and Rogers, and we asked them, well, you know, can you show us where your points of presence are so that we can connect our network to you? And they declined. They they refused to to share their network maps. They played a kind of a a game of go fish. I don't know if you played that before, but as a child, you know, do you have any fives? You know, you you could give them a location, and they would tell you if they had something in proximity to it, but you couldn't get access to their network infrastructure. And this becomes an issue not just for okay. Whatever the appropriate tab is.
[01:14:43] Sophie: Can I sit down?
[01:14:43] Curtis Heimerl: Yes. Yes. Alright.
[01:14:56] Steve Song: Oh, it's
[01:14:57] Curtis Heimerl: I I shared the wrong one. Okay.
[01:15:00] Steve Song: We've got time. Can can can I share the
[01:15:03] Curtis Heimerl: whole screen? So stop sharing. Looks good. And then you've got screen share down here, it's gonna ask you which one. So why you close it? I'll grant you screen share again. So now you have to pick the right window. So that's gonna share this window, which which is not the window you wanna share.
[01:15:30] Steve Song: Can I can I share
[01:15:31] Curtis Heimerl: the desktop? Tab tier. No. Just I guess it's just window. Okay. Alright. Oh, no. There it go. Share entire screen.
[01:15:38] Steve Song: Oh, it's going.
[01:15:38] Leonardo Maccari: Damn it.
[01:15:39] Curtis Heimerl: There's a button. So stop sharing. Start sharing again. We're gonna get it. I wanna see Steve's slides.
[01:15:50] Leonardo Maccari: And then I grant.
[01:15:51] Steve Song: Oh, sorry.
[01:15:52] Leonardo Maccari: Go ahead. Grant.
[01:15:59] Curtis Heimerl: Share time screen. Now you should be able to tab over.
[01:16:07] Steve Song: Yay. Okay. So this is our goal. We want to we want to build as much information about terrestrial fiber infrastructure as there is about undersea cable infrastructure. And a lot of our work is working with regulators and with operators and changing norms about sharing. And the interesting thing is that there are lots of network operators that share maps of of their of their fiber optic infrastructure. And I think the we can see an example here. This is broadband in InfraCo in in South Africa, and it's great. I mean, this is this is a very positive sign of an operator that shares not only the roots of their network, but the access points, the points of presence. But this is essentially it's a JPEG image. Right? And a JPEG image is not very useful if you're trying to combine that data with other sources with you know, and to integrate it, say, into a GIS system. And so what we've been developing are some tools to help you get from here into into the structured data format that is the OpenFiber data standard. So and and first of all, I mean, one of the one of the easiest things you can do is actually, import, that image as an overlay into Google Earth. And so I can take, for instance, this is this is a map that's publicly available of of BTCL in Bangladesh. That's their national fiber optic optic backbone. Again, it's a it's a PNG image and a little frustrating, you know, as, you know, as a as a as a standalone image. But if I import that into Google Earth as an overlay, then I can stretch that map out over Google Earth and and align the map with the with the borders of Bangladesh and begin to actually then trace that network. And it is it's a little painstaking to do. So, you know, it may be a few hours of your time to to do it, but that you can actually then begin to identify the points of presence on the network and the and the roots and save that as a KML file. Now that's a that's a big step forward, you know, being able to have a KML file instead of a a PNG image. But sadly, I mean, I don't and I don't know why Google did this. When they when they built the KML standard, they thought we won't need any metadata. And so, you know, you have, I think, essentially two fields, a name and a description that you can embed into KML. And so it's not really good for for structured data. So to go the the next step in order to to take that KML file that you've built, that you've traced of this map, we built another tool, which is a, an online converter. It's it's also Python code that you can that can use at the command line. But I can take, a KML file from any source and, and convert it into the Open Fiber data standard. So in this case, I'm actually gonna take one, you know, a little more robust example. This is GRID online. It's the fiber optic network in Namibia run by Namibia Power. And they happen to offer a downloadable KML file that that you can save and and then upload to this KML con to OFDS converter. And as you upload it, you can begin to add metadata about the network. You can describe its status. You an important distinction in fiber operators are is whether you're leasing that capacity or you're the actual owner of the network. And so in in this case, the owner is Namibia Power, and Grid Online is the provider name. And this is a a very modest amount of meta information. We'll get to how you add more meta information shortly. And there are a number of op other options for processing KML because people do the wackiest things with KML in terms of structuring data. And and that's it, really. It it parses the KML, breaking it up into individual roots and points of presence, and converts it into the structured data format that we we call the Open Fiber Data Standard. Now once it's exported, you you have a JSON file and and a and a couple of GeoJSON files, GeoJSON being an I an IETF standard for describing structured geographic data. And the conversion, you can see there. It's it gives you a little proof of concept there that the the conversion has has gone okay, and I can download that that zip file of those files into my download directory. And now we can go to the next step. Having the structured data is actually imported into a GIS system. And in order to do that, we've built a plug in for QGIS, which is arguably the most popular open source GIS tool available. It's called OFDS Studio. It's available in the plug in library. It's it's like two clicks to install it as a as a plug in. And once you once you install it, you can instantly create these these data structures that you see on the left hand side here, which capture rich information about the points of presence and roots on the network. And I can take that that KML file that I've downloaded, which I need to unpack. But hang on.
