[00:01.310 --> 00:02.850]  My name is Cora.
[00:03.170 --> 00:12.930]  I'm a second year electrical engineering PhD student at the City College of New York.
[00:13.130 --> 00:16.330]  I'm advised by Tashar, Joyce, and Rosario Gennaro.
[00:16.690 --> 00:32.200]  I would say that my research interests are in developing security and privacy solutions for people for at-risk populations.
[00:32.680 --> 00:36.040]  So, and vulnerable populations, so whether that be children or the elderly.
[00:36.720 --> 00:42.780]  And today I'm going to be talking about my ongoing research that has to do with mobility, movement, and communication.
[00:43.420 --> 00:44.380]  So, yeah.
[00:46.800 --> 00:59.000]  So, for the purposes of this talk, we can define an Internet shutdown as the deliberate interruption of Internet services that causes a partial or complete loss of connectivity on a network.
[01:00.500 --> 01:10.720]  And just to give you an idea of what this looks like on a global scale, these are a couple headlines just from 2024 alone of instances that have happened.
[01:10.980 --> 01:12.580]  And I'll give you guys a second to take a look.
[01:16.310 --> 01:23.830]  And what I think is important to notice here is that these are not one-off instances of Internet shutdowns that are happening.
[01:23.950 --> 01:26.630]  These are repetitive, they're targeted, and they are malicious.
[01:26.710 --> 01:32.250]  So, they're something that we need to keep an eye on and watch as things continue to develop.
[01:32.250 --> 01:38.830]  And, unfortunately, 2024 was the worst documented year for Internet shutdowns to ever be recorded.
[01:39.130 --> 01:44.530]  There were 296 instances of Internet shutdowns across 54 different countries.
[01:44.530 --> 01:51.110]  To help illustrate what that looks like, I took the Access Now dataset and made a fun little colorful map.
[01:51.290 --> 01:58.850]  So, all of the countries that are highlighted in red, those are the countries that experienced an Internet shutdown in 2024.
[01:58.850 --> 02:07.610]  But what is important to notice are the countries that are highlighted in purple because they are seven of the 54 that experienced an Internet shutdown.
[02:07.970 --> 02:11.830]  They imposed it for the first time within the calendar year, which is something to watch.
[02:12.050 --> 02:16.530]  So, as you saw on the graph previously, we're not seeing anything start to decline.
[02:16.750 --> 02:24.970]  They're still climbing rapidly, which is underscoring the importance of why we need to talk about Internet shutdowns on a global scale.
[02:25.870 --> 02:28.430]  So, let's talk a little bit about how they happen.
[02:29.690 --> 02:32.070]  There are a number of trigger events.
[02:32.250 --> 02:36.490]  These are catalyst happenings that posture a nation for an Internet shutdown.
[02:36.770 --> 02:45.990]  In 2024, the most common ones that were recorded were instances of social and political unrest, elections, exams, and government coups.
[02:46.750 --> 02:48.790]  And research supports this.
[02:48.990 --> 03:05.430]  There was a research document put out by Georgia Tech that found that Internet shutdowns are nine times more likely to co-occur with protests, 16 times more likely to co-occur with elections, and 286 times more likely to co-occur with government coups.
[03:05.430 --> 03:08.470]  So, we see the correlation here.
[03:08.610 --> 03:16.150]  And now that we have an idea of what causes them and how they get rolling, we can look at what the workflow generally looks like.
[03:16.150 --> 03:25.030]  So, typically, you have your government authority, and this can be national, regional, militant, whatever it might be, and they can issue a set of directives.
[03:25.070 --> 03:27.550]  You often see it done legally or extra legally.
[03:27.830 --> 03:32.710]  So, when I say legally, you often see this rooted in legislation.
[03:33.270 --> 03:39.270]  There's some sort of text that vests power into the authority themselves to be able to issue an Internet shutdown.
[03:39.990 --> 03:42.350]  And they can really do it for a variety of reasons.
[03:42.490 --> 03:44.430]  It can be in national interest.
[03:44.710 --> 03:46.130]  It could be for public safety.
[03:46.550 --> 03:49.330]  And you'll notice that the language there is very vague.
[03:49.510 --> 03:51.750]  That leaves a lot of space for these sorts of things to happen.
[03:52.470 --> 03:55.890]  And extra-legal methods, there are a number of them.
[03:56.250 --> 03:59.690]  They're often just cases of force.
[03:59.690 --> 04:12.610]  There have been recorded instances of subsea fiber optic cables being intentionally cut or power being shut off in certain areas where network infrastructure exists to restrict the use of the Internet.
[04:12.610 --> 04:19.250]  And that's a way to sort of get around it looking intentional because it looks sort of like a technical accident.
[04:19.250 --> 04:31.650]  And more often than not, the ISP or telecoms are either partially owned or completely owned by the government, which makes it very easy.
[04:31.790 --> 04:36.370]  It's very much of a tap on the shoulder to be able to get Internet shutdowns rolling.
[04:37.750 --> 04:47.990]  And that is a big problem because the ability to share information, organize, and communicate is vital during times of political and social unrest.
[04:48.250 --> 04:53.390]  And this is exactly why the Internet is targeted by governments in these instances.
[04:54.150 --> 05:07.030]  To give a couple examples, shutting down the Internet is a great way to control elections because you can restrict the flow of information, make it very difficult for a citizen to cast an informed vote.
[05:07.230 --> 05:12.930]  By returning the information that they're able to access is only what they want you to see.
[05:13.490 --> 05:25.090]  If there are some sorts of human rights violations going on and they're often very grave, shutting off the Internet is a way to prevent people from sharing that that's happening to the greater world and holding countries accountable.
[05:25.610 --> 05:28.590]  And it also prevents people from being able to protest.
[05:28.950 --> 05:42.270]  If the Internet is shut down, it's very difficult to organize, come together, plan where we're going to meet and how, which puts up all of these barriers, which creates a really large problem.
[05:43.230 --> 05:51.790]  And losing Internet access is a major hindrance and safety risk for protesters who are actively resisting oppressive regimes.
[05:52.310 --> 06:01.110]  So, for those affected, if someone was injured while they were protesting, it is now impossible to make a cell phone call to get EMS services.
[06:01.410 --> 06:04.470]  It's also very hard to keep in contact with protest organizers.
[06:04.810 --> 06:08.170]  There's no way to send a message and figure out where people are rallying together.
[06:08.170 --> 06:19.670]  And it also is very hard to contact your loved ones, whether that's your family or people that you're actively protesting with, to be able to know if people are safe and figure out where everybody is.
[06:21.050 --> 06:26.170]  So, we have this large issue of Internet shutdowns, particularly in the protest setting.
[06:26.170 --> 06:31.250]  So, how can we communicate during Internet shutdowns when this is happening?
[06:31.570 --> 06:33.690]  So, we'll enter ad hoc Mesh messaging.
[06:33.930 --> 06:37.290]  There's been a lot of talk of Mesh ad hoc this year, so I'm sure we're all familiar.
[06:37.750 --> 06:46.650]  But, to review at a very high level, what Mesh is, is a decentralized peer-to-peer communication scheme that uses short-range links.
[06:46.650 --> 06:51.870]  So, that can be Bluetooth Low Energy or Wi-Fi Direct to allow peer-to-peer communication between devices.
[06:52.930 --> 06:55.970]  We know that Mesh is proximity dependent.
[06:56.250 --> 07:01.310]  So, that means that the range of Wi-Fi Direct or Bluetooth directly impacts who you're able to connect with.
[07:01.550 --> 07:14.690]  And for people that are outside of that one-hop range, there is multi-hop, which allows messages to travel further and further, even if they're outside of that, like, physical range of whatever link that you're using.
[07:16.290 --> 07:21.630]  And Mesh is particularly good to use in a protest scenario for a couple different reasons.
[07:22.310 --> 07:24.910]  Like I mentioned earlier, Mesh is decentralized.
