[00:01.320 --> 00:08.320]  So we created a set of tools to study global peer-to-peer file sharing and the distribution of media on the Internet.
[00:09.260 --> 00:16.720]  Through Alpha60, we are learning the quantity, location, and time of uploads and downloads via the BitTorrent protocol.
[00:17.240 --> 00:21.440]  We put our anonymized results in public on our GitHub.
[00:22.120 --> 00:30.700]  And since 2018, we have collected data on approximately 568 media texts and over 4 billion downloads.
[00:31.120 --> 00:35.200]  That's approximately 7 million downloads per text.
[00:35.720 --> 00:37.320]  What do we sample?
[00:37.600 --> 00:43.500]  We sample media texts, by which we mean films, TV series, music, and books.
[00:43.500 --> 00:51.000]  We also sample leaks, such as leaks related to the Russian-Kranian War.
[00:51.820 --> 00:55.140]  So I think we should put invasion in that sentence.
[00:55.440 --> 01:01.360]  We sample leaks pertaining to Iran, to China, to Shara Discord leaks.
[01:02.060 --> 01:16.460]  We also sample files pertaining to 3D printable weapons, also known as ghost guns, like the Glock 19, allegedly used by Luigi Mangione to murder United Healthcare CEO, Brian Thompson, in December 2024.
[01:17.700 --> 01:21.460]  What Alpha60 measures that other systems don't?
[01:21.660 --> 01:30.160]  So, Alpha60 captures a more complete picture of information flows than official systems like Nielsen, Alpha60.
[01:30.160 --> 01:30.160]  So, what does Alpha60 do?
[01:30.700 --> 01:36.100]  By quantifying and mapping informal, quote-unquote, pirate network activity.
[01:36.820 --> 01:44.560]  Alpha60 measures global information exchanges, not single country ratings, which is mostly what you get from official systems.
[01:45.200 --> 01:51.960]  Alpha60 makes our data public via our GitHub, whereas streamers offer little data transparency.
[01:52.960 --> 01:59.540]  Alpha60's methodologies are cross-platform, unlike streamers, which use proprietary per-platform metrics.
[01:59.600 --> 02:17.100]  Alpha60 measures peer-to-peer sharing of information that is expressly prohibited or illegal, such as leaks of classified government documents, queer media texts in China, Russia, Saudi, and other countries with anti-LGBT laws, and U.S. media texts that are not given official releases in China and other countries.
[02:21.240 --> 02:23.860]  Okay, these are some of our generated outputs.
[02:24.640 --> 02:31.460]  This is an animation of Game of Thrones Season 7, Episode 1.
[02:31.760 --> 02:36.140]  This is the first five days cumulative downloading around the world.
[02:36.320 --> 02:41.260]  And so you can see certain places are awake, other places are sleeping.
[02:41.260 --> 02:47.600]  And then as time goes on, after the premiere on HBO, certain cities wake up.
[02:47.600 --> 02:51.800]  And as the circles get bigger, that indicates more downloading.
[02:53.060 --> 02:56.180]  Purples and pinks indicate higher res.
[02:56.600 --> 02:58.680]  Greens and blues indicate lower res.
[02:58.880 --> 03:03.680]  And when downloading is really hot, the name of the city will appear in letters.
[03:03.680 --> 03:06.760]  So you can see now Seoul is really awake.
[03:06.900 --> 03:08.440]  Athens is also very awake.
[03:08.620 --> 03:13.860]  But if I forward it a little bit, you'll see different cities start to wake up.
[03:13.960 --> 03:15.820]  Toronto, Sao Paulo, and so on.
[03:17.060 --> 03:20.760]  Okay, and then let me just advance one slide.
[03:21.780 --> 03:24.060]  All right, I'm going to let Benjamin talk about this one.
[03:24.880 --> 03:27.660]  This is a different map.
[03:27.660 --> 03:28.700]  We're going to go into the maps later.
[03:29.060 --> 03:32.940]  This is Stranger Things, the last Stranger Things, the weekly.
[03:33.260 --> 03:34.700]  This is a much shorter one.
[03:34.700 --> 03:39.760]  So these are the maps.
[03:40.100 --> 03:47.560]  Week one, two, three, four, five, six, seven, eight.
[03:47.840 --> 03:48.380]  That's it.
[03:48.540 --> 03:49.580]  I'll talk about the orientation.
[03:50.260 --> 03:52.140]  Well, I'll talk about the maps a little later.
[03:52.320 --> 03:52.500]  Okay.
[03:54.100 --> 03:54.840]  All right.
[03:55.260 --> 03:57.960]  And then, oh, I think this is still you.
[03:57.960 --> 03:58.560]  Yeah.
[04:00.060 --> 04:01.780]  This is another output.
[04:02.100 --> 04:07.080]  We are... This is an experimental map of Black Panther.
[04:07.360 --> 04:09.340]  This is just the North...
[04:09.340 --> 04:15.040]  This is a slice of the Americas just for the high-resolution peers or high-depth peers.
[04:15.220 --> 04:19.320]  So this is 1080p peers in the Americas.
[04:20.240 --> 04:25.040]  And you can... Basically, orange is 4K.
[04:25.540 --> 04:27.180]  Pink is 1080p.
[04:27.180 --> 04:29.420]  So this is just a 1080p slice.
[04:29.820 --> 04:40.220]  But you can really see the kind of big... The big usage in Rio, South America, which was a little bit surprising.
[04:40.380 --> 04:44.200]  I don't... I don't think the Americans stuff is not surprising.
[04:44.200 --> 04:53.200]  But what's kind of interesting about this map is the Caribbean area and the circle of high activity in the islands...
[04:53.200 --> 04:54.580]  In the islands part.
[04:55.260 --> 04:58.700]  And this is much smaller.
[04:59.160 --> 05:03.320]  It shows you just an idea of kind of like the distributions we're looking at.
[05:03.420 --> 05:10.200]  And so this is a 4K distribution from 2018 for the same text, Stranger Things.
[05:10.520 --> 05:18.940]  And you can see this is a little bit more indicative of the wealth of the people downloading and who has access to 4K display technology.
[05:18.940 --> 05:23.880]  So you can see that, again, East Coast United States really heavily represented.
[05:24.260 --> 05:29.960]  And then the Sao Paulo Rio also as hotspots in this.
[05:30.060 --> 05:35.360]  So we have a whole bunch of mapping techniques that we've developed as part of this.
[05:36.480 --> 05:45.320]  And part of this is refusing to use projections that are known flawed to represent Internet phenomena in the year 2025.
[05:45.320 --> 05:46.640]  We've got to get beyond this.
[05:46.880 --> 06:01.420]  So I work with some people at Constant in Brussels, Philip Riviere, and we're using a Cahill Keys projection modified, which was originally an anti-war, anti-nuke projection, which seems appropriate for us.
[06:02.440 --> 06:04.540]  And this is actually a map.
[06:04.540 --> 06:08.040]  I'm using this just to show the Americas slices here.
[06:08.040 --> 06:16.380]  We see the lines, this fiber, and then we have various other infrastructure markings on this, Tor exit nodes.
[06:16.680 --> 06:25.800]  We have points of presence for CDNs and a couple other network niceties in this.
[06:26.040 --> 06:32.780]  I have... this next slide is just the CDNs for North America.
[06:32.780 --> 06:37.040]  And this would be stuff like Netflix runs its own CDNs.
[06:37.120 --> 06:38.120]  Then we have Fastly.
[06:38.400 --> 06:39.200]  We have Cloudflare.