[01:22:56] Leonardo Maccari: Hopefully
[01:23:04] Steve Song: sorry. Give me one second here. I will just, unpack it. Okay. I can import that that JSON file, and suddenly you have a full GIS system that that captures fiber roots and points of presence. And now in those points of presence, for instance, I can begin to then add richer data about those about the individual services available at points of presence, the kinds of infrastructure, the kind of availability, the a range of issues related to what kind of services are available at that point of presence. Similarly, when it comes to the the individual spans, we you can capture information such as, is it underground infrastructure? What ITU standard is the fiber optic cable? How many strands of fiber are are on that network? And with the goal of creating a robust GIS system for managing data about any terrestrial fiber optic network. And the advantage of doing this is once everyone's speaking the same language about about fiber optic networks, you can begin to integrate that data together. You can do benchmarking. You can develop one tool that works for one regulator, will work for every regulator everywhere in terms of of analyzing and using the a common language for describing infrastructure. We place a lot of emphasis there is a lot of emphasis internationally from the ITU and from other organizations about the resilience of submarine cable infrastructure. And we've missed half of the discussion by not having as in-depth and robust a discussion about terrestrial fiber optic infrastructure. It's the two pieces together that actually tells us how resilient the global Internet is. And so I invite you to to explore some of these tools to try it out yourself. And one of the the great things is you can contribute to building a, a map, and, the the map is here. It is a sparsely populated map at the moment, but, all of the data for this map is stored in a GitHub repository, and you're welcome to to build your own map of where you particularly live or a place that you're particularly interested in, submit it as a pull request to the repository, and dynamically update a global crowdsourced map of fiber optic infrastructure. Thank you.
[01:26:28] Curtis Heimerl: I'm I'm first in the queue, so we have to, like, share this. So oh, and then, yep, people are jumping in. This is awesome for just the start. I think that's a really, really cool effort. Is the like, I guess the core question is why just fiber? Sort of, you can imagine any Internet link being relevant down here. Is it the is it, just that fiber is less churny than other tech? Is it that, you know, the I I guess yeah. Why why just fiber?
[01:27:05] Steve Song: Well, in my opinion, not just fiber, but fiber is a great place to start. It's really the it's the foundation. And so we hope to build out from there. At the moment, we only the standard only covers national and middle mile fiber infrastructure. This year, we intend to extend it to fiber to, you know, FTTx networks, fiber passive infrastructure, possibly to undersea cables. But it's not limited to fiber as a technology in any way. I mean, the another interesting source of open data is tower infrastructure. And, it's it's there are norms here that that just don't don't track across different countries. So for instance, Canada, you can download a CSV file of every tower, every radio, every antenna, every altitude, azimuth. All of that data is public data data downloadable, you know, to your desktop. The same with The Netherlands and Switzerland. They make that data available, but other countries make you know, it's it's absolutely out of the question that you would share. So a lot of the a lot of the conversations we're having about the Open Fiber data standard is really just about norming and and kind of creating a conversation where sharing more information becomes comfortable for for everyone.
[01:28:24] Curtis Heimerl: Andrew?