[07:25.350 --> 07:28.330]  And to understand why that matters, we can compare it with a centralized structure.
[07:28.790 --> 07:33.030]  In a centralized network, all functionality is funneled through one single point.
[07:33.230 --> 07:36.110]  So, in this little diagram, that would be the cell tower.
[07:36.110 --> 07:42.270]  Every message has to go from the person to the cell tower to the receiver, which creates this critical point of failure.
[07:42.490 --> 07:50.830]  And we know that these critical points of failure are targeted by governments, which is why it is such a big problem when it comes to communicating in these spaces.
[07:52.390 --> 07:59.390]  Decentralized completely skips the cell tower, and it can just go from peer-to-peer, which is why we want to use technology like this.
[07:59.530 --> 08:03.070]  It removes the bottleneck and can still function during Internet shutdowns.
[08:04.010 --> 08:07.470]  And Mesh also benefits from the network effect.
[08:07.990 --> 08:13.750]  The network effect being the more people that you have involved in your network, the better it's able to operate.
[08:14.030 --> 08:21.170]  So, when there's just a couple people that are on a Mesh network, your range is limited and it's spotty as people continue to move around.
[08:21.390 --> 08:28.510]  But as you have more and more people, the range gets further, the connection gets stronger, and it overall is a good thing.
[08:29.630 --> 08:32.510]  So, you can think about this sort of like a bucket brigade.
[08:32.930 --> 08:43.610]  So, you can have one person that has a pail of water, it's getting passed down a line of people, and the pail will go person to person to person until it arrives at wherever it's trying to go.
[08:43.950 --> 08:48.670]  And the more people that you have in your bucket brigade, the further and faster it's able to travel.
[08:48.870 --> 08:50.710]  That is how I remember it, at least.
[08:50.710 --> 08:52.810]  So, this is good.
[08:53.030 --> 08:58.670]  We know that Mesh works well in this context, but how is it performing in real life?
[08:59.670 --> 09:01.090]  I would call it complicated.
[09:01.790 --> 09:05.530]  So, we can take a look at FireChat first.
[09:05.730 --> 09:08.470]  I'm sure some people in the room are familiar with FireChat.
[09:09.030 --> 09:14.550]  It was extremely popular during a number of mass protests that happened worldwide.
[09:14.550 --> 09:21.030]  Particularly, we point at the Umbrella protests in Hong Kong in 2014, and the Elab protests in 2019.
[09:21.450 --> 09:27.090]  They saw a massive spike of installations during the instances of these two protests.
[09:27.410 --> 09:29.590]  It was more than they were able to keep up with.
[09:29.890 --> 09:34.510]  And it was, as the article says here, adored by protesters.
[09:34.810 --> 09:39.130]  But there were a lot of issues that made it an unsafe choice for protesters to use.
[09:39.130 --> 09:45.910]  And just four years later, it was removed from the Internet, and it's not supported any longer.
[09:46.190 --> 09:49.770]  So, it was plagued with operational and security problems.
[09:50.070 --> 09:54.490]  This is a quote that was from the University of Toronto Citizen Lab.
[09:54.650 --> 09:56.950]  They ran a security analysis on FireChat.
[09:57.250 --> 10:05.130]  They say that all messages sent and received, as well as the list of chat channels the user has joined, are stored unencrypted on the device.
[10:05.130 --> 10:11.390]  Which is a real problem when you're trying to protest in places where the government is malicious towards you.
[10:11.570 --> 10:16.830]  It also collected users' real names and their real email addresses while they were registering.
[10:17.130 --> 10:19.370]  And this was all stored locally on the device.
[10:20.130 --> 10:25.810]  And it got to be such a problem that FireChat actually eventually made a statement to the press.
[10:25.810 --> 10:32.750]  And they said, quote, people need to understand that this is not a tool to communicate anything that would put them in a harmful situation.
[10:32.950 --> 10:38.350]  If it were to be discovered by someone who's hostile, it was not meant for secure or private communications.
[10:38.830 --> 10:41.450]  So, I think that quote sort of explains itself there.
[10:42.410 --> 10:43.790]  Let's take a look at another tool.
[10:44.010 --> 10:44.670]  This is Bridgeify.
[10:45.250 --> 10:46.730]  Bridgeify still exists now.
[10:47.070 --> 10:48.270]  It's still available to download.
[10:49.390 --> 10:54.210]  It was... it's probably one of the most well-recognized Mesh apps, I would say.
[10:54.530 --> 10:59.990]  And it saw a lot of use during the eLab protests in Hong Kong.
[11:00.510 --> 11:04.130]  Again, similarly to before, it saw a massive spike of downloads.
[11:04.410 --> 11:07.310]  There were millions of open chats going on at the time.
[11:08.130 --> 11:14.550]  And today, it boasts around 11 million downloads between Apple and Google download stores.
[11:14.710 --> 11:15.910]  Which is something to keep in mind.
[11:16.190 --> 11:21.430]  Despite all of this, it has a lot of functionality problems.
[11:21.750 --> 11:23.470]  Quite lackluster, if you were to ask me.
[11:24.090 --> 11:26.090]  So, these are screenshots I took a couple of days ago.
[11:26.210 --> 11:27.110]  What it was looking like.
[11:27.890 --> 11:28.690]  Not great.
[11:29.930 --> 11:32.910]  And these are two reviews that I pulled from the Google Play Store.
[11:33.010 --> 11:35.250]  We can keep in mind that they're reviews, so take it with a grain of salt.
[11:35.250 --> 11:41.350]  But if you look at the number of reviews that they have that are negative, I think there's probably something to them.
[11:41.850 --> 11:50.230]  So, the review on the top essentially describes that a user was trying to set up Bridgeify to use the text with their mom.
[11:50.270 --> 11:53.150]  If there was a protest or something going on and they were to get separated.
[11:53.790 --> 11:58.070]  And they put the phone essentially in perfect conditions right next to each other.
[11:58.610 --> 12:01.470]  Enabled the connection and they weren't able to find anything.
[12:01.630 --> 12:03.710]  And this is just like sitting at the dining room table.
[12:03.710 --> 12:08.090]  And the user below actually tried to use it during an actual instance of a protest.
[12:08.510 --> 12:11.710]  And they were unable to get it to work at all.
[12:11.910 --> 12:15.310]  So, it doesn't bode well for the application.
[12:15.810 --> 12:17.590]  And unfortunately, there is more.
[12:18.910 --> 12:22.890]  In 2021, there was a team of researchers that put out this paper.
[12:23.510 --> 12:26.830]  Mesh messaging and large-scale protest breaking Bridgeify.
[12:27.150 --> 12:33.030]  And they were questioning whether or not Bridgeify was a good tool to use in the protest setting.
[12:33.910 --> 12:38.510]  Because they saw issues in the way that it was conceived.
[12:38.710 --> 12:40.810]  Because it was not designed to be used that way.
[12:40.810 --> 12:55.450]  And the researchers essentially said that Bridgeify currently permits its users to be tracked, offers no authenticity, no effective confidentiality protection, and it lacks resilience against maliciously crafted messages.
[12:55.450 --> 13:01.810]  So, they put out this report, and it was pretty ground-shaking for the people that were currently using the application.
[13:02.290 --> 13:11.950]  The researchers found that you could produce entire social graphs to show who I messaged on a daily basis, like who I'm connected to, and so on and so forth.
[13:11.950 --> 13:22.590]  You could read their messages entirely, you could impersonate anyone, and even caused shutdowns of the Bridgeify network with a single maliciously crafted message.
[13:23.850 --> 13:25.950]  So, Bridgeify acknowledges this.
[13:26.110 --> 13:30.590]  And they say, okay, thank you for letting us know.
[13:30.590 --> 13:35.410]  We're going to implement the signal protocol to try and alleviate these issues.
[13:35.970 --> 13:38.130]  So, that's what they said that they did.
[13:38.130 --> 13:41.810]  And then, shortly after, they said that the issue was patched.