[06:39.280 --> 06:44.920]  We have all these people that are trying to get you that video streaming so that it's not janky at your home thing.
[06:45.020 --> 06:47.660]  What they use is servers all over the world to kind of cache data.
[06:47.800 --> 06:48.880]  And that's what a CDN is.
[06:49.040 --> 06:53.220]  So in the distribution of video on the Internet, CDNs are really important.
[06:53.740 --> 06:57.800]  Of course, peer-to-peer operates as its own CDNs.
[06:57.800 --> 07:01.420]  So this is kind of what peer-to-peer is competing against.
[07:03.180 --> 07:03.940]  Here's our GitHub.
[07:04.220 --> 07:05.040]  We've talked about it.
[07:05.140 --> 07:09.040]  This is how we put structured data in public about all our releases.
[07:09.460 --> 07:10.360]  Here it is.
[07:11.520 --> 07:14.120]  If you have questions, shoot us an email.
[07:15.060 --> 07:27.040]  The other thing we do is we have a website that talks about the bar bet, which is how are these media texts in relation to other texts?
[07:27.360 --> 07:36.160]  So from our sampling, we see in 2024 that Dune 2, hella, 26!
[07:36.960 --> 07:43.820]  26 weeks, we're looking at about 89 million downloads, unique peers during that period.
[07:43.820 --> 07:46.460]  That is an all-time record that still stands.
[07:47.140 --> 07:49.880]  We'll see what Stranger Things later in this year does.
[07:50.700 --> 07:53.360]  But Squid Game, also very impressive.
[07:53.980 --> 08:02.380]  Fallout, Top 3, Godzilla, Inside Out, Arcane, What If, House of the Dragon, Shogun, et cetera, et cetera.
[08:02.660 --> 08:13.760]  So this is a kind of a way to get relations about media texts that aren't usually relatable across platforms, film, and television, and some other things, which we'll get into later.
[08:15.940 --> 08:21.340]  Okay, so now we're going to do a deeper dive into volume of peer-to-peer traffic sampled by Alpha 60.
[08:21.640 --> 08:37.040]  So these are the number of media texts sampled by our toolkit per year from the year we started 2018 to now, the total number of downloads that we tracked per year, and the average number of downloads per text.
[08:37.040 --> 08:44.840]  So you can see that we've been mostly increasing our sampling, although 2021, California was still under COVID lockdown.
[08:44.960 --> 08:54.660]  We sampled quite a few more texts, but we've sort of evened out into the 60s per year 2022, 2023, 2024.
[08:54.660 --> 09:09.810]  But if you look at the total number of downloads column, you'll see that even though we maintain this roughly the same volume of text sampled for between 2022 and 2024, the number of downloads that we were tracking greatly increased.
[09:09.810 --> 09:22.530]  So in 2022, 65 texts yielded 353 downloads, but in 2024, 61 texts, we saw almost 2 billion downloads across those texts.
[09:22.730 --> 09:33.450]  So the average number of downloads of texts has been increasing pretty dramatically where, you know, in 2022, we were seeing 5.4 million downloads per text on average.
[09:33.450 --> 09:39.210]  In 2024, we're seeing 32.6 million downloads per text on average.
[09:39.210 --> 09:43.190]  And we only have, you know, up until mid-August in 2025.
[09:43.190 --> 09:52.070]  But it does look like the 2024 downloads per text numbers are sort of, you know, at roughly the same place.
[09:52.130 --> 09:56.670]  They haven't sunk back down to the 2023 or 2022 numbers.
[09:56.670 --> 10:03.770]  Hey, didn't you give an interview to Wired Magazine in 2023 saying it was a golden age of piracy?
[10:03.790 --> 10:08.610]  Okay, my interview to Wired Magazine was actually, or my op-ed for Wired was actually in 2020.
[10:08.990 --> 10:10.410]  So I was a little early.
[10:10.670 --> 10:12.350]  Okay, but these are some receipts for that point.
[10:12.350 --> 10:16.230]  I predicted that COVID would yield a huge spike in piracy.
[10:17.290 --> 10:20.690]  But it looks like that came a little bit later than COVID.
[10:21.410 --> 10:23.770]  Okay, this is just fun facts.
[10:24.290 --> 10:37.870]  So when we compare the total number of downloads that Alpha60 tracks to the total number of subscribers that Netflix has, you'll see that we were sort of on that par in 2019 and 2020.
[10:39.710 --> 10:46.630]  So it's not really, I mean, it's not like a solid data point that we need to advertise or anything.
[10:46.630 --> 10:56.470]  But in 2019 and 2020, the volume of downloads we were sampling was roughly equivalent to the volume of subscribers that Netflix had.
[10:56.650 --> 11:03.310]  And then our downloads far outstripped the number of Netflix subscribers in 2021 and 2022.
[11:03.770 --> 11:14.010]  And starting 2023, our total downloads that we were sampling was roughly equivalent to the total number of subscribers of all major streamers combined.
[11:14.730 --> 11:21.190]  Netflix, Amazon Prime, HBO Max, Hulu, Disney Plus, Paramount Plus, and Apple TV Plus.
[11:21.410 --> 11:26.570]  So just so you know how much we are downloading or how many downloads we track.
[11:26.750 --> 11:32.770]  Okay, now we're going to dig deep into one case study, one universe, which is the Star Wars television universe.
[11:33.810 --> 11:34.750]  Yeah, right.
[11:35.570 --> 11:36.890]  Thank you for that.
[11:38.170 --> 11:47.110]  I just want to remind people that Disney Plus is an American company, and it is not available everywhere.
[11:47.850 --> 11:50.330]  As a matter of fact, it was pulled in Russia after the war.
[11:50.330 --> 11:54.670]  So no surprise that we saw a lot more downloads in Russia after that.
[11:55.390 --> 12:10.430]  But here's an idea of kind of like, as we show you more maps, and as you can kind of put this together yourself, you can see that places that Disney neglects are Russia, China, large parts of Africa, large parts of the Middle East.
[12:11.110 --> 12:14.430]  They have a subsidiary in India called Hot Star that they bought.
[12:14.630 --> 12:19.190]  So we're going to count that as Disney Plus, even though it's technically a separate entity, but it's the same content.
[12:19.390 --> 12:20.970]  So we're going to go with that.
[12:22.430 --> 12:24.670]  Here is our sample set.
[12:25.690 --> 12:32.710]  From 2019 to 2025, these are all the specific episodes of MediaText that we sampled.
[12:32.930 --> 12:36.710]  And you'll notice that we're not sampling every set.
[12:36.710 --> 12:45.190]  We're sampling kind of the first two or three episodes, the last two or three episodes, and then we can interpolate kind of results from that.
[12:46.130 --> 12:50.410]  So this is the sample set for the rest of this talk.
[12:51.590 --> 13:03.650]  And then what I want to do is just kind of clarify a little bit of things when we say sampling, so that when we start saying things like billion, we are not downloading a billion movies.
[13:03.650 --> 13:05.850]  Just let's be really clear about that.
[13:06.030 --> 13:12.470]  What we are doing is we have a set sample length of 10 weeks, 15 weeks, 26 weeks.
[13:12.470 --> 13:15.390]  This is somewhat arbitrary.
[13:15.770 --> 13:17.670]  Nielsen uses a five-week limit.
[13:17.990 --> 13:24.690]  And we wanted to kind of establish a long tail and try to come up with more interesting metrics.
[13:25.190 --> 13:27.090]  The procedure is we collect torrents.
[13:27.410 --> 13:29.630]  We create a torrent set for each MediaText.
[13:30.210 --> 13:31.390]  So let's say and or 101.
