[00:00.540 --> 00:02.980]  This is YouTomb, a free culture hack.
[00:03.700 --> 00:09.840]  One thing I've really enjoyed about The Last HOPE so far is there is an awesome emphasis on community.
[00:10.020 --> 00:11.220]  There's a lot of community going on.
[00:11.320 --> 00:15.740]  We, of course, have a physical community here in this room that comes together periodically.
[00:16.500 --> 00:19.400]  And there are a lot of virtual communities out there as well.
[00:19.680 --> 00:27.720]  And I'd like to congratulate everyone who has helped to make those communities and be part of those communities and also, of course, come here and talk about them.
[00:28.780 --> 00:30.300]  YouTomb, a free cultural hack.
[00:30.460 --> 00:31.040]  Panel discussion.
[00:40.370 --> 00:41.230]  Hi, guys.
[00:41.510 --> 00:42.690]  My name is Christina.
[00:42.950 --> 00:43.970]  I'm a senior at Harvard.
[00:44.030 --> 00:47.510]  And I'm with the Harvard Free Culture and the MIT Free Culture Group.
[00:47.730 --> 00:50.210]  And this is YouTomb, a free culture hack.
[00:50.530 --> 00:55.730]  So what YouTomb is, is basically it's a graveyard of YouTube videos.
[00:55.890 --> 00:59.610]  It's where we track YouTube videos that are taken down for any reason.
[00:59.610 --> 01:03.430]  But we sort of focus on the ones that are taken down for copyright complaints.
[01:04.170 --> 01:11.330]  As you can see here, the left side of the site is where the most recent videos that are taken down for copyright go.
[01:11.630 --> 01:16.270]  On the right side, you can see all the videos that we're scanning, you know, with the latest ones first.
[01:17.110 --> 01:18.530]  And yeah, that's the site.
[01:21.310 --> 01:21.710]  Okay.
[01:21.950 --> 01:25.770]  And just so you know, Christina and I are going to be kind of tag teaming the first part of this talk.
[01:25.950 --> 01:26.750]  I'm Dean Jansen.
[01:27.370 --> 01:32.370]  I'm also part of the free culture movement and have been involved with free culture at Harvard and MIT for a while now.
[01:34.810 --> 01:43.570]  So we're going to start with kind of the history of the project, why we put together YouTomb, what we wanted people to get out of it, what we wanted to get out of it.
[01:43.690 --> 01:46.770]  And then we're going to let Quentin talk a little bit about the back end.
[01:46.910 --> 01:48.370]  He's one of the programmers behind it.
[01:48.970 --> 01:53.470]  And Oliver has been doing a lot of really interesting research on some of the data that we've come up with.
[01:55.810 --> 02:02.550]  So this post came up on the Google blog last October, and a friend sent it to me.
[02:02.730 --> 02:06.070]  And as soon as I saw it, some alarm bells went off.
[02:06.170 --> 02:17.450]  Basically, they're saying that YouTube and Google are going to be using kind of an algorithm full of, like, video...
[02:17.450 --> 02:18.190]  Media...
[02:18.190 --> 02:19.470]  Excuse me.
[02:20.330 --> 02:20.830]  A bunch...
[02:20.830 --> 02:21.090]  Like...
[02:21.090 --> 02:22.350]  Automatic filtering.
[02:22.350 --> 02:22.710]  Yes.
[02:23.110 --> 02:30.850]  So they're going to be filtering videos automatically based on a thumb printing system of a database of things that copyright holders have given to them.
[02:31.030 --> 02:36.510]  If YouTube videos have matches, then the copyright holders will be notified.
[02:37.050 --> 02:41.470]  And the quotes I pulled out here, video identification goes above and beyond our legal responsibilities.
[02:41.910 --> 02:44.890]  It will help copyright holders identify their workers on YouTube.
[02:45.150 --> 02:52.330]  That sort of set off some alarm bells in my head because really Google is using machines to decide what is and isn't fair use.
[02:52.330 --> 03:01.870]  And for anyone who's not familiar with fair use, that's a part of the copyright law that allows us to actually use a copyrighted work without the permission of the copyright holder.
[03:02.190 --> 03:09.010]  And we can use it in specific instances where we're criticizing things or where we're making commentary on something.
[03:09.010 --> 03:12.230]  And so, as soon as I found this out, I was like, what the crap?
[03:12.410 --> 03:14.290]  Like, this is not cool at all.
[03:15.710 --> 03:23.090]  And so, this slide just shows sort of why I think I was validated in being a little bit nervous.
[03:23.730 --> 03:26.470]  Some of you may recognize the clouds here.
[03:26.590 --> 03:30.090]  They were used in the anonymous message to Scientology video.
[03:31.730 --> 03:38.930]  The original video was actually just a time-lapse video that some guy shot in his backyard and put into the public domain.