[13:41.910 --> 13:48.110]  They went on to continue to push the application out for people that were experiencing Internet shutdowns.
[13:48.110 --> 13:51.950]  This is in 2022, during the Russian invasion of Ukraine.
[13:52.370 --> 14:00.290]  And it's not... I wouldn't say it's like a super bold advertisement, but if you look at the language, it's definitely suggesting that it's there to use if you follow certain steps.
[14:00.590 --> 14:03.870]  And these were both pulled from the Bridgeify Twitter.
[14:03.870 --> 14:10.530]  The picture on the bottom is something that they posted just showing the sheer amount of downloads that they were getting while this was happening.
[14:11.710 --> 14:15.110]  So, who thinks that everything was okay?
[14:17.530 --> 14:18.130]  Maybe.
[14:18.930 --> 14:19.230]  Ugh.
[14:19.930 --> 14:21.650]  So, another paper came out.
[14:21.870 --> 14:23.470]  And it is called Breaking Bridgeify.
[14:23.750 --> 14:25.890]  Again, it was the same exact team of researchers.
[14:26.230 --> 14:31.690]  And they found that the signal protocol was implemented completely incorrectly when it came to this.
[14:31.810 --> 14:33.990]  And there were still some glaring security issues.
[14:35.270 --> 14:38.630]  It was essentially a vicious man in the middle attack.
[14:38.910 --> 14:41.930]  This is a screenshot that researchers actually posted a video.
[14:42.130 --> 14:43.870]  You can find it on Twitter if you want to watch it.
[14:44.030 --> 14:46.650]  It's about a minute and 30 seconds long in real time.
[14:46.690 --> 14:51.070]  So, it's pretty horrific to watch how insanely dangerous this is.
[14:51.070 --> 14:54.890]  But essentially, you have your two users, Alice and Bob.
[14:55.030 --> 14:56.470]  They're communicating using Bridgeify.
[14:56.730 --> 14:59.570]  And there is a malicious third-party involved.
[14:59.890 --> 15:04.810]  And by the end executing the code that they had written, they were able...
[15:04.810 --> 15:10.110]  The third-party was able to read the encrypted messages entirely.
[15:10.790 --> 15:13.110]  And that is...
[15:13.110 --> 15:14.890]  That was a major problem.
[15:15.070 --> 15:15.350]  Yikes.
[15:15.350 --> 15:19.750]  So, where Bridgeify stands today, I don't see...
[15:19.750 --> 15:21.930]  As I reviewed the research, I don't see a lot of consensus.
[15:23.470 --> 15:26.510]  I see the work sort of drop off around like 2022, 2023.
[15:26.950 --> 15:34.030]  I would still be very hesitant to use the application if you were looking for some sort of mesh tool to be communicating with.
[15:34.870 --> 15:40.130]  And there's a lot of reasons to be cautious because the consequences are a grave.
[15:40.390 --> 15:42.550]  The New York Times recently put out this article.
[15:42.550 --> 15:43.670]  I don't know if you all saw it.
[15:43.850 --> 15:48.830]  But essentially, they interviewed people that actively protest during the Hong Kong ELA protests.
[15:49.090 --> 15:54.010]  And they were interviewing people that had been imprisoned for taking part in the protest.
[15:54.290 --> 16:05.550]  And they found that the outside of like the mental and physical damage that they received, it's the societal damage that's really just...
[16:06.250 --> 16:08.050]  It's hard to get your head around.
[16:08.050 --> 16:17.730]  There was one activist who told her story and said that she was arrested and imprisoned for participating in a largely peaceful demonstration.
[16:18.170 --> 16:25.870]  And she was stripped of her social work license and she's still unable to get her license back to this day and can't find work.
[16:25.870 --> 16:33.090]  So it underscores why it's so important for us to trust the tools that we are using when the stakes are so very high.
[16:33.310 --> 16:38.250]  I definitely recommend reading the article because there's more instances that they talk about in it.
[16:38.530 --> 16:41.330]  But we have to ask ourselves a question.
[16:41.750 --> 16:50.650]  What do FireChat and Bridgeify have in common besides the obvious of the security problems and the not working super well?
[16:57.030 --> 16:59.390]  Yes, but no for this one.
[17:00.190 --> 17:00.830]  Sorry?
[17:03.610 --> 17:06.170]  I am going to point to something different.
[17:06.470 --> 17:14.790]  Is that both Bridgeify and FireChat were not... were used as protest tools, but they were not designed to be protest tools.
[17:15.390 --> 17:18.610]  And you might ask yourself, OK, they weren't designed for this.
[17:18.770 --> 17:21.870]  What was their intended use case and how did it end up here?
[17:22.650 --> 17:24.810]  Surprisingly enough, it was music festivals.
[17:25.070 --> 17:34.950]  It was intended to be used in places like Electric Forest and Burning Man or very crowded areas with mass crowds like a very packed stadium where a network could be overloaded.
[17:34.950 --> 17:39.830]  And even some instances of people trying to use them on planes when you're out of the reach of a cellular tower.
[17:41.290 --> 17:45.110]  So, you might say, OK, that's still a crowd.
[17:45.190 --> 17:46.410]  Why is it different?
[17:47.270 --> 17:51.150]  Crowds are not all the same and they're actually pretty different from each other.
[17:51.570 --> 17:55.130]  Although it's technically groups of people, they move differently.
[17:55.130 --> 17:56.330]  They behave differently.
[17:56.550 --> 18:00.690]  And that's something that we have to take in mind when we're developing mesh tools.
[18:01.270 --> 18:12.790]  This brings us to this major disconnect that we have in the research community is that we have mesh tools that are being deployed into venues that they're not built for or tested against.
[18:13.070 --> 18:15.350]  So, we could talk a little bit more about what that looks like.
[18:16.150 --> 18:22.570]  I would say that this is very large, like largely from the academic space, but mesh tools are often evaluated, used in simulation.
[18:23.670 --> 18:26.770]  NS3 is usually the one that people point to to do this.
[18:27.390 --> 18:31.970]  And essentially how it works is you give NS3 a mobility model.
[18:32.210 --> 18:37.150]  A mobility model tells you how the nodes in your network that you're testing are supposed to move.
[18:37.430 --> 18:42.930]  And this can be directional, this can be velocity, the time that it takes to make those sorts of decisions.
[18:43.310 --> 18:49.190]  It's simulated upon and then you get an outcome to tell us, OK, this is how well this mesh scheme worked.
[18:50.010 --> 18:56.250]  And the problem with that is because we don't have a good understanding of how people physically move.
[18:56.390 --> 19:05.770]  And this is important in the mesh context because in a mesh network, as people move around and change their position, you're constantly restructuring the way that the network works.
[19:05.930 --> 19:09.110]  It's like if you picked up your routers and just started playing musical chairs with them.
[19:09.290 --> 19:10.950]  It becomes very complicated.
[19:10.950 --> 19:25.570]  So when we don't have an understanding of it and we substitute it for something that's random, you're going to end up with this like harmful cycle where we're developing tools like perhaps they're OK for protest scenarios and then they're deployed and they don't work well.
[19:25.750 --> 19:33.430]  And we need to find a way to accurately evaluate and build tools that have the protesters in mind from beginning to end.
[19:34.870 --> 19:36.750]  So we can talk about how the simulation works.
[19:36.870 --> 19:39.410]  I know I went through that fast at a higher level.
[19:39.630 --> 19:41.390]  So you have your mobility model.
[19:41.610 --> 19:45.130]  This is how the nodes are supposed to move around.
[19:45.270 --> 19:46.030]  This is like their speed.
[19:46.250 --> 19:47.310]  This is where they're angled.
[19:47.450 --> 19:48.370]  This is how they make decisions.
[19:48.830 --> 19:50.230]  And then you have your traffic model.
[19:50.410 --> 19:55.350]  So that is how frequently are the nodes that are in your simulation sending messages.