[13:33.110 --> 13:38.190]  I can't actually remember the size of that set right now, but they're all between 100 and 500.
[13:39.070 --> 13:41.190]  We put in a cloud samplers.
[13:42.190 --> 13:43.990]  That architecture is a separate talk.
[13:44.130 --> 13:44.950]  We're not going to talk about it.
[13:45.490 --> 13:51.290]  And then we just run that every three minutes for weeks or for half a year.
[13:52.190 --> 13:54.890]  Again, we are sampling metadata only.
[13:55.190 --> 13:57.050]  We are not downloading movies.
[13:57.510 --> 14:04.730]  We consider the ability to track distribution, a feature, not a bug, of the BitTorrent protocol.
[14:06.550 --> 14:08.130]  Okay, here's some results.
[14:10.010 --> 14:11.590]  There's a lot of ways to look at this.
[14:11.810 --> 14:14.030]  So I'm going to present a couple different views of this data.
[14:16.070 --> 14:18.550]  This is normalized start.
[14:18.770 --> 14:19.510]  So what does that mean?
[14:19.670 --> 14:26.090]  So that because we're starting things over different years, we're going to normalize everything to the first week that they were released.
[14:26.090 --> 14:34.570]  So on this, you can kind of see this is a weekly cumulative of unique peers or unique downloads.
[14:35.090 --> 14:53.730]  And you can see on the right side labeling, you can see that per week Mandalorian 208 at the time of the creation of this graph was sampling very steadily at about 2.5 million peers a week.
[14:53.730 --> 14:58.090]  Which is pretty impressive for peer generated CDN.
[14:58.650 --> 15:08.710]  And you can see the other, the next performant title would be the Acolyte series, which Abigail is going to have some commentary on later.
[15:09.410 --> 15:11.750]  And then Ahsoka also coming in really strongly.
[15:12.410 --> 15:16.650]  You see kind of Mandalorian fading a little bit in Andor first season.
[15:17.010 --> 15:23.250]  Obi-Wan and Mandalorian third season kind of tailing, tailing all the rest.
[15:24.550 --> 15:26.010]  Here's another way to look at it.
[15:26.250 --> 15:30.890]  And that is instead of like every week, we think of every week cumulative.
[15:30.890 --> 15:35.150]  This makes the situation a little bit different.
[15:36.030 --> 15:42.870]  We can see that in terms of the unique peers, Acolyte is very far ahead of Mandalorian.
[15:43.230 --> 15:45.770]  I did this graph before we had a complete sample.
[15:46.090 --> 15:53.510]  So I can tell you now that Acolyte 101 did indeed surpass Mandalorian 208 at the 26 week mark.
[15:53.510 --> 16:03.570]  So Acolyte, that Acolyte series is actually the top sampled Star Wars property that we have dealt with so far.
[16:05.030 --> 16:13.410]  And then when I presented this result in San Francisco, they were asking about, we would like to see this in time.
[16:13.590 --> 16:15.250]  So you normalize these starts.
[16:15.450 --> 16:16.070]  It's really interesting.
[16:16.070 --> 16:19.370]  But tell me about how this was actually received at the time.
[16:21.610 --> 16:22.590]  So here we go.
[16:22.590 --> 16:25.170]  So 2019, bottom right.
[16:25.310 --> 16:26.390]  Sorry, the slide is not labeled.
[16:26.490 --> 16:27.290]  That little green line.
[16:27.470 --> 16:32.610]  So these are all normalized for the kind of bigger downloads we see later in the years, like 2025.
[16:32.910 --> 16:35.030]  So Mandalorian, huge success.
[16:35.370 --> 16:37.190]  These are like the first four or five weeks.
[16:37.350 --> 16:39.770]  We hit like, I don't know, 6.5 million.
[16:41.670 --> 16:42.710]  This is the next year.
[16:42.950 --> 16:47.550]  So it's kind of continuing in through the last, through the first part of the year.
[16:47.550 --> 16:48.530]  Where it kind of tops out.
[16:48.630 --> 16:52.130]  Mandalorian 101 tops out around 19 million.
[16:53.190 --> 16:57.810]  And then you see a dark green line on the right side of the screen.
[16:57.990 --> 17:00.790]  That's the second season of Mandalorian starting.
[17:04.330 --> 17:07.230]  And then, so that's continuing in the same line.
[17:07.430 --> 17:12.930]  You'll notice if you have a visual memory that these lines from Mandalorian are roughly the same.
[17:13.050 --> 17:13.790]  They're the same slope.
[17:13.990 --> 17:14.910]  They're about the same height.
[17:16.310 --> 17:23.790]  So Mandalorian season one, season two, we're inferring are relatively the same in terms of distribution.
[17:24.490 --> 17:27.850]  We did sample all the Star Wars animated things.
[17:27.910 --> 17:30.390]  We think they're really important for the Star Wars universe, for the Star Wars fans.
[17:31.270 --> 17:35.270]  But this is not really what we're going to dwell on, on this talk.
[17:35.870 --> 17:37.690]  Okay, here's where things start to get a little bit more interesting.
[17:38.510 --> 17:39.830]  Book of Boba Fett.
[17:41.030 --> 17:42.010]  I liked it.
[17:42.710 --> 17:44.950]  Not as many people did the other ones.
[17:45.430 --> 17:50.970]  Obi-Wan Kenobi, you know, there was some Mandalorian in that, but maybe people didn't realize it.
[17:52.010 --> 17:54.350]  Again, these are kind of underweight, the other ones.
[17:55.410 --> 18:01.270]  Andor season one starts out, kind of ends in a nice place.
[18:01.370 --> 18:05.350]  I would say Mandalorian season one S, maybe a little lower, a little flatter line.
[18:06.690 --> 18:08.950]  Then we get to Mandalorian season three.
[18:09.910 --> 18:11.630]  Maybe that's underperforming a little.
[18:12.310 --> 18:16.730]  And so on these lines, we see a dark line is episode one.
[18:16.910 --> 18:19.730]  The dash line is the last part of that.
[18:19.990 --> 18:24.350]  So we see that the dash line is...
[18:24.350 --> 18:25.870]  It's the last episode of the season.
[18:26.370 --> 18:27.090]  Oh, yes.
[18:27.270 --> 18:28.250]  It's the last episode of the season.
[18:28.470 --> 18:30.430]  So it's about the same slope.
[18:31.250 --> 18:33.080]  So it's about the same, but eh.
[18:34.420 --> 18:37.560]  2023, things start to pick up at the end with Ahsoka.
[18:38.480 --> 18:42.060]  Ahsoka, you can see the slant on that line is a lot higher.
[18:42.320 --> 18:44.560]  So that just means like more interest.
[18:45.400 --> 18:50.180]  And you can see the dotted line on Ahsoka is the end of Ahsoka season one.
[18:50.180 --> 18:52.740]  And so you're seeing that those slopes match.
[18:53.000 --> 19:00.380]  So you're looking at that series having continuity through the life of it, where there's as much interest at the beginning as there is at the end.
[19:01.480 --> 19:02.340]  Whoa, what's this?
[19:02.680 --> 19:06.860]  So it turns out Ahsoka ends up above Mandalorian on this.
[19:07.360 --> 19:09.020]  Um, which is pretty interesting.
[19:09.420 --> 19:15.900]  Um, and then the real interesting thing is Acolyte, which looks like, wow, that looks like a huge success here.
[19:16.220 --> 19:21.180]  Um, also, um, wow, that it ends stronger than it starts.
[19:21.520 --> 19:22.660]  That's also cool.