[01:28:29] Leandro Navarro: Hi. Thank you, Steve. And, also, thank you for your question before, Curtis, because it relates to my experience. In the Guifi.net community network, like, more than twenty years on this, we started, with lots of community networks in Spain. And suddenly, someone, Ramon, came with, an implementation of a mapping system and some other tools. And people said, wow. That's wonderful. And I think one of the key successes of Guifi.net was to provide the tools for mapping the network in an interoperable way, which was a requirement to create a community network, kind of umbrella organization for smaller community networks in in villages. But when it when Fiverr came, people were reluctant to share the mapping of of them because they were they had fear that someone will come and cut the free Fiverr or something like that. So but in the recent years, we found that with the with the explosion of interest commercial interest on Fiverr, we had controversies, even lawsuits with maintainers who claimed that the Fiverr network was theirs even though it was paid by by the community, and even they got tax benefits from donating, that infrastructure to the to the common good. And, now the Guifi.net Foundation is in this law law suites, and they see the importance of, of, making public claims about who invested, who deployed a network, that the network belongs to someone and not some to someone else. So so, this open data is is critical for the survival of the foundation even and to claim this is mine, not yours. And but in general, I think there is reluctance to share this information. And, I mean, in general terms, I I would think about the idea of the requirement for transparency or accountability on on infrastructure that somehow is public. And, I mean, connected to the existence of an inter interoperable way of representing this information that is the, I don't know, legal regulatory requirement, not just to be cool, but to be to to to have a requirement to be transparent, to be accountable, to be responsible for for those deployments. I I don't know if you if you've explored that area and and to what extent this data standard is also an instrument to enable this kind of required national or regional transparency about infrastructure investment or ownership or whatever on on fiber networks or other type of public infrastructures in a
[01:31:19] Leonardo Maccari: way? The
[01:31:23] Steve Song: issue of transparency is one of the hardest things we deal with in in our advocacy for the open fiber data standard. There I mean, we see the same thing in undersea cables that, you know, we can't share maps of detailed maps of undersea cables because of threat actors. And so to begin with, our focus is on well, okay. That we respect your position, but let's at least agree to use a common language for describing the infrastructure. And while we're using that common common language, we can begin to have a conversation about how this information should be shared. Maybe much of it shouldn't be public. Maybe it should be available to the regulator or to licensed operators or to to researchers under certain conditions. Those issues, I think, are going to be negotiated nationally, and and we may see different norms of sharing emerge in different countries. And we don't we don't preach in terms of what's what's correct here, but we do think it's that there's a missing conversation here about what what can be made available and to who.
[01:32:35] Curtis Heimerl: Alright. Last up, have John.
[01:32:43] John Levine: Yeah. I'm I'm John Levine. I think you just answered my question. But first, I think this is great. I I believe there are two fibers running down my street, and I know who owns one of them, but I don't know who owns the other one. You know? And having and having clearer maps would be good. Yeah. But I I was the issue obviously is that the maps also told the bad guys where to cut. You know? And I'm wondering if you sort of some some sort of framework about, you know, which which which ones are sensitive enough that you might wanna hide where they are versus the one that runs down my street where nobody cares?
[01:33:17] Steve Song: I think there are there I mean, there are two answers to that. One is I think there's an important conversation about that about bad actors and the dangers of making that available information available versus the dangers of not making that information available.
[01:33:34] John Levine: There's no question that there's there's real benefits here. You know? So But rebalancing.
[01:33:39] Steve Song: Yeah. My point being that, you know, in, you know, the largest operator in Nigeria and Ghana, MTN, report over 10,000 fiber cuts a year. And, you know, it's like Hanlon's razor never ascribed to malice that which can be explained by confidence. You know, most of these cuts are actually accidental. And if there was more available public data, they would actually experience less disruptions rather than more disruptions to the bad actors.
[01:34:07] John Levine: The crooks might know that that's fiber rather than rather than copper, so it's not worth anything, that kind of stuff?
[01:34:12] Steve Song: No. I I think it's just a lot of a lot of civil works that are happening without any sort of coordination.
[01:34:19] Leonardo Maccari: Okay. Yeah.
[01:34:21] Curtis Heimerl: Yeah. I'm I'm gonna have a little sidebar with Steve afterwards. Because I'm fundamentally confused about why bad actors cut fiber, like, as an as a structure, like, especially international fiber. There's other routes. I don't, like
[01:34:32] John Levine: Yeah. No. I I from what I know, I I think you're right that it's mostly like, What?
[01:34:37] Curtis Heimerl: Yeah.