[19:56.150 --> 19:57.550]  How large are the messages?
[19:57.710 --> 19:59.230]  That also gets fed in.
[19:59.530 --> 20:02.510]  And then you have your mesh protocol, whatever it might be that you're testing.
[20:02.510 --> 20:08.710]  All gets fed into your simulator and then it spits out some performance metrics and lets you know how the scheme did.
[20:09.490 --> 20:11.070]  So let's think about that for a second.
[20:12.170 --> 20:20.610]  Let's say you were taught to drive in an empty parking lot, no cars, and you got your license for driving around in this empty parking lot.
[20:20.750 --> 20:25.190]  And then one day you're told to go drive in Midtown and you get in a car accident.
[20:25.190 --> 20:27.330]  Are you a bad driver?
[20:27.690 --> 20:34.470]  Or were you evaluated and taught how to drive in a place that's not conducive to the rest of the environments that you would drive?
[20:35.090 --> 20:43.850]  And I would say that this is the core of the issue here, is that we're evaluating our mesh tools on mobility models that don't match where they're meant to be deployed at all.
[20:43.850 --> 20:47.810]  And you're seeing this massive disconnect in the applications that are being created.
[20:48.210 --> 20:52.030]  Let's take a look at what this type of mobility model looks like.
[20:52.190 --> 20:57.830]  This is the, I'd say like the de facto model that is reached for all the time in academia.
[20:57.830 --> 20:59.510]  It's called the random waypoint model.
[20:59.510 --> 21:08.550]  You're often going to see the base version of this, or it's going to be some sort of derivative that might have a couple other options to determine the way that the nodes move.
[21:08.690 --> 21:11.290]  But generally, at a high level, it works like this.
[21:11.750 --> 21:14.170]  So you have your 2D grid space.
[21:14.370 --> 21:20.550]  And that's where these simulations take place in a blank grid, which is already an issue, but we'll talk about that later.
[21:21.230 --> 21:23.870]  And you have your node that will start in an original position.
[21:23.870 --> 21:29.550]  And it will be in this space, and then it will take a moment to think and decide where it wants to move.
[21:30.010 --> 21:35.230]  Once it's decided a target location, it's going to decide how fast it wants to get there.
[21:35.330 --> 21:37.150]  And then it's going to execute that movement.
[21:37.410 --> 21:42.730]  And then once it arrives in the next block, it's just going to repeat that over and over and over again.
[21:43.090 --> 21:43.890]  And that is it.
[21:44.030 --> 21:47.770]  That, at a very high level, is how the random waypoint model works.
[21:48.290 --> 21:51.690]  And you end up with something that looks a little bit like this.
[21:51.690 --> 21:57.090]  You have a very, like, zigzagged, jumping-around motion.
[21:57.370 --> 22:02.650]  And I think just looking at it like this, we know that this isn't how people move in real life.
[22:03.270 --> 22:04.970]  It's making a lot of assumptions here.
[22:05.150 --> 22:09.630]  It's assuming that people move randomly at all, which is something that we'll discuss.
[22:10.010 --> 22:14.310]  But it also assumes that we have this machine-like consistency when we move around.
[22:14.310 --> 22:22.370]  So, for the entire day, if I decided to move from place to place to place, I said that I was going to move at X miles per hour.
[22:22.510 --> 22:24.290]  I would have to do that every single time.
[22:24.490 --> 22:25.910]  We know that's not how people move.
[22:26.010 --> 22:27.050]  Say you're carrying something.
[22:27.230 --> 22:28.750]  Say you're walking in a difficult terrain.
[22:29.110 --> 22:30.390]  Your speed is going to change.
[22:30.470 --> 22:31.850]  And that's something that needs to be reflected.
[22:31.850 --> 22:37.810]  And going back to this 2D grid situation, the world doesn't look like a 2D grid.
[22:38.030 --> 22:39.530]  There's obstacles in our way.
[22:39.630 --> 22:40.250]  There's buildings.
[22:40.450 --> 22:45.610]  There's pieces of architectural influence that we have to take in mind when we're trying to simulate.
[22:46.070 --> 22:54.390]  Particularly when we're trying to simulate a protest that might be taking place in an urban setting, where you have big buildings and streets and cars that are around.
[22:54.930 --> 22:58.810]  And finally, it assumes that we move with no memory.
[22:59.050 --> 23:02.430]  So, in the random waypoint model, it doesn't matter where the node has been.
[23:02.650 --> 23:06.490]  It doesn't take that into account when it comes to deciding where it's going to go.
[23:06.630 --> 23:09.890]  And we know that that's not how we move in real life.
[23:10.770 --> 23:15.050]  Human movement cannot be represented by any form of randomness at all.
[23:15.150 --> 23:18.170]  Because it has so many different outside influences.
[23:18.450 --> 23:19.870]  We're going to talk about a few of them.
[23:19.870 --> 23:26.470]  But I would point to crowd science, relationships, and architectural features as being some of the major points for this.
[23:27.390 --> 23:28.950]  So, let's talk about crowd science.
[23:29.190 --> 23:33.370]  This is something that I think we need to talk about more when it comes to mesh.
[23:33.410 --> 23:42.110]  Because I think it brings a lot of the solutions that we're looking for and starts to point us in the right direction when it comes to developing models that work really well.
[23:42.110 --> 23:50.070]  But crowd science is... it is a field of study that seeks to understand and manage the behavior of large human groups.
[23:50.230 --> 23:57.530]  And what you see from a lot of the literature is that large groups have patterns in the way that they move.
[23:57.550 --> 24:00.010]  And it's heavily influenced by certain factors.
[24:00.290 --> 24:03.070]  This is an article that was published in 2022.
[24:03.070 --> 24:08.650]  It's very interesting, especially for our context when it comes to, like, mass dense crowds.
[24:09.210 --> 24:18.090]  Essentially, what the researchers found was that the way that one person moves in a dense crowd is directly impacted by their optical neighbors.
[24:18.390 --> 24:28.710]  And what they're saying is that if I'm walking in a big crowd of people, whoever's standing directly in front of me, they're going to drive the way that I move.
[24:28.710 --> 24:31.730]  And they found that people will match speed and direction.
[24:31.890 --> 24:38.090]  And this is a completely unconscious thing that people do, but it happens when you're in these very large crowds.
[24:38.470 --> 24:39.950]  Which is something that we have to keep in mind.
[24:40.130 --> 24:44.130]  Because go back and take a moment to think about that random waypoint model.
[24:44.290 --> 24:47.670]  They're bouncing around with no direction at all.
[24:47.850 --> 24:54.750]  If you're trying to simulate that as a crowd on a large scale, you're going directly against what crowd science has found to be true.
[24:55.930 --> 25:01.370]  It tells us some more pretty interesting things that we should take in mind when we're developing these mobility models.
[25:01.830 --> 25:04.750]  It defines that there's two different types of crowds.
[25:04.970 --> 25:07.990]  And they actually have different movement signatures.
[25:07.990 --> 25:10.010]  And they behave very differently from each other.
[25:10.390 --> 25:12.970]  So, on the left, we have a physical crowd.
[25:13.110 --> 25:18.710]  And a physical crowd can just be defined as a mass of people that are co-occupying a space.
[25:18.710 --> 25:21.050]  But they're all there for their own independent reason.
[25:21.730 --> 25:24.930]  So, this woman might be headed home from work.
[25:25.090 --> 25:27.410]  And this guy might be going to a doctor's appointment.
[25:27.770 --> 25:29.830]  But they're both in the same space.
[25:30.010 --> 25:31.090]  And that forms the crowd.
[25:31.350 --> 25:34.330]  But there is also a psychological crowd that exists.
[25:34.610 --> 25:38.210]  And this is when you have a large mass of people co-occupying a space.
[25:38.690 --> 25:41.230]  But they share a sense of self.
[25:41.530 --> 25:43.210]  And they've gathered for a shared reason.