[19:23.740 --> 19:25.440]  Abigail, why do you think that got canceled?
[19:25.920 --> 19:26.400]  Right.
[19:26.620 --> 19:42.020]  So I think that there's, um, just a couple of hypotheses we have about the success of Ahsoka and Acolyte, which, um, if there are Star Wars people here, you already know this, that Ahsoka is a character that debuted in the animated series called Clone Wars.
[19:42.180 --> 19:47.240]  Um, Clone Wars, um, you know, like some people think redeemed the terrible prequels.
[19:47.420 --> 19:53.080]  And Ahsoka has been able to like build her fandom over, um, more than a decade.
[19:53.080 --> 20:01.020]  So when a live action Ahsoka, um, show debuted, perhaps that accounts for this, you know, outweighed success.
[20:01.020 --> 20:14.300]  But it doesn't really, that, that hypothesis doesn't really account for the fact that Obi-Wan Kenobi, uh, beloved legacy character from the original trilogy, doesn't get as much, um, play in the pirate network as Ahsoka does.
[20:14.440 --> 20:18.040]  So there wasn't as much hunger to see live Obi-Wan as there was live Ahsoka.
[20:18.300 --> 20:27.980]  Um, but it's notable that Ahsoka and Acolyte, the two highest performing texts, according to Alpha 60 in the Star Wars universe, are both led by women of color.
[20:28.340 --> 20:29.760]  I just want to note that.
[20:30.180 --> 20:38.820]  And the Acolyte, um, was canceled after one season, despite what we are tracking as its very strong performance.
[20:38.820 --> 20:44.820]  It's like, you know, it's, it's basically top performance in the Star Wars television universe.
[20:44.820 --> 20:54.400]  Now, the Acolyte had some very distinct characteristics in the storytelling, one of which was that it was all people of color who were leads in the cast.
[20:54.660 --> 21:10.100]  Um, another one is that the Acolyte is the story of a young, um, Padawan, you know, somebody who was in Jedi training, who, um, goes over the dark side, led by a Sith Lord, who tempts her sexually.
[21:10.400 --> 21:16.680]  So I think that's maybe the first time that's happened in the Star Wars universe, that there is, um, that evil wins.
[21:17.040 --> 21:24.140]  And evil wins through, um, you know, of course, lust is probably a part of the dark side, but we never saw that on screen before.
[21:24.400 --> 21:33.060]  So I think that there's, um, some reason, some story reasons why the Acolyte was canceled, but it is notable, um, according to our findings.
[21:35.720 --> 21:38.680]  And then this, that brings us to this year, uh, to date.
[21:38.860 --> 21:40.760]  So again, the sampling's not done.
[21:41.100 --> 21:42.600]  We're still at mid season on this.
[21:42.720 --> 21:46.260]  And this, we have three confusing lines instead of, uh, just two.
[21:46.460 --> 21:48.340]  Um, so we have the dark one is...
[21:48.340 --> 21:48.860]  Oh, you're saying it's Andor.
[21:48.980 --> 21:49.760]  Andor, sorry.
[21:50.160 --> 21:51.540]  Uh, Andor, this is Andor season two.
[21:51.940 --> 21:54.660]  The dark line is the episode one.
[21:55.000 --> 21:59.740]  The small dots are the last three of Andor.
[22:00.260 --> 22:00.640]  And then these...
[22:00.640 --> 22:01.000]  One season.
[22:02.740 --> 22:11.800]  Um, and then these little dash lines are us sampling season one, right when season two was released.
[22:12.100 --> 22:16.500]  So what we're seeing here is long tail, right?
[22:16.720 --> 22:23.780]  And we're seeing that after the second season was released, season one got a lot more popular.
[22:23.780 --> 22:30.060]  So this is important to think about in terms of what's going on with media and seriality and streaming.
[22:30.400 --> 22:40.040]  And why are they doing things like all these sequels and then breaking up things like Wednesday into part one, part two, et cetera, et cetera.
[22:40.040 --> 22:41.960]  And they're doing the same thing with stranger things.
[22:42.160 --> 22:43.520]  So this could be a reason.
[22:45.480 --> 22:45.960]  Okay.
[22:46.280 --> 22:48.980]  And I just feel like I have to gratuitously show you map.
[22:49.200 --> 22:57.480]  This is a map of the Mandalorian distribution, uh, all throughout the, uh, this Pacific plate.
[22:59.520 --> 23:02.640]  Um, again, really strong representation in South America.
[23:03.560 --> 23:05.540]  And, um, this is Europe.
[23:05.720 --> 23:11.460]  But what, one of the things we really notice is that Europe has about 50, over 50% of piracy always.
[23:11.460 --> 23:13.800]  Asia kind of brings up 27%.
[23:13.800 --> 23:16.660]  America is kind of like around 20%.
[23:16.660 --> 23:21.580]  So this is kind of like routine for media distribution on, on, uh, distributed networks.
[23:21.980 --> 23:25.180]  The other things to note here is, uh, Africa.
[23:25.500 --> 23:30.760]  So Africa is tremendously understudied in media distribution in this country.
[23:31.140 --> 23:42.740]  Um, and, um, you can see the parts of Africa that are lit up by this text are, um, Um, uh, Ivory Coast.
[23:43.020 --> 23:50.060]  And if you remember the network maps, you can see that there's very strong network linkages to that part of Africa.
[23:50.860 --> 23:54.220]  And South Africa also is the major player here.
[23:54.460 --> 24:00.200]  Um, the other thing of note here is, of course, uh, India, super hot.
[24:00.600 --> 24:07.040]  And I didn't do, I didn't do separate high-def low-res, uh, separations on this.
[24:07.040 --> 24:12.260]  But in general, Korea, Beijing are the hotspots for 4K downloading.
[24:12.560 --> 24:20.140]  This is definitely related to the more advanced network infrastructure in those countries where they can download a 4K movie.
[24:20.240 --> 24:21.440]  No, no big, no big deal.
[24:23.360 --> 24:23.940]  Okay.
[24:24.100 --> 24:29.900]  So now we're going to get into, um, the rankings of the top downloading countries and cities for specific titles.
[24:29.900 --> 24:37.420]  So I just chose three titles in the Star Wars TV universe, um, Ahsoka, Andor, and Acolyte, probably because they all begin with A.
[24:37.660 --> 24:45.280]  So here are the top downloading countries and cities for Ahsoka season, season one, episode eight.
[24:45.480 --> 24:46.580]  So the season finale.
[24:46.580 --> 24:52.020]  So you'll see, um, um, downloading countries, Russia, Korea, USA are the top.
[24:52.200 --> 24:55.480]  Downloading cities, Moscow, Seoul, and Shanghai are the top.
[24:55.740 --> 24:58.380]  And you can take a look at the rest for a moment.
[24:59.200 --> 25:02.720]  And then we'll look at Andor season two, episode 10.
[25:02.980 --> 25:05.860]  Top downloading countries, Russia, Korea, USA.
[25:06.640 --> 25:09.480]  Top downloading cities, Moscow, Shanghai, Seoul.
[25:09.480 --> 25:12.820]  And take a look at the rest of the lists.
[25:14.240 --> 25:20.480]  And then Acolyte season one, episode seven, top downloading countries, Russia, Korea, Turkey.
[25:20.800 --> 25:24.100]  Top downloading cities, Moscow, Seoul, Shanghai.
[25:24.320 --> 25:28.000]  And you can take a look at the rest of the lists.
[25:29.520 --> 25:30.160]  Okay.