[01:34:37] John Levine: You know? And and particularly, you know and I and I and I know that certainly in where I I've seen places where there's basically signs on the poll that says this is fiber. It's not copper. You know?
[01:34:47] Leonardo Maccari: Mhmm. Yeah. Alright. Cool. Alright.
[01:34:51] Curtis Heimerl: Thanks, Steve. And last up, we have Leandro. The clicker should work.
[01:35:05] Leonardo Maccari: Do you have the
[01:35:06] Curtis Heimerl: timer? Do. I mean, we have twenty five minutes, so you're gonna run us out? I'll I'll start the timer, though. It'll be twenty minutes.
[01:35:12] Leandro Navarro: Okay. So, thank you very much. I'm Leandro Navarro. I used to be a professor at UPC, but, I I took freedom, from from the sole obligation. So I, as Curtis said before, I took, like, a participatory action research in a in a strict manner, and now I'm more like part of of different communities. One of them is UPC. We have a a an NGO at UPC where we do something related to to action research, which is which is called service learning. We work with students that because UPC has a fund for for for cooperation for development projects. And then every year or every six months, there are projects we in which students volunteer to apply their learnings into into the field, into practical examples, like connecting schools, hospitals, communities in different places, sometimes locally, but most of the time in the global South through pan counterpart organizations, which which facilitate this exchange. The students go and and try to help, but also they learn a lot from those experiences. And, and we support that from different initiatives. Also, the Catalan chapter of the Internet Society, We we support these initiatives, and we apply for the Beyond the Net program in which, we get support for buying equipment. UPC pay tends to pay the trips. And and we have also the e reuse initiative in which we work with social enterprises who collect devices and for reuse. Devices can be routers, put put antennas, point to point links, and user devices as well. Because without a user device, without a mobile phone or a laptop, we cannot really connect to to the Internet. And and this is field work. And in fact, for instance, Maria is a it's a it's an activist who was involved in the e reuse initiative and the reuse of electronic devices. And she started a PhD after that, and then she finished the PhD. And now is she's back to the to the activist groups. Eduardo and Viviana, they belong to two communities in in Argentina, which one is a social enterprise. The other one is a university that also has a refurbishment lab, and and they provide connectivity and devices for students. So well, this is the the second presentation, I think, about the an Internet draft that that has was adopted in the previous meeting by by the Gaia Group. This presentation is similar to the previous one, so I'm gonna skip a lot of already presented details. In the previous meeting, we discussed that we use the term digital sovereignty in the sense that communities that have devices connectivity and so forth, they they want to take their own decisions. Let's say, like, found out that sovereignty is is more and more used in in a national environment. So we we translated that into digital autonomy or autonomy, the concept of being able to decide and choose yourself and not depend on a third party and external party. And, also, all these initiatives have to do with not just passively use the Internet, but typically, what it will come as kind of the concept of meaningful use, which depends on who is defining it. It's a bit different, but there are commonalities in in it. There's also a new, I mean, feedback in in the list so that I that we try to incorporate it into the draft and discussing about security aspects and a bit some tweaks to the document. So we tend to think that community level is something that you deploy and then you use it for forever. But, but in fact, it is not the case. I mean, devices have a lifetime, and, they need to be repaired, re replace, relocated because the users are somewhere else and and so forth. So it's a kind of dynamic, environment. And in our, like, certain experiences with partnerships with Global South communities, we find that sometimes you deploy, you know how to deploy, you know how to repair, you know how to set up a a network infrastructure. But but then when you come back the following year, you realize what happened in in the meanwhile. You suspect that from the communications, but when you go back and once one year or six months later, you you find how things are. So there is a big need, and we have increased that effort over the years in in supporting communities to have repair capacity and and be being able to manage this kind of pool of devices, which are which are the basis for their network infrastructures and and connectivity services. So yeah. I mean, deployment, setting up a network, providing coverage is is it's a requirement, but it's not enough. You need to have all the elements in the chain working all the time, which means you need to know how to procure, repair, reset, update, whatever, all the elements of of infrastructure. And also, of course, make sure that not just you provide the the the technical means, but also you need to also help the community to understand how to use that thing in a constructive manner. So the the document, I don't I I recommend you to read it if you're interested. It's hard to cover all the pages in in one presentation. It describes some principles, also some operational practice, some examples, some practical examples that we've worked with them and in which any device becomes part of the connectivity infrastructure. And and, well, there is a focus on something specific, and we are not really trying to find any protocol in particular. In the section seven, you will find a description of of the different communities