[25:43.210 --> 25:56.070]  And researchers have found that when you have a psychological crowd, they actually tend to move slower and much closer together as they travel as a big group.
[25:56.270 --> 26:02.010]  And there is increased social interaction that happens between the different people that are in the crowd.
[26:02.010 --> 26:03.730]  And that's something to keep in mind again.
[26:03.730 --> 26:14.450]  Because, go back to the mobility models, we don't have a way to represent that when it comes to people having relationships, moving closer together, and being influenced by the nodes around them.
[26:15.930 --> 26:21.530]  This is a clip pulled from a newsreel of a civil protest that happened in Myanmar.
[26:21.930 --> 26:23.570]  And I want to point out a couple of things.
[26:23.850 --> 26:25.950]  I'll give you guys a minute just to look at it as it is.
[26:26.010 --> 26:29.290]  It's been blurred just for privacy reasons to protect those that were involved.
[26:29.290 --> 26:30.890]  It's a little blurry up here.
[26:31.210 --> 26:34.910]  But I'll give you guys a second just to watch and kind of notice some patterns.
[26:38.370 --> 26:45.950]  Does anyone want to point something out that they notice about it that might go against this random waypoint model that we've been talking about?
[26:46.830 --> 26:48.770]  They're all going in a straight line?
[26:49.190 --> 26:49.510]  Yeah.
[26:49.750 --> 26:52.110]  There's this, like, channeling pattern that we see.
[26:52.570 --> 26:53.530]  Anybody else?
[26:54.330 --> 26:59.190]  It seems like the physical kind of thing, the psychological kind of thing, yeah.
[26:59.190 --> 27:00.050]  Yeah.
[27:00.330 --> 27:01.570]  You see, like, this...
[27:01.570 --> 27:05.030]  There's different behaviors, although it's like one large group of people.
[27:05.170 --> 27:08.830]  People are motivated by different things, although they're in the same space.
[27:09.150 --> 27:09.990]  One more person.
[27:10.250 --> 27:13.270]  Something that you notice about this that might go against what we talked about.
[27:13.910 --> 27:17.470]  People that are going in the same direction are closer to each other.
[27:17.750 --> 27:18.070]  Yeah.
[27:18.430 --> 27:22.050]  People that are going in the same direction are closer to each other.
[27:23.010 --> 27:25.490]  So, please forgive me for my drawings.
[27:25.710 --> 27:26.530]  I couldn't find my mouse.
[27:26.530 --> 27:27.910]  But here's what we have.
[27:28.090 --> 27:31.950]  I've kind of highlighted for you what the different patterns look like.
[27:32.090 --> 27:38.690]  And when we talked about architectural features having an influence on the way that people move, you can see that really clearly here.
[27:38.690 --> 27:51.030]  If you take a look along the sides of the buildings, I would look specifically here, under the awnings and sort of around the sides of the buildings, you see, like, a high density of people.
[27:51.270 --> 28:02.270]  Even though there's definitely a lot of free space for people to move around and give themselves more room, according to CrowdScience, we actually, as people, when we're in big crowds, we don't like to diffuse ourselves.
[28:02.270 --> 28:06.910]  We like to stay close together and we don't make use of the space the way that you would think that we do.
[28:06.910 --> 28:14.170]  And these channels of people that you see traveling, you said it perfectly, they want to stay very close together.
[28:14.170 --> 28:18.170]  And we can point that back to the psychological reasons that we've highlighted earlier.
[28:18.170 --> 28:26.290]  And something else is that we have these, this is a still now, but we have groups of people that travel together.
[28:26.710 --> 28:33.030]  These bonded nodes, I guess you could call them, people that might have arrived to the protest together.
[28:33.230 --> 28:34.630]  Maybe that's your significant other.
[28:34.790 --> 28:35.850]  It's someone you work with.
[28:35.970 --> 28:36.770]  It's a sibling.
[28:37.070 --> 28:41.050]  You're not going to tolerate a lot of distance being put between you and that other person.
[28:41.050 --> 28:44.510]  And you see people traveling together in pairs.
[28:44.510 --> 28:48.730]  You see this much more often than you see someone wandering around on their own.
[28:50.170 --> 28:53.170]  So let's put these back next to each other again.
[28:53.630 --> 28:58.230]  And after we've talked about it, the differences are startling.
[28:58.810 --> 29:07.830]  We're trying to develop tools to be deployed in this environment using this understanding of how people move and behave.
[29:07.830 --> 29:15.550]  And this, I would argue, is why you're seeing a lot of failure when it comes to mesh tools specifically being built in the protest context.
[29:15.750 --> 29:21.890]  Because we're trying to design them on a mobility model that does not accurately represent how people move at all.
[29:21.890 --> 29:25.750]  So, we know a few things now.
[29:25.950 --> 29:29.710]  That human crowds have traceable mobility patterns.
[29:29.930 --> 29:37.690]  And that the media, whether that's videos, pictures, whatever it might be, from protests, don't resemble the mobility models that we have.
[29:38.010 --> 29:42.390]  And that our mesh applications that are meant for use in protests don't work well.
[29:42.550 --> 29:43.910]  So we know these three things.
[29:44.230 --> 29:47.550]  And you're probably asking yourself, what do we do about it?
[29:47.550 --> 29:57.730]  And that is the core of what my research is looking at, is to solidify our understanding of how people move and protest to be able to build strong mobility models.
[29:57.730 --> 30:03.930]  And this is sort of the flow of what this can look like in the future as the project develops.
[30:04.250 --> 30:09.890]  So, we're going to develop better mobility models by conducting interviews and watching footage.
[30:10.150 --> 30:18.150]  And once we've done this, we can actually do very representative evaluations of how new mesh schemes are working.
[30:18.150 --> 30:28.010]  And once we find one that works, there's room to build a better mesh application that's meant for protesters, that's kept them in mind from start to finish of the development.
[30:28.390 --> 30:36.030]  And we're here right now, is we need to understand the way that people move to build these strong mobility models.
[30:36.590 --> 30:39.990]  So we can talk a little bit about what that approach looks like.
[30:40.770 --> 30:44.410]  My research questions, these have been cut down for brevity because they're kind of rude.
[30:44.690 --> 30:53.430]  But, essentially, what we're looking to understand is how do various people in the protest setting communicate when there's an Internet shutdown that's going on?
[30:53.830 --> 31:01.810]  Secondly, we want to understand how these different actors are physically navigating their environment during protests when there is no Internet access.
[31:02.410 --> 31:08.810]  And finally, we want to know what kind of features that they would want in a secure communication system that can be used during an Internet shutdown.
[31:08.810 --> 31:15.070]  And we're hoping to glean this from our interviews that we have that I'll tell you a little bit more about.
[31:15.310 --> 31:18.270]  But before we got to that, we did a lot of groundwork.
[31:18.390 --> 31:28.830]  And this was a massive review of publicly available media to sort of contextualize the study and give us an idea of what the movement patterns look like from afar.
[31:29.170 --> 31:35.870]  And by watching them, we were able to identify patterns and develop questions, sort of like what y'all saw in that clip before.
[31:35.870 --> 31:42.510]  So to show them a little clearer, we see these like bonded clusters of people.
[31:42.650 --> 31:44.610]  This is those like those relationship nodes.
[31:44.850 --> 31:46.370]  This is you with your friend.
[31:46.510 --> 31:48.930]  This is you with your sibling traveling around the protest.
[31:48.930 --> 31:51.750]  You're not going to tolerate a lot of distance being put together.
[31:51.950 --> 31:57.050]  So one thing that I want to make sure that we get in this new mobility model is a way to represent this.
[31:57.210 --> 32:03.910]  A way to represent nodes that travel together will only tolerate up to X distance between each other.
[32:04.090 --> 32:07.690]  And if they are to be separated, they come back together at some point.
[32:09.090 --> 32:11.690]  Another thing that we saw was the channeling.