[25:30.160 --> 25:34.740]  So here's our comparison across Ahsoka, Andor, and Acolyte.
[25:34.740 --> 25:39.680]  There are nine countries that are top countries for all three texts.
[25:39.920 --> 25:46.320]  Russia, Korea, USA, Brazil, Turkey, France, China, the Netherlands, the United Kingdom.
[25:46.660 --> 25:50.940]  There are seven cities that are top cities for all three texts.
[25:51.160 --> 25:56.100]  Moscow, Seoul, and Shanghai are the top three for all three texts.
[25:56.260 --> 25:57.720]  But not always in that order.
[25:57.960 --> 25:59.740]  So what's going on in those three cities?
[25:59.740 --> 26:00.480]  I don't know.
[26:00.480 --> 26:04.800]  Maybe they just have an especially strong Star Wars fandom in those places.
[26:05.120 --> 26:15.100]  And then the other cities that are consistently in the top cities lists are St. Petersburg, Amsterdam, Istanbul, and Ashburn, Virginia.
[26:15.880 --> 26:20.380]  Now, Ashburn, Virginia is a place that...
[26:20.380 --> 26:21.200]  Oh, okay.
[26:21.460 --> 26:23.900]  So a comparison across Ahsoka...
[26:23.900 --> 26:35.360]  So continuing the comparison across these three texts, let's just hypothesize, you know, do countries that don't have access to Disney Plus download a lot?
[26:35.580 --> 26:37.720]  And so Russia is a top downloader.
[26:38.200 --> 26:43.880]  China is a top downloader, but all the rest of these countries do have legal access to Disney Plus.
[26:43.880 --> 26:47.500]  So it's not always about legal access.
[26:47.840 --> 26:58.520]  The other thing to note is that, you know, the other question I asked is, well, do countries whose first language is in English download texts whose language is English?
[26:58.680 --> 27:00.980]  So the USA is a top downloader.
[27:01.100 --> 27:02.540]  The UK is a top downloader.
[27:02.680 --> 27:07.220]  They are, you know, speakers of English and that's the language of Star Wars.
[27:07.220 --> 27:14.380]  But nevertheless, all the other countries on this list are not English first language countries.
[27:14.640 --> 27:18.520]  So language isn't, you know, apparently a super motivator.
[27:18.680 --> 27:20.360]  It does seem like...
[27:20.360 --> 27:32.780]  I mean, because we know that Disney Plus releases texts with subtitles and dubs and they're really good, the pirate network gets those files right away as well.
[27:33.000 --> 27:42.220]  So it probably doesn't matter what the language of the original, you know, what the original language is, because people in any language can consume these texts right away.
[27:42.500 --> 27:44.420]  OK, these slides just basically say that.
[27:45.760 --> 27:46.440]  All right.
[27:46.580 --> 27:53.880]  What about Ashburn, Virginia, which was in a top downloader of all of those three Star Wars TV shows?
[27:54.360 --> 27:59.420]  There were other cities that appeared in the top 10 for several of those shows.
[27:59.580 --> 28:01.520]  One is Yekaterinburg, Russia.
[28:01.840 --> 28:04.140]  Another is Novosibirsk, Russia.
[28:04.440 --> 28:23.060]  So then I just started hypothesizing about what might be driving super traffic in cities that we just haven't heard of, you know, because every other top downloading city except for these three are like global cities like Amsterdam, like, you know, Istanbul.
[28:23.660 --> 28:25.460]  So what's up with these three?
[28:25.660 --> 28:29.840]  So the first thing I looked at was the number of data centers in each of these cities.
[28:30.360 --> 28:33.360]  And Novosibirsk only has six data centers.
[28:33.540 --> 28:40.580]  Yekaterinburg only has nine data centers, although it is called a critical regional data center hub in Russia.
[28:40.580 --> 28:43.420]  But Ashburn, Virginia has 143 data centers.
[28:43.600 --> 28:48.000]  So a lot of the traffic we see in Ashburn, Virginia might be VPN traffic.
[28:49.040 --> 28:51.280]  Hosted by AWS East.
[28:51.620 --> 28:52.300]  Right, right.
[28:52.480 --> 28:52.900]  Amazon.
[28:53.340 --> 28:58.380]  So another hypothesis I wondered about is what about military installations?
[28:58.680 --> 29:04.180]  Because just anecdotally, what is there to do if you're in the military and stationed on a base?
[29:04.180 --> 29:07.740]  You pirate a lot of TV or you pirate a lot of movies.
[29:08.080 --> 29:11.720]  So I just looked at how many military installations are in each of these places.
[29:12.700 --> 29:17.580]  Andrews is in Ashburn, Virginia, or at least within an easy drive of there.
[29:17.940 --> 29:24.580]  But Yekaterinburg is the headquarters of one of only five military districts in Russia.
[29:24.580 --> 29:27.980]  So Yekaterinburg is a major military hub of Russia.
[29:28.200 --> 29:30.300]  And Novosibirsk also has...
[29:30.300 --> 29:37.780]  It has the 41st Army stationed in Novosibirsk and several other military units, including Air Force bases.
[29:38.100 --> 29:46.100]  So it does seem like military presence might be a reason, might account for some of this super downloading there.
[29:46.100 --> 29:47.440]  I looked at universities.
[29:47.900 --> 29:56.080]  There's just a kind of, you know, a kind of conventional wisdom that universities do a lot of downloading.
[29:56.700 --> 30:02.080]  There are a couple of universities, you know, within 10 to 35 minutes of Ashburn, Virginia.
[30:02.420 --> 30:06.000]  There are nine universities each in Yekaterinburg and Novosibirsk.
[30:06.060 --> 30:07.680]  So that could be driving traffic.
[30:07.880 --> 30:15.200]  Maybe the orange cells are the maybes and the green cells are things that I think really are probably drivers.
[30:15.200 --> 30:18.880]  I also put a category in here called other.
[30:19.240 --> 30:34.540]  And I just wanted to say you'll see the green cells under the Russian cities that Yekaterinburg and Novosibirsk are cities that rank at the second fastest downloading speeds among Russian cities.
[30:34.540 --> 30:38.740]  They tie in terms of downloading speeds with St. Petersburg, Russia.
[30:38.740 --> 30:48.620]  So Moscow is by far, has by far the fastest downloading speeds in Russia, but there are three cities at that second fastest level.
[30:49.080 --> 30:51.960]  And Yekaterinburg and Novosibirsk are two of them.
[30:52.280 --> 30:59.880]  So a high number of data centers, one or more military bases and very fast Internet may be contributing factors.
[30:59.880 --> 31:00.800]  Sorry, that's a misspelling.
[31:01.220 --> 31:01.320]  Sorry, that's a misspelling.
[31:01.320 --> 31:05.040]  In high volume, far above average peer to peer to peer downloading.
[31:05.700 --> 31:06.220]  Okay.
[31:06.480 --> 31:07.240]  Thank you so much.
[31:14.800 --> 31:17.420]  We will accept questions if you have them.
[31:17.960 --> 31:18.280]  Yes.
[31:18.480 --> 31:18.980]  Oh, okay.
[31:19.060 --> 31:19.280]  Sorry.
[31:19.760 --> 31:20.480]  You first.
[31:20.680 --> 31:21.060]  And then you.
[31:21.380 --> 31:21.500]  Yes.
[31:22.040 --> 31:26.360]  Are you for 2025 doing only murderers in the building?
[31:26.600 --> 31:26.740]  Yes.
[31:26.920 --> 31:27.220]  We have them.
[31:27.820 --> 31:34.200]  We restricted this just to Star Wars just to kind of like contain us a little.