that we we have used for for developing the document in which in all of them, there is a con component of connectivity, a component of network and or user devices or both with different particularities, different regions involved, which means we can identify things which are in common, things which are different, things which are the same idea but implemented differently and in in different places. And that covers Africa, Latin America, and and Europe, Spain mostly. So in the document, you will find some definitions about what is meaningful connectivity, digital autonomy, what we call circular device management, which is not just like an environmental thing. It's also like devices can be reused by the same or different people over a long lifetime if if there is local capacity to repair and and refurbish and upgrade these devices. And and a a common interesting thing is that that connects very well with community networks is that in many in all these cases, there is there is a community ownership of the devices. There's a legal term called for that. It's Roman law, so it is not invented yesterday. And in which well, people want to to use the devices, not own the devices. Owning devices is another thing. It's like a imagine, like, a long term library of things. So you get a loan, a device that belongs to the community. You are supposed to return it for reuse or for recycling because it's it's not yours completely. It's it was given by the community to you. You well, if you're a student in the university when you finish your studies, maybe you should return it. Or if you are in a community, as long as you live there and you use it for whatever community program, it it's yours in in a way. And then, well, in the, you will see several, like, core principles that are typically in in this operational experiences we we we include. I I won't read it at all, but you can see. For instance, I said, one of the typical examples is this idea of collective access models, which applies for the connectivity, an inclusive model for accessing the network, but also an inclusive model to accessing devices because with as I said, without devices, connectivity doesn't mean anything. We don't have any sense in the in our brains to to create IP packets. And, again, I mean, all the local capacities, which sometimes, as as it happens with community networks, when the communities grow, they need to formalize themselves better. And, also, they some of the volunteers that typically learn by doing might end up into permanent jobs or part time jobs, which are also an element for economic sustainability of of the local network. And, of course, there's an governance element and so forth. Not only principles, but also practices. I mean, there are some typical tasks that happen in in these communities. Like, you you see here the words that we use. I mean, repair training, capacity building, providing programs to to orient connectivity to certain benefits or users, collective collective decisions, ways at at some point, ways to provide transparency, accountability about I mean, like, accounting books in a in an engine in an medium sized NGO or something like that. You need to prove you are managing your resources properly. For reuse, particularly, it's very important to make sure that, data sanitization is done, correctly. Otherwise, people fear that, my files will appear somewhere else or by someone and can access my personal documents and
[01:46:21] Leonardo Maccari: and
[01:46:21] Leandro Navarro: some other aspects. So this is collected in the in the document. There are the meaningful use. There there were some comments in the list, and and we incorporated in the draft, not just the the words, but also the examples. So we we collected from the different communities that we were that we're working with them. Examples of what is the meaningful use they are They they were targeting or they they achieved through this idea of managing devices, network, and user devices as commons. In the document, there's a connection, but it does hasn't changed much related to well, I mean, the device ability has to do with the right of participate and benefit from from the Internet. Autonomy means that you need to be able to repair your own devices, find the spare parts, so forth. In in Global South projects, we when we go at the following year, we also take the e waste that is there back to home because recycling capabilities in in certain places are not really there. Regarding security, of course, is not only the security, the use of the Internet, which means training and support for community members to for for safe practices and to avoid traps or things in in on the Internet, but also the the quality of the process in in in managing these devices and preusing them in different ways, like, quality checks, sanitization, inventory management, which allow a small scale initiative to to become larger and to work for for the size of the of their community. And, well, different aspects related to sustainability as well. They are discussed in document. Well, this is not different from maybe the previous presentation in the last IETF. Why why it's connected to the Gaia research? If you have any question, we can discuss. It also includes a bit of more in the in the newest draft. It op it goes more in detail into possible open research questions, things that deserve some extra thoughts. And and, well, the last point is well, this this sentence in italics is is kind of the last sentence in the document, which I think is a good conclusion. It's understanding how device ecosystems interact with connectivity infrastructure, community participation, and sustainability objectives might contribute to a more resilient and inclusive Internet connectivity models. But it's not the fault. The whole document is just a sentence. So my question to you is, what do you think needs to be included before the document, let's