[32:12.490 --> 32:18.870]  So even when there is room for people to diffuse and make space, they're much more likely to just stay closer together.
[32:18.870 --> 32:22.210]  And we can point back to the crowd science to explain why that happens.
[32:22.210 --> 32:24.630]  But we need a way to represent this as well.
[32:24.790 --> 32:35.490]  Because the idea that we have that these big crowds of people are sort of like these like mindless, have no direction in the way that they move because there's so many people.
[32:35.690 --> 32:37.730]  It's just not accurate to how people move today.
[32:38.150 --> 32:42.010]  Crowd science tells us that they're self-organizing and we need a way to represent that.
[32:42.010 --> 32:49.550]  Another feature that we definitely need to have in our new mobility model is going to be a scatter.
[32:50.490 --> 32:58.670]  Particularly when you're looking at the protest context, you have hostile actors, whether that's law enforcement, whatever it might be.
[32:58.910 --> 33:04.830]  We don't have a way to represent them coming on to our 2D grid space to interact with the people.
[33:04.830 --> 33:15.030]  And what we noticed through a lot of review is when you have a hostile actor and they come in contact with a group of people, you actually see people clump together and scatter apart.
[33:15.290 --> 33:21.530]  Which even if it doesn't make the most sense in terms of their safety, think back to those optical neighbors.
[33:21.790 --> 33:26.690]  When there's a lot of things going on at once, we're just much more likely to follow the person that's in front of us.
[33:26.830 --> 33:32.090]  So even if it's not the best escape route that's available, we're traveling in groups.
[33:32.090 --> 33:36.250]  And this is something that we need to consider when we're talking about mesh.
[33:36.630 --> 33:43.390]  Again, mesh is constantly restructuring itself when people move around whatever environment that they're in.
[33:43.490 --> 33:51.790]  And if we go from this, like a tight group of people, to this, we have like little cells of people breaking off in different directions.
[33:51.930 --> 33:53.610]  We need to be able to account for that.
[33:53.710 --> 33:59.450]  When it comes to people using mesh networks, we want to try and prevent people from being unable to message with each other.
[33:59.450 --> 34:03.310]  So this motion is something that we want to be able to simulate.
[34:04.750 --> 34:07.670]  So, we did our media review.
[34:08.010 --> 34:12.090]  And the next phase in this project is going to be an interview study.
[34:12.270 --> 34:15.470]  And we're looking to talk to three specific kinds of people.
[34:15.710 --> 34:17.610]  We're looking for civilian demonstrators.
[34:17.610 --> 34:19.510]  We are looking for protest organizers.
[34:19.510 --> 34:21.070]  And we're looking for journalists.
[34:21.470 --> 34:23.670]  And we're doing this for a particular reason.
[34:24.330 --> 34:34.370]  When you think about all of the different types of people that might be present at a protest, there's going to be subtle differences in the way that they behave and the way that they move.
[34:34.550 --> 34:41.510]  And we want to be able to draw on all of that, to paint ourselves a really vivid picture of what movement might look like.
[34:41.510 --> 34:48.910]  So, for protest organizers, what we noticed in the review is that there tends to be this scouting behavior.
[34:49.150 --> 35:03.770]  So, if someone is in charge of making sure that everyone stays safe and the protest continues to move in the right direction, they're probably going to run out ahead or hang far back and give people an idea of, hey, there's police that are present at this street and this street,
[35:03.890 --> 35:04.550]  so stay aware.
[35:04.550 --> 35:08.910]  So, you're going to have, like, this wider lens view of what's going on.
[35:09.090 --> 35:12.710]  You often have multiple in the same demonstration that are leading it forward.
[35:12.750 --> 35:14.410]  That gives us different perspectives.
[35:14.810 --> 35:22.590]  And they have an understanding of what the intended motion is supposed to be, which is something that we'd want to draw on when we're trying to understand mobility.
[35:23.550 --> 35:28.330]  For civilian demonstrators, they are in the center of the action.
[35:28.330 --> 35:31.210]  You can think about, like, that very dense crowd of people.
[35:31.210 --> 35:36.170]  And the way that they'll be... they've experienced the protest is through their personal perspective.
[35:36.570 --> 35:48.790]  And they might be missing a little bit of context as to what is happening, like, way out ahead of them or way from behind them because they're just seeing what's in front of them because of how dense those crowds might be.
[35:48.970 --> 36:00.470]  But we want to understand for people that are in the center of what's going on, how was your movement influenced and were you using any sorts of devices to try and communicate with people when the Internet went out?
[36:00.630 --> 36:02.290]  What did you do to continue to communicate?
[36:02.670 --> 36:09.570]  So they have a very, like, an interesting way of evaluating what's going on as a whole.
[36:09.990 --> 36:12.530]  And finally, we want to talk to journalists.
[36:13.190 --> 36:22.070]  One thing to keep in mind is that I would say journalists are often conflated as being a part of the protest when really they're there to report.
[36:22.370 --> 36:29.970]  So you'll see journalists sort of hanging out on the edge of crowds because they want to get an idea of what's going on.
[36:30.210 --> 36:36.930]  So we can't just combine journalists in with the rest of the people that are there demonstrating for the cause because they're there to report.
[36:37.110 --> 36:40.170]  And you're going to see a difference in the way that they move because of that.
[36:40.170 --> 36:44.230]  So I would call them semi-independently motivated to move.
[36:44.450 --> 36:48.570]  And that's going to give us a rotating perspective that's going to be important for this.
[36:48.790 --> 36:50.110]  So let's do a little diagram.
[36:50.890 --> 36:55.330]  Let's say we have a northbound protest and we have our civilian demonstrators.
[36:55.910 --> 36:57.530]  They're going to be the most tightly packed.
[36:57.710 --> 37:01.050]  They're going to be right together, right in the center of the street.
[37:01.410 --> 37:10.150]  And then we have our protest organizers that are going to hang out way in the front, way in the back, giving us an idea of what's going on to protect the safety of the people that are involved.
[37:10.770 --> 37:18.470]  And we have our journalists that are hanging out on the edge of all of the action to try and get an idea of what's going on to be able to report on that.
[37:18.790 --> 37:32.330]  And when you think about those three different positions, the three ways that they're going to move, the ways that they're motivated to move, you have this very interesting perspective where you have journalists sort of looking in on everything going on.
[37:32.330 --> 37:35.390]  The civilian demonstrators are seeing what's ahead of them.
[37:35.490 --> 37:37.990]  And the protest organizers can see beyond that.
[37:38.210 --> 37:47.970]  And by drawing on those three very unique perspectives, we can get a really strong idea of how people are moving, even though we're not physically present there to be able to see ourselves.
[37:47.970 --> 37:54.650]  And that brings me into the ethics of the project, which is definitely something that's important to discuss here.
[37:56.050 --> 38:02.530]  Protesters are an at-risk population, and that requires the utmost care that we can provide when working with them.
[38:02.670 --> 38:13.610]  So things that we're keeping in mind as we're developing this interview study is that any of the footage and media that we reviewed in the groundwork phase, that's not going to be released at any time.
[38:13.610 --> 38:16.730]  That was just for us to get an idea of what's going on.
[38:16.730 --> 38:30.650]  And when we get to the interview portion, when we have people giving their testimonies about how they moved and why they moved a certain way, we're going to make it so you can redact that at any time, even after you've said everything and you decide, actually,
[38:30.650 --> 38:32.190]  I don't know if I'm comfortable with that.
[38:32.570 --> 38:34.730]  Take it back to make sure that people stay comfortable.
[38:35.310 --> 38:41.810]  And we want the goal of this to improve our understanding of protester movement.
[38:41.930 --> 38:43.970]  We don't want it to be something that's predictive.
[38:43.970 --> 38:46.750]  And this is something that we're keeping in mind as we're developing.
[38:47.050 --> 38:58.150]  And sort of the train of thought that we had for this was that, again, we know that the current mesh tools that we have don't work well enough to be widely trusted or used.