[31:34.520 --> 31:37.300]  But yeah, we're definitely doing only murderers.
[31:37.480 --> 31:39.220]  And we have done all seasons of only murderers.
[31:39.580 --> 31:40.480]  Do you have data on that?
[31:40.800 --> 31:41.400]  Yes, we do.
[31:41.780 --> 31:42.600]  Just go to your website?
[31:43.060 --> 31:43.420]  The GitHub?
[31:43.620 --> 31:46.460]  Yeah, the GitHub does have the data for that.
[31:46.700 --> 31:47.040]  All right.
[31:47.700 --> 31:48.060]  Yes.
[31:48.060 --> 32:00.000]  I wanted to know if you've profiled any other types of users that would be likely to, one, have, you know, the know-how into pirating media?
[32:00.280 --> 32:01.880]  Like, what kind of persons are there?
[32:02.060 --> 32:05.740]  I mean, the definition has kind of changed from, like, I think it's 10 years ago.
[32:06.200 --> 32:13.370]  I think you were technologically savvy and having more, or greater access to devices and tools.
[32:13.590 --> 32:21.470]  You might be, you might have more knowledge than even millennials and Gen Z in how to, uh, touring, uh, movies and such.
[32:21.630 --> 32:21.830]  Mm-hmm.
[32:21.970 --> 32:43.430]  And then I also wanted to know, like, you took the top three, um, shows, and I know you're not screenwriters, but, like, would, would people from non-English speaking countries or people who aren't as cognizant of American culture be a place of greater importance on plot structure for how good a show is?
[32:44.350 --> 32:53.850]  Because anecdotally, I have, uh, cousins who are very verbose in the entire Netflix catalog, more so than I, I have to finger on the postings.
[32:53.970 --> 33:05.010]  You probably see everything, um, without having been there and really absorb and dissect, um, everything about the show into a high caliber, like, almost Shakespearean.
[33:05.010 --> 33:05.350]  Mm-hmm.
[33:05.410 --> 33:13.870]  And so they're really, they might have more of a critical edge, like, as, more so as a movie critic because it's, like, culturally different.
[33:14.190 --> 33:14.810]  Yeah, yeah.
[33:15.330 --> 33:19.930]  Okay, I'll, I'll speak to the first question and then I think we both have things to say about the second question.
[33:20.110 --> 33:21.330]  So thanks for those great questions.
[33:21.550 --> 33:32.310]  So the first question was about who is downloading and could younger generations have more technical know-how and so participate more in the peer-to-peer network than, um, older generations.
[33:32.310 --> 33:35.030]  So, I mean, I think that's probably true.
[33:35.490 --> 33:41.570]  Um, I also have done interviewing of, um, media pirates, media and information pirates.
[33:41.810 --> 33:47.890]  And so you can read my book that's coming out next year called The General Library from MIT Press.
[33:48.210 --> 33:54.510]  And The General Library came from, that term came from one of my interviewees who misspoke.
[33:54.510 --> 33:58.750]  She said, I download all my science textbooks from The General Library.
[33:58.750 --> 34:00.650]  I said, Oh, what's that?
[34:00.850 --> 34:02.970]  And I've never heard of that site.
[34:02.970 --> 34:05.550]  And she said, Oh, it's called Gen Lib.
[34:05.810 --> 34:06.910]  It stands for General Library.
[34:07.090 --> 34:10.430]  And I mentally knew, Oh, no, that's not correct.
[34:10.590 --> 34:11.610]  It's called Lib Gen.
[34:11.790 --> 34:13.130]  It stands for Library Genesis.
[34:13.510 --> 34:22.830]  But I found her term General Library to be so useful and helpful for just denoting that there's always been a way to get texts illicitly.
[34:22.830 --> 34:29.190]  There's always been a way to not have to pay for certain media and information and to still acquire it.
[34:29.310 --> 34:34.610]  And so I call all of the pirate network for all of time, the general library.
[34:34.870 --> 34:37.050]  And who uses the general library?
[34:37.050 --> 34:54.430]  My book actually ended up focusing exclusively on people of color, queer people, women, disabled people, and other minorities, because the general library is immensely helpful to those groups, because, you know, not just for, of course, socioeconomic reasons,
[34:54.510 --> 34:58.430]  but also because information is really restricted for a lot of groups.
[34:58.430 --> 35:09.550]  For example, girls who grow up in very conservative religious households cannot access a lot of media, they are told not to watch or listen to a lot of kinds of media.
[35:10.030 --> 35:22.450]  And the same with children who are suspected of being queer, their parents put prohibitions, bans, give them talks about how we don't allow that kind of media in this household, and so on.
[35:22.450 --> 35:32.030]  So my book really focuses in on how people who are disadvantaged or minoritized or marginalized really turn to the general library for media and information.
[35:32.310 --> 35:40.230]  And one thing that I argue is that they have, these are people who don't have a lot of privilege, but what they have is technological privilege.
[35:40.230 --> 35:52.970]  So I feel like what you're talking about is technological privilege, if you know how to pirate, if you know enough about how to work a technical system, then you can get all the media and information that you want.
[35:53.290 --> 36:03.130]  And people are used, like, for example, college students who have scholarships to college, but they don't, they aren't given a ton of extra money to afford all of their science textbooks.
[36:03.130 --> 36:08.470]  They use their technological privilege in order to close what I call the resource gap.
[36:08.830 --> 36:17.090]  So I think that's really important that young people are leveraging their technological privilege to, you know, try to make their way in this world.
[36:17.390 --> 36:22.490]  So I think that's a really important, you know, finding of my interviewing.
[36:22.490 --> 36:39.270]  But as for the second question, which is also really great about whether, for example, like, people who don't speak English as their first language might turn to American media and prefer plot over dialogue or something like that.
[36:39.450 --> 36:44.530]  I do think we see, do we see, like, a preference for science fiction over comedy, for instance?
[36:44.530 --> 36:56.890]  Like, I do think comedy is very dialogue heavy, but that science fiction has a lot of tropes that are just, you know, commonly globally understood kinds of significations.
[36:57.130 --> 36:59.450]  So that's just like, I don't know, one hypothesis.
[37:00.230 --> 37:01.470]  She's the media expert.
[37:02.110 --> 37:11.570]  So yeah, and that, or the thinking is that comedy does not translate as well globally as things like sci-fi.
[37:12.110 --> 37:17.090]  So we were, we do sample some comedy and we do only murders.
[37:17.350 --> 37:18.030]  Only murders is comedy.
[37:18.030 --> 37:19.070]  And some other things.
[37:19.790 --> 37:21.910]  But we, we don't know.
[37:22.070 --> 37:23.830]  And we don't have the tools.
[37:23.930 --> 37:31.450]  We'd certainly be interested in thoughts about how to do structural analysis of text to kind of like quantify complexity, that kind of thing.
[37:31.450 --> 37:34.010]  So if you're, if you're doing research in that area, let us know.
[37:34.270 --> 37:35.570]  We would love to work with you on a dataset.
[37:35.930 --> 37:41.090]  But getting back to your point about demographics, we're only capturing IP addresses and time.
[37:41.390 --> 37:46.410]  So we don't actually have demographic data like this subscriber was 17 years old.
[37:46.550 --> 37:48.350]  That's not information we have.
[37:48.790 --> 37:53.470]  However, with advanced databases, you can pick apart IPs.
[37:53.470 --> 38:04.450]  And so things that I would associate with a younger demographic would be downloading video on a phone, using a mobile network to download.
[38:04.770 --> 38:08.890]  And we recently got access to those kinds of databases from IP info.