say, closes? By just rereading, I found some some some more things like ordering alphabetical ordering of of definitions and and a few typos. But other other than that from from our perspective, we find it is is quite mature. And then how do well, if if that works, let's say, do we move forward and adopt, which is the process of adoption. It's it's something good to know. So but if in the end, this is a kind of overview of of the latest iteration on the document, and and glad to share and discuss with you whatever you want to comment. Thank you. Oh, yeah. One one second is, I just included in the in the in the document, the acknowledgments to the to the to the funders and the supporters and and the people who contributed effort into into this, which, overall, it takes more than ten years of of different activities. This is example, for instance, of community center in Argentina in Rosario. This is the refurbishment lab in in Tau in also in Argentina. This is a program that the National University of La Plata has in which students that have no not enough resources, they they get the laptops. This is the few weeks ago, there was an a public event where two to two to 300 students got laptops from the from the program. These laptops were prepared in that warehouse where you can see they are very organized in sections and processes. And at the same time, in the university, they run a computer repair program and refurbishment program to train not just the university students and technicians, but also other actors in the community, other other NGOs that also support vulnerable population or even citizens who want to buy secondhand device instead of of a new one. And then with that payment, pay the workers in this social inclusion NGO and also provide resources for the initiative to be able to donate devices for students as well. So thank you.
[01:51:43] Curtis Heimerl: Alright. So join up in the queue if you have questions. I am first in the queue, which, you know, just as some context, GAIA is not an RFC machine. We've produced one RFC in the entirety of GAIA's operation, which is, I think, nearing ten years if not at that. I think it was, you know, RFC 7962 sounds right. Maybe before that, it was in grad school. Right? But, anyway, so we're really excited to kinda step back into that space, and the chairs have been talking about updating the old RFC. But I think the core question really is is is this the appropriate place for this? And, you know, you used to be a chair, and so this is
[01:52:23] Ben Roberts: a home
[01:52:23] Curtis Heimerl: community. But sustain is, I think, next door. It should it be at sustain given the focus? Or I think you've done a really good job contextualizing it within Gaia. Like, the importance of this space for connectivity and for getting technology out to these areas is really critical, but it has a big sustainability focus. Have you talked to them? What do you think about that? Why why is this the right home base?
[01:52:47] Leandro Navarro: Yeah. But, yeah, as as I said, if we produce one NF RFC, we I think we can produce a second one. And, also, you know, the the tools are are quite different from from previously. So for me, it was a learning exercise how to use the tools to produce a document that then it compiles and generates a text, XML, HTML, and PDF outputs of it. So so it seems it works. And regarding the the the binding to Gaia and to Sustain, of course, it has an implication for sustainability, but but I think the main focus is I mean, this is this is is there. For instance, when we use circular as a concept, you know, circularity is a kind of strange word because I mean, an apple is circular. You know? You have a tree. You eat it. There are seeds, and there's a new apple tree and comes back and sounds like a circular. But a a computer device, a router, is not circular by by any means. I mean, it's spiral at most, but never circular. I mean, you do not recover a router by putting into the ground, and you you get another router out of it. So, I mean, is this straight off that if you end up providing Internet access for everyone, how how do you deal with environmental implications of producing so many devices? I don't know. So it's there. But we also use the term circular in a sense of of use. In community networks, it's typical that if if the population is somewhere else, you might bring one access point, one link, and put it somewhere else to expand coverage in a new area. And also, if you do that, well, those guys need devices to access to use that connectivity. So we also look at that I mean, routers, switches have a very long lifetime, laptops, medium, mobile phones, very short, and but still they are reusable. So if the devices belong to the community, it makes sense that you can recirculate them to to other people. So I think the the main point of the document is documenting these practices in which communities are autonomous enough to procure, maintain, reallocate, and even in some in some cases recycled these devices to recover parts. And and and this this kind of circular management of of all the infrastructure is a requirement for these community networks to be, of course, sustainable in in in a mental environmental terms, but also sustain over time, be able to evolve, maintain the infrastructure. Because as I said before, when we do these collaboration projects, we find that typically, there is lack of money to deploy a network, but also there's lack of money and skills to to maintain it over the the long time and to deliver quality of service. We were discussing before that some people might say, well, use the cloud because it's it's it's stable. So if we can manage to have a networks which depend on local technicians, local decision makers, everything would be more like a sustaining in the long term. And I think this the the major the dimension of the work is is community networking or the Gaia, environment.