[38:58.150 --> 39:09.250]  And we can't release some sort of application to the protesters that are involved in certain demonstrations because it puts them at major risk, both legally and bodily.
[39:09.430 --> 39:10.990]  We don't want to put anyone in danger.
[39:11.290 --> 39:21.990]  And I would say it's highly unlikely for us to be able to contact an ISP or a telecom and ask them for communication data to get an idea and do traces.
[39:21.990 --> 39:32.810]  So that brings us to our next best option, which is observation and interview, to get a strong idea of how people are moving to be able to address this problem.
[39:34.410 --> 39:39.550]  Considering all of that, I'm very optimistic about the direction of the study.
[39:39.550 --> 39:49.730]  I think that this project is the first stepping stone in a series of stones that we can use to develop tools that can more directly serve the needs of civil society.
[39:50.190 --> 39:57.930]  And on that note, I recognize that I'm in a room full of experts and people here have all different kinds of knowledge and expertise.
[39:57.930 --> 40:07.130]  And I would say that this project is an interdisciplinary one and we need to pull from multiple places to get an idea of how to attack and solve this problem.
[40:07.130 --> 40:08.690]  So we want your input.
[40:09.050 --> 40:10.850]  I think we have a little bit of time left.
[40:10.950 --> 40:16.910]  So I would love if people have like questions, comments or suggestions for how we should go about this project.
[40:17.090 --> 40:20.570]  We really want to hear them and draw on all of your knowledge.
[40:21.230 --> 40:24.550]  And with that, thank you very much.
[40:29.590 --> 40:31.850]  Do you have one question that came from the online?
[40:32.550 --> 40:33.390]  Maybe we'll start there.
[40:34.050 --> 40:36.910]  What do you make of Ceno Browser and Briar?
[40:37.090 --> 40:39.190]  These were built for Autocratic Environies.
[40:40.010 --> 40:40.530]  Yes.
[40:40.830 --> 40:44.170]  So Briar is one that I took a look at.
[40:44.350 --> 40:46.390]  I think it's very interesting as a solution.
[40:47.990 --> 40:52.350]  My understanding of Briar is that there's some other factors at play.
[40:52.470 --> 40:53.470]  It's not true mesh.
[40:53.630 --> 40:56.770]  And we're looking like directly at that true mesh situation.
[40:56.770 --> 41:04.230]  But I would love in the future to be able to evaluate it and see because generally the reviews that I read are quite positive of how Briar works.
[41:06.470 --> 41:07.230]  Yeah.
[41:07.930 --> 41:27.070]  So one thing, um, one thing about like protests, of course they happen in an environment where things can't with things are mute that environment, which is kind of mutable when you think about like police barricade streets or kettle protesters and if we aren't necessarily like direct attacks on the
[41:27.070 --> 41:27.990]  protesting crowd.
[41:28.550 --> 41:33.010]  Um, have you, are you planning to account for that in your modeling as you move forward?
[41:33.550 --> 41:33.910]  Absolutely.
[41:34.270 --> 41:41.250]  We want to be able to account for like the physical features that can be put in place that could like directly block someone's path.
[41:41.470 --> 41:48.330]  So whether that be putting up barricades or perhaps there's like a line of cars in the way or like a line of people that you just don't want to interact with.
[41:48.470 --> 41:49.650]  We want to be able to model that.
[41:49.790 --> 42:01.930]  So that's absolutely something that we're looking into and we're hoping to get an idea from the interview study of what is most common, uh, and what people are seeing the most as an interruption to the way that they're trying to move.
[42:02.610 --> 42:03.210]  Yeah.
[42:04.090 --> 42:13.610]  Uh, so you've been talking about, uh, you know, communication technologies that use Bluetooth as their physical layer, uh, at least with the existing part.
[42:14.110 --> 42:20.250]  Um, I'd hope we've had two or three talks about MeshTastic, which use the ISM band.
[42:20.410 --> 42:24.750]  Why is there any discussion about how they handle their routing in progress situations or...?
[42:24.750 --> 42:29.890]  So my lab is very interested in MeshTastic and it's something that we're hoping to evaluate.
[42:30.170 --> 42:36.470]  Uh, this project, this mobility project is actually based off of the work of my advisor, Tashar Joyce.
[42:36.790 --> 42:40.310]  He gave a talk here last year, uh, that was on this project.
[42:40.470 --> 42:51.010]  I don't know if anyone might have seen it, but essentially as they wrote their paper and they did their evaluation, they noticed that these mobility models were an issue and that's sort of where my research comes out of.
[42:51.150 --> 42:56.310]  So as a lab, we're very interested in evaluating it at some point because the technology is really cool.
[42:56.910 --> 42:59.910]  Uh, for, for the moment, we're looking specifically at this instance.
[43:02.310 --> 43:02.830]  Yeah.
[43:03.510 --> 43:12.290]  Are you concerned that publication of better models or movement, uh, can be useful for police and for repressivity?
[43:13.170 --> 43:13.690]  Right.
[43:13.950 --> 43:18.670]  So this is something that we have thought about a lot when we're considering the ethics of this project.
[43:19.250 --> 43:32.310]  And I think that it's a complicated question because if we want to try and advance mesh technology specifically for this use, this use case, we need a better idea of how people move.
[43:32.430 --> 43:38.410]  But at the same time, we don't want to make something that can be flipped and used as a way to harm the people that we're trying to protect.
[43:38.970 --> 43:43.650]  Um, it's something that we've thought a lot about, in my opinion.
[43:43.910 --> 43:53.430]  I think that to make strides towards developing something better, something that can help keep people safe, it might be something that we have to dip our toes into, but very carefully.
[43:54.090 --> 44:08.030]  And, uh, the process of us designing the study when it comes to gathering information, we're trying to do it very intentionally, making sure that we keep people as private as possible to protect them so they don't face any repercussions for, uh, you know,
[44:08.390 --> 44:10.450]  uh, contributing to our research.
[44:10.570 --> 44:12.490]  But it's absolutely something we've thought about.
[44:12.590 --> 44:16.990]  And I think it's a, it's a hard question to answer, but that's sort of my position on it.
[44:17.170 --> 44:22.050]  Someone's going to do this anyways to better send ads at you or whatever.
[44:22.470 --> 44:29.850]  So might as well do the work and make sure that, you know, it's promoted to, you know, good social, you know, work.
[44:30.130 --> 44:30.570]  Absolutely.
[44:31.470 --> 44:31.910]  Yeah.
[44:32.650 --> 44:50.810]  So I look at this as a, a cat and I'm scared, because uh, as most people who are protesting due to, I fell or fuck with that, which has a certain frequency gain for carry away, police disease, uh, over a lot of those weapons, those, those biggest, don't not worry about national network,
[44:50.950 --> 44:53.170]  we have both network, you have imperatively, you have nothing.
[44:53.610 --> 44:53.770]  Right.
[44:53.970 --> 44:55.890]  Shout there if you get that way.
[44:56.610 --> 45:01.990]  Um, and then it also wouldn't change the pre-paces or you start hopping through PC, we'll catch up with that.
[45:02.230 --> 45:10.030]  So it's really, you know, but, Lauren Burskent has people who are looking at lower right now per sample.
[45:10.350 --> 45:12.350]  So figure out how to, how to.
[45:13.550 --> 45:14.090]  Absolutely.
[45:14.510 --> 45:27.050]  Uh, that's another thing that we're hoping to understand, uh, during our interview process, because you see two divergent, like schools of thought when it comes to using technology at protests, at least from what we reviewed.
[45:27.250 --> 45:30.830]  So you see people say, absolutely like have your phone.
[45:30.950 --> 45:34.050]  You need your phone to be able to communicate with people and let them know that you're safe.
[45:34.270 --> 45:37.670]  Then you see another group of people that say, do not bring your phone.
[45:37.790 --> 45:38.830]  That's extremely dangerous.