[38:09.330 --> 38:11.890]  And we're seeing some really interesting things there.
[38:11.890 --> 38:19.150]  And I would say indirectly, I would say that mobile downloadings, I would associate more with younger people.
[38:20.030 --> 38:21.370]  I can't prove that.
[38:21.590 --> 38:25.170]  But I would say like that's probably a pretty strong argument.
[38:25.830 --> 38:28.450]  So we, yeah, we do have those kinds of separations.
[38:28.710 --> 38:30.650]  We've seen some really interesting things there.
[38:32.090 --> 38:36.110]  We have seen Spain.
[38:36.110 --> 38:39.930]  It depends on the mobile company in the country that you're doing.
[38:40.050 --> 38:45.490]  But we're seeing very large variances in wireless downloading for media techs.
[38:45.650 --> 38:55.690]  And one of the things that's really interesting to us is there's like a third of the mobile downloads for media in Spain are wireless.
[38:56.050 --> 38:57.070]  They're on a mobile device.
[38:57.790 --> 39:06.390]  And what's interesting is that has also trans... that has jumped continents to other countries in the Spanish diaspora.
[39:06.550 --> 39:10.890]  So you're seeing the same thing in like Mexico and some of these other Spanish language countries.
[39:11.150 --> 39:16.970]  So I don't know how related that is to demographics, but that's kind of where I would go with that.
[39:17.390 --> 39:17.650]  Yeah.
[39:18.090 --> 39:31.170]  And maybe the other thing we can pick apart with IPs is if you think younger people are more concerned with privacy, are using things like hosted services and VPNs, that's also data that we can get.
[39:31.330 --> 39:33.370]  But age, yeah, we don't know.
[39:33.450 --> 39:34.290]  We don't know that for sure.
[39:34.790 --> 39:35.010]  Mm-hmm.
[39:35.730 --> 39:39.610]  So we're tracking multiple torrents across the globe.
[39:39.870 --> 39:46.790]  Were you able to see like for an individual peer like the group of torrents that they might be seeding?
[39:46.790 --> 39:53.170]  And also like aggregate kind of see like this kind of peer is into this stuff and other peers are into other stuff.
[39:53.290 --> 39:54.550]  And like what would be grouped together?
[39:54.830 --> 39:57.370]  We haven't done that but could, is the answer.
[39:58.170 --> 39:58.310]  Interesting.
[39:58.630 --> 39:59.530]  Back really quick.
[39:59.950 --> 40:01.810]  Yeah, we're really quick with a hand all the way in the back, yeah.
[40:02.190 --> 40:07.990]  Can you talk a little bit more about the evolution of where the files are, how they're coming from?
[40:07.990 --> 40:13.010]  The DRM and where you're at, where you're expecting that to be a challenge in the future.
[40:13.230 --> 40:18.450]  I'm fascinated by how days after something can be released on Netflix that it's available.
[40:19.350 --> 40:20.470]  Whether it's...
[40:20.470 --> 40:21.830]  Oh, we do...
[40:21.830 --> 40:22.570]  I don't...
[40:22.570 --> 40:34.750]  I don't know if this is answering your question, but one of the things that's really interesting, especially in HBO in the earlier years of streaming, is they would have zero hour downloads with correct subtitles in many different languages.
[40:34.750 --> 40:37.070]  Don't tell me that isn't leaked officially.
[40:37.850 --> 40:38.670]  Like, for sure.
[40:39.350 --> 40:40.590]  Like, zero...
[40:40.590 --> 40:43.570]  Zero hour leak with multiple language subtitles is perfect.
[40:44.130 --> 40:45.730]  That no torrent team can do that.
[40:45.910 --> 40:48.110]  Not even the best Chinese sub teams can do that.
[40:50.310 --> 40:51.170]  In the back.
[40:52.250 --> 40:56.470]  The pirates in the audience, I'm sure we're hoping for source code, but...
[40:57.490 --> 40:58.970]  Email me, this is GPL.
[40:59.270 --> 40:59.910]  Oh, yes.
[41:00.810 --> 41:10.710]  If you don't mind talking about the BitTorrent protocol, and where are these samples in the protocol we can find to reproduce this?
[41:11.830 --> 41:14.450]  I use LibTorrent, which is GPL software.
[41:14.790 --> 41:16.730]  All the software for L60 is GPL.
[41:17.550 --> 41:18.970]  You could read the sources.
[41:20.030 --> 41:21.230]  And just...
[41:21.230 --> 41:23.030]  That was probably the best answer for you.
[41:23.510 --> 41:23.590]  Yeah.
[41:24.270 --> 41:24.330]  Yeah.
[41:25.410 --> 41:26.610]  And the product with the mask.
[41:26.790 --> 41:28.710]  How do you anonymize your results?
[41:28.710 --> 41:29.250]  Yeah.
[41:29.250 --> 41:31.230]  We strip the last octet and we don't release IPs.
[41:33.970 --> 41:34.430]  Yeah.
[41:35.390 --> 41:38.210]  What percent of pirates do you see are using VPNs?
[41:38.310 --> 41:43.010]  And also, what countries do you see are, like, the most generous viewers?
[41:43.850 --> 41:46.130]  Oh, those are two great questions.
[41:48.090 --> 41:52.870]  The VPN data is really interesting, and I don't have it at the top of my...
[41:54.610 --> 41:57.630]  That's not something I have paged in right now.
[41:57.630 --> 42:04.230]  But I would say that they're at shockingly low VPN usage in Russia.
[42:04.350 --> 42:07.030]  This may actually be because of the network blocks.
[42:07.290 --> 42:08.150]  Very high.
[42:08.470 --> 42:11.690]  Much higher than expected VPN usage in the United States.
[42:12.550 --> 42:13.670]  And in...
[42:15.170 --> 42:21.370]  It's, like, Germany, Netherlands, France are all pretty high on that.
[42:22.230 --> 42:23.630]  We've got five minutes left.
[42:24.010 --> 42:27.050]  Also, which countries do you see are, like, the most generous viewers?
[42:29.370 --> 42:30.870]  It depends on the media text.
[42:31.110 --> 42:31.770]  We do...
[42:31.770 --> 42:42.530]  You could look at our GitHub, and we do track the seeders, as I showed a lot of data about downloading, but we do also have these rankings of countries and cities for the seeding as well.
[42:42.530 --> 42:54.630]  I guess one of the things we were surprised to find is that everybody is kind of, like, thinking that the peer-to-peer network is very moochy, but we were surprised to see the percentage of seeders.
[42:54.850 --> 42:56.750]  We're looking at, like, 23%.
[42:56.750 --> 42:57.690]  So...
[42:57.690 --> 42:59.090]  Really pretty high.
[42:59.790 --> 43:00.330]  Right there.
[43:00.550 --> 43:00.830]  Yeah.
[43:01.570 --> 43:05.550]  Yeah, so, I deal with Wilson data, and I know that it...
[43:07.830 --> 43:09.630]  You're stuck with five weeks.
[43:09.890 --> 43:10.030]  Yeah.
[43:10.690 --> 43:11.710]  Yeah, it's not good.
[43:11.850 --> 43:15.670]  But I'm just wondering if you've ever had the chance to compare your data to Nielsen?
[43:15.910 --> 43:17.790]  Yes, we didn't have time to do that.
[43:17.930 --> 43:18.430]  Let's talk about it.
[43:18.730 --> 43:21.470]  Yeah, we did a special data analysis.
[43:21.850 --> 43:23.750]  We knocked down to five weeks.