[01:56:34] Curtis Heimerl: Yeah. I I think that's correct. And it really I you know, there is overlap just with Gaia and sustain. And so we gotta put them on a list so people can go to both instead of having them running concurrently as we are at this ITF. And that's on me. So, maybe in the future, there's just discussions to have with them, but that's a good answer to me. We have Ben.
[01:56:57] Ben Roberts: Hi. Thank you. Ben Roberts. I'm from from Kenya, so representing the global South, I suppose. Good work you're doing. From you're asking questions, you know, what from a research point of view, what can really help this. In schools in Kenya, nearly every single device is a recycled device. It's generally, there have been companies operating for years that have been recycling desktop PCs coming out of call centers or something, laptops as well. So everything is generally recycled. Windows 10 came out in 2015 and is now in its sunset phase. So nearly most devices would be Windows 10, quite a lot of Windows eight. So they wouldn't find much any other operating systems. They'd be the main operating systems of devices. In terms of Linux devices, almost nothing. You know, but they're going to be the two predominant ones. Almost none would be Windows 11. So I agree with you about local technicians being able to repair and recycle and get boards. I see this. There's some friends of mine in Mount Kenya who do exactly this. They go around, collect them all up, and then they fix them up, then the broken ones, and then give some back to the schools, and then give the other ones to new schools to circulate them. But I think there's a lack of certainly lack of technical capacity around knowing then what open source OS you can use instead that will actually work and be supported. So if you say, okay, if you've got this much RAM in the CPU, what is the best version of Linux to use that is still going to carry on being supported for security fixes. So I think some research around what operating systems are out there in those recycled devices is absolutely key and can be done quite easily. And then also best practices and dissemination of those best practices, what OS to use, and how to do that. Right? If you don't have if you've got a computer and it's a brick, how do you how do you install a version? We don't have CDs anymore. So how do you so teaching people capacity building on how to get a working OS onto those to make them actually have a ongoing life since Windows and Microsoft have withdrawn from the conversation.
[01:59:21] Leandro Navarro: Yeah. Yeah. I mean, I subscribe for all your points. In fact, this slide, I think, reflects a lot of, the working ingredients of a of of an operational, infrastructure for, in this case, university students in in La Plata. Yeah. If you don't have all these ingredients at your network is fragile. It works as long as the funder or the initial volunteers or guys from outside decided to do. And, typically, people don't touch it because if the the white guys from the North are not there, if they do something wrong, they it would spoil things. So better better degraded than nothing. So, yeah, we we find all these critical so that whatever form of community networks for school, for a for a certain community, for a health center relies on on all these elements. You you see on the top left, you not you not only see piles of devices which are moved as they are processing in the place, but also you you find some things which are not which are not a computer. They are, like, special tools for data wiping for other processing of testing whatever components, which improve the quality of the the process. So these guys are very experimented, and I didn't discuss. But for instance, with all these initiatives, we develop we are developing open source tools in which we try to incorporate their processes into the tool so that they can do all this process in a more efficient manner. So when you move one device that has been already inspected to the other pile, we can record that into the into the inventory system of the of the factory, let's say, of the refurbishment factory. Without these people doesn't know which operating system would work in that a new desktop or laptop computer. We we try to install Linux devices in some schools, and, well, it was an interesting experience, let's say. In one case, it was funny. He said, well, do do that if you want and find out that the laptop, how slow it will be when it runs Windows. And and they tried, and then they said, okay. Maybe the Linux is is not that bad at the end. But but it's a lot of, knowledge that it's granted in in the IETF, but it's not in in in the community. So this is also work has to be done is meaningful access means not only knowing how to use the Internet, but also how to how to maintain the infrastructure. But, well, I mean, I encourage you to read the document because I would be able to talk too too long. But I think the point is made. Thank you.
[02:02:21] Curtis Heimerl: Alright. We are at time. Thank you all for coming to GAIA. See you all in San Francisco. I hope if you're interested in talking, please reach out to us. And, of course, this is active working group RFC activity. So if you're interested, you know, the mailing list is the place for that. Make sure to join that where the discussion will continue. So thank you all. See you around. Oh, I had to turn off.