[45:38.830 --> 45:40.470]  You can be tracked, leave it at home.
[45:40.470 --> 45:42.150]  Or if you need a phone, bring a burner.
[45:42.450 --> 45:48.610]  And we as an academic community don't have a good grasp on what that actually looks like in practice.
[45:48.730 --> 45:57.590]  And I think once we go through the interview process and we're able to get an idea of what people are carrying and why they're carrying it, we can start to tailor solutions.
[45:57.870 --> 46:02.470]  Because it is true that there's always going to be an oppressive power that's pushing back against us.
[46:02.530 --> 46:04.150]  So we need to be able to innovate constantly.
[46:04.430 --> 46:07.770]  And I think that this interview will put us in a great position to do so.
[46:08.370 --> 46:11.690]  It's that you need to, you know, we're playing chess, right?
[46:11.810 --> 46:13.110]  And it's like, oh, okay, fine.
[46:13.250 --> 46:14.310]  Hey, this is awesome.
[46:14.490 --> 46:16.370]  I can communicate securely with everyone.
[46:16.370 --> 46:20.730]  But it's ISM ban, like, as far as like, oh, knocking, you know, that out?
[46:21.230 --> 46:21.630]  Easy.
[46:22.030 --> 46:25.330]  You know, we can knock out hope, you know, and that's fantastic.
[46:25.490 --> 46:25.910]  No problem.
[46:26.230 --> 46:29.470]  They can shut down civil service if you really wanted to.
[46:29.630 --> 46:34.010]  It's not a question of what tactics do they want to use against protest or how far they want to go with that.
[46:34.170 --> 46:37.090]  Because, you know, that also denies then information at data.
[46:39.130 --> 46:46.530]  I think they're in a better place to sort of knock out, um, you know, uh, consumer grade stuff.
[46:46.690 --> 46:46.850]  Right.
[46:46.990 --> 46:48.070]  But, you know, all keeping our own.
[46:48.230 --> 46:51.250]  Oh, no, I mean, it's not that information about the protestors in my point, no?
[46:51.730 --> 46:51.970]  Yeah.
[46:52.590 --> 46:53.210]  You're bold.
[46:53.450 --> 46:54.310]  So thank you.
[46:54.670 --> 47:00.650]  Uh, so I was wondering, what actual issues are you seeing with having bad models of crime movement?
[47:00.790 --> 47:05.510]  So you showed, like, that we have bad models, but like, what actually goes wrong?
[47:05.510 --> 47:06.230]  Sure.
[47:06.410 --> 47:07.210]  That's a really good question.
[47:07.590 --> 47:10.870]  So there is one paper and I can find it and I can send it to you.
[47:10.970 --> 47:20.590]  It is a little bit older, but they found that the random waypoint model in itself, when you're simulating, it has this tendency to take the nodes.
[47:20.770 --> 47:25.150]  So like, you know, it's randomly placing people, uh, with, you know, no order as to why they're going somewhere.
[47:25.550 --> 47:36.170]  And it has this tendency as the simulation runs over time to actually cluster everybody in the middle of that grid space, which goes against what we know, how people move.
[47:36.370 --> 47:48.630]  And there was also some talk in that paper of as time goes on, uh, because it's like an, like an endless simulation, uh, there's this decay that happens and people aren't able to complete their movements.
[47:48.850 --> 47:51.750]  So I think that all traces right back to it being random.
[47:51.810 --> 47:54.270]  Uh, we don't have a good idea of how people move.
[47:54.370 --> 48:01.570]  And when we try to use these features to try and supplement that, you see these, like these movement patterns that just don't match.
[48:02.610 --> 48:05.070]  Probably have time for one more question, I would say.
[48:05.470 --> 48:05.630]  Yeah.
[48:06.170 --> 48:06.310]  Yeah.
[48:06.670 --> 48:13.070]  Um, just on the topic of you brought up how you can't just release the beta, then have people test them and feel pretty obvious reasons.
[48:13.350 --> 48:15.990]  So just going out there, how's your team considered?
[48:16.570 --> 48:19.590]  And of course, you'd have to find a group that's willing to do this, or groups.
[48:20.050 --> 48:27.290]  Um, but just having people take them to the live scenario, and then you need, uh, uh, or communicating on dummy data over it.
[48:27.710 --> 48:29.750]  Talking about, I know, 20 for breakfast and stuff.
[48:30.730 --> 48:33.650]  I think that is an interesting approach.
[48:34.130 --> 48:46.090]  I would be very hesitant to issue anyone that's in these sorts of situations, any sort of technology or like anything that could bind them back to us, that could cause trouble for them.
[48:46.350 --> 48:50.510]  I think it would be great way to collect mobility and trace and have an idea.
[48:50.790 --> 49:01.810]  Uh, it would also cut out a lot of the, the work when it comes to like, how do we make it as easy as possible for people to tell us, you know, if you asked me, how did you move around today, Cora?
[49:01.990 --> 49:03.090]  I would probably shrug at you.
[49:03.210 --> 49:12.570]  So we're trying to figure out ways to make that like retelling easier, whether that has to be some sort of like tactile model that you move around, or if we're just giving you pen and paper and you draw.
[49:12.850 --> 49:16.350]  But, uh, that's definitely an interesting approach, and I'd love to talk more about it.
[49:20.160 --> 49:21.380]  One, one more?
[49:21.820 --> 49:22.300]  Okay.
[49:22.740 --> 49:24.060]  Uh, one more question.
[49:25.280 --> 49:25.760]  Hi.
[49:25.920 --> 49:26.780]  This is really great.
[49:27.620 --> 49:41.640]  I'd also say a lot of the types of mesh networks, then one of the things that is usually interesting about how we think of mesh is that, uh, often there's a presumption that sort of the network emerges spontaneously as an emergent property.
[49:41.980 --> 49:47.240]  I think, you know, with mesh networks, they're, first of all, they always, uh, ad hoc, and they're not always available.
[49:47.580 --> 49:47.900]  Mm-hmm.
[49:47.960 --> 49:49.440]  And they're not always perfectly portable.
[49:49.940 --> 49:52.340]  Sometimes they have state always certain announced priority.
[49:52.740 --> 50:09.820]  So I'm wondering if you guys ought to, or the book of the thinking about ending keys, the incorporating some of these on other aspects of mesh, for example, the group of people might mark certain nodes as more influential in advance, or, uh, place certain,
[50:09.820 --> 50:12.200]  uh, relay devices in the area.
[50:12.360 --> 50:22.060]  And, you know, like adding some stationary or some hierarchy or other things that aren't just purely boggated, Albert ad hoc, um, you know, um, emerging.
[50:22.800 --> 50:23.780]  Yeah, absolutely.
[50:24.080 --> 50:25.720]  It's definitely something that we've talked about.
[50:25.940 --> 50:30.420]  So, uh, I'll point you to my advisor's paper that mine is built off.
[50:30.500 --> 50:36.960]  It's called Amigo, uh, Amigo by Jashara Joyce, and you can find it online in archive.
[50:37.400 --> 50:41.560]  But essentially, they have a very interesting way of approaching what that looks like.
[50:41.680 --> 50:42.520]  They use clique routing.
[50:42.780 --> 50:49.000]  Uh, we've also talked about in lab ways that we could deploy, like, what you described like a deployable device.
[50:49.160 --> 50:51.180]  We described it as some sort of spine.
[50:51.540 --> 50:57.740]  Like, perhaps, because you see this happening a lot in urban environments going down a city street, you have buildings along the city street.
[50:58.000 --> 51:09.640]  Perhaps there's people that are willing, uh, willing to have some sort of device that helps propagate the signal down further to help people when their mobility isn't ideal for the communication we're trying to do.
[51:09.960 --> 51:17.140]  But absolutely, I'd love to talk more about it, because it's definitely something we've thought about, like, for the further down, uh, development of this.
[51:18.760 --> 51:19.580]  Thank you, guys.