[43:25.110 --> 43:31.290]  And one of the things that we find really interesting is that, for film, we have a box office mojo.
[43:31.490 --> 43:35.570]  We have a little bit more established ideas of popularity.
[43:35.570 --> 43:41.290]  And so, the pirate numbers we're seeing are very similar to box office.
[43:41.570 --> 43:43.010]  So, we correlate very well with box office.
[43:43.430 --> 43:45.990]  Nielsen is often made of junk, frankly.
[43:46.330 --> 43:46.610]  Yeah.
[43:46.610 --> 43:49.890]  And it was barred by all the streamers for many years.
[43:50.070 --> 43:51.670]  They just kind of recently got into it.
[43:51.710 --> 43:52.950]  I think their methodology is...
[43:52.950 --> 43:56.670]  Well, I'm not going to say, because I don't want to get into legal issues.
[43:56.670 --> 44:01.210]  But, yeah, I would be suspicious of Nielsen, for sure.
[44:01.330 --> 44:02.770]  Yeah, I'm highly suspicious of Nielsen.
[44:03.130 --> 44:03.450]  Oh, okay.
[44:03.470 --> 44:03.650]  Yeah.
[44:04.290 --> 44:04.770]  Right there.
[44:04.990 --> 44:06.030]  How do you...
[44:06.030 --> 44:06.990]  Where do you...
[44:06.990 --> 44:07.670]  These are actually related.
[44:07.830 --> 44:13.710]  Where do you source your torrents, and how do you choose what you will be evaluating so that's representative?
[44:14.650 --> 44:16.050]  Yeah, I don't know if it's representative.
[44:16.670 --> 44:16.970]  Yeah.
[44:16.970 --> 44:20.030]  And we use a combination of top-sides and DHT shockers.
[44:20.810 --> 44:21.310]  Mm-hmm.
[44:21.990 --> 44:26.050]  And what we sample is often based on whatever...
[44:26.650 --> 44:29.550]  I don't know, whatever we think might be important to sample.
[44:29.850 --> 44:34.010]  So, there's not a standardized methodology from 2018 till 2025.
[44:35.310 --> 44:39.510]  But, like, you know, our first target was Game of Thrones final season.
[44:39.770 --> 44:45.350]  So, if we think in advance that anything's going to blow up the Internet, we try to sample that.
[44:45.350 --> 44:46.730]  But we also try to sample...
[44:46.730 --> 44:51.470]  I mean, we also do sample a lot of women-led, minority-led, queer-led texts.
[44:51.630 --> 44:55.110]  Because there's not a lot of data on those texts.
[44:55.270 --> 45:01.670]  And so, we want to just, like, make sure they're represented in our piracy rating system.
[45:02.270 --> 45:04.030]  But then we also love sci-fi.
[45:04.350 --> 45:05.110]  So, then we do...
[45:05.110 --> 45:08.230]  I think we kind of oversample in the sci-fi genre.
[45:08.670 --> 45:10.370]  Because that's just, like, what we're watching.
[45:10.730 --> 45:13.670]  So, I think some of it is just, like, we like Wednesday.
[45:13.670 --> 45:15.370]  We like only murders.
[45:15.370 --> 45:16.650]  So, we're going to sample that.
[45:16.770 --> 45:18.790]  So, there's not really a standardized method.
[45:18.930 --> 45:26.970]  But I think if this project grows, we are going to have to come up with a more standardized methodology for picking.
[45:27.290 --> 45:29.010]  So, I think that's a really good question.
[45:29.010 --> 45:34.710]  One of the real questions was, hey, there's an internalized racism in the U.S.
[45:35.150 --> 45:36.110]  media production industry.
[45:36.110 --> 45:40.850]  Where they think that black-led texts can't be popular outside the United States.
[45:40.970 --> 45:41.970]  And that's total bullshit.
[45:42.330 --> 45:47.450]  So, we were specifically sampling to disprove that.
[45:47.670 --> 45:50.810]  Because we think that's spunk and we want to move on from that.
[45:50.810 --> 45:51.730]  Yeah.
[45:52.710 --> 45:53.750]  Two questions.
[45:54.170 --> 45:54.930]  Do you...
[45:54.930 --> 46:01.390]  Have you looked at the data that's, like, immediately following, I don't know, like, Netflix subscription price increase?
[46:01.710 --> 46:02.190]  Yep.
[46:02.190 --> 46:03.170]  That's the question.
[46:04.210 --> 46:06.130]  That would be interesting.
[46:06.390 --> 46:08.470]  And we have noticed a huge...
[46:08.470 --> 46:13.470]  We have noticed a steady increase over the last couple of years, which is because as the streaming prices increase.
[46:13.630 --> 46:16.930]  But it's a little weird because there's also things like the Ukraine war.
[46:17.310 --> 46:20.970]  And that massively influenced kind of the behavior we're seeing.
[46:21.410 --> 46:24.290]  I had a second question that was completely out of the direction.
[46:24.430 --> 46:28.450]  So, you said that the data you release from Angular IPs and it's anonymized...
[46:28.450 --> 46:30.430]  Yeah, we don't ever release any IPs.
[46:30.430 --> 46:38.390]  So, with law enforcement, certain texts in certain places in the U.S. are illegal for children and stuff like that.
[46:39.050 --> 46:41.910]  And law enforcement can go from IP to...
[46:41.910 --> 46:43.230]  Who was that that downloaded?
[46:43.730 --> 46:44.850]  Do you...
[46:44.850 --> 46:48.090]  Have you had historically any encapsulated law enforcement?
[46:48.750 --> 46:52.330]  And do you, like, consider that when you store data internally?
[46:53.090 --> 46:54.930]  Well, we do strip the last octet.
[46:55.070 --> 46:57.850]  So, we do consider it when we store internally.
[46:57.850 --> 47:02.150]  But we have not had the joy of dealing with that.
[47:02.650 --> 47:02.990]  Yeah.
[47:03.190 --> 47:03.990]  And we do...
[47:03.990 --> 47:05.530]  We do...
[47:06.950 --> 47:09.210]  We obey takedowns.
[47:09.430 --> 47:13.530]  So, there's automated content takedown services.
[47:13.850 --> 47:17.690]  And we have 24 hours to take them off our server or else we're null routed.
[47:17.870 --> 47:20.710]  So, we're really serious about that in general.
[47:21.110 --> 47:21.250]  Yeah.
[47:22.050 --> 47:27.490]  We have been asked that question by a person who I strongly suspected.
[47:27.490 --> 47:30.990]  had a state affiliation in a foreign country.
[47:31.810 --> 47:34.630]  And that pertained to the 3D printed guns.
[47:34.990 --> 47:38.250]  So, he didn't press the issue with me.
[47:38.490 --> 47:52.070]  But I think we will have to formulate some kind of response for that query of, you know, like the query was, could you just release the IP addresses just to law enforcement?
[47:52.070 --> 47:53.550]  You know, that was the question.
[47:53.550 --> 47:55.290]  And I think that's a good...
[47:55.290 --> 47:58.670]  That's a question that you're correct to keep on our radar.
[47:59.910 --> 48:01.450]  I'm sorry, but...
[48:01.450 --> 48:01.910]  You're right to react.
[48:02.330 --> 48:02.350]  Okay.
[48:02.410 --> 48:02.630]  Awesome.
[48:02.750 --> 48:03.230]  Thank you so much.
[48:03.230 --> 48:03.590]  Thanks everybody.
[48:03.590 --> 48:03.790]  Thank you, everybody.
[48:04.010 --> 48:04.930]  Thank you, everybody.