[03:39.110 --> 03:44.850]  He posted it on both Vimeo, which is what you can see above, that's a video site, and he also posted it to Google Video.
[03:44.850 --> 03:55.950]  And interestingly enough, the anonymous faction used that footage in their message to Scientology.
[03:56.250 --> 04:00.630]  And then, later on, the Huffington Post on their, what was it?
[04:01.230 --> 04:10.670]  23.6 comedy website, Ariana Huffington, those folks used some of this footage in a spoof of the message to Scientology video.
[04:10.670 --> 04:26.490]  And then, Google actually sent BrownPow, the original creator, a takedown notice saying, hey, we've taken your stuff off of Google Video because it uses some copyrighted material from this thing that Huffington had put up.
[04:26.650 --> 04:31.270]  So, basically, their automatic scanner had pulled this thing out, said, take this down.
[04:31.390 --> 04:40.250]  And then, I think Huffington, they allowed it, but still, this person was really upset because the automated system had just taken his stuff down.
[04:40.250 --> 04:44.310]  And assumed that it was a derivative work of something that was actually a derivative work of his.
[04:44.450 --> 04:46.230]  So, that's just kind of an inherent problem.
[04:47.810 --> 04:52.230]  So, I immediately thought of the Google Chilling Effects project, which is...
[04:53.790 --> 04:57.170]  It started out as an independent project where they would...
[04:57.170 --> 05:05.970]  The EFF and some other people were watching search results that were taken out of the Google system due to copyright or other reasons.
[05:05.970 --> 05:07.310]  I think obscenity is one of them.
[05:08.150 --> 05:09.970]  And then, they highlight what's been taken out.
[05:10.070 --> 05:12.810]  Now, they're cooperating with Google, but it's sort of a similar project.
[05:13.070 --> 05:16.630]  So, in realizing all this, I tossed up the bat signal.
[05:18.230 --> 05:19.030]  And then...
[05:19.550 --> 05:20.750]  I'll let you take over here.
[05:22.590 --> 05:22.990]  Yeah.
[05:23.290 --> 05:24.630]  So, Dean threw up the bat signal.
[05:24.870 --> 05:28.190]  And here, this is where the bat cave is, the undisclosed location.
[05:28.770 --> 05:30.530]  It's at Harvard, apparently.
[05:31.550 --> 05:42.950]  Dean and I are both members of Harvard College Free Culture, which is an organization that focuses on basically the intersection between the digital world and art and the copyright and all of that good stuff.
[05:43.530 --> 05:46.750]  We're basically the free software movement for everything that isn't software.
[05:48.010 --> 05:49.410]  And so, he brought it to our attention.
[05:49.570 --> 05:50.810]  This was at the beginning of the school year.
[05:50.890 --> 05:52.730]  So, we were like, yeah, that's a really interesting project.
[05:53.010 --> 05:55.150]  We have a ton of other interesting projects.
[05:56.170 --> 06:00.830]  And then we basically just sat on our asses for a while and didn't do anything, which sucks.
[06:00.950 --> 06:04.610]  Until the war came home, so to speak.
[06:04.990 --> 06:08.350]  This is our friend Kevin Driscoll in a very tripped out picture.
[06:09.070 --> 06:11.690]  He's a teacher, a DJ, an awesome dude.
[06:11.730 --> 06:16.810]  And he was totally silenced by this really unfortunate automatic takedown thing.
[06:17.170 --> 06:21.710]  So, Kevin's a lifelong native of Boston and therefore a big Red Sox fan.
[06:21.710 --> 06:22.870]  I'm sorry if that's not kosher here.
[06:24.650 --> 06:29.930]  And he put up this video, which was taken on his Motorola Razr.
[06:30.050 --> 06:35.950]  And it's only 15 seconds long of the last pitch of the World Series last year.
[06:36.150 --> 06:38.630]  As you can see, the quality is absolute shit.
[06:38.830 --> 06:45.330]  And it's about seven seconds of the video followed by eight seconds of shaking because him and all his friends are jumping up and down.
[06:45.710 --> 06:50.050]  But within three days of the video going up, he saw this.
[06:50.670 --> 07:03.570]  This video is no longer available due to a copyright claim by MLB Advanced Media because apparently they were afraid that people were going to watch this shitty seven-second clip from a cell phone instead of watching their actual DVD or whatever it is that they were selling.
[07:04.370 --> 07:07.670]  So Kevin forwarded this incident to us.
[07:07.930 --> 07:10.850]  And Harvard Free Culture as a whole was like, Aw, hell no.
[07:11.090 --> 07:12.450]  You did not just go there.
[07:14.090 --> 07:18.270]  So we reconvened in the Batcave and we are like, Alright, we have to do something.
[07:18.670 --> 07:19.710]  Let's build this tracker.
[07:20.290 --> 07:23.350]  We are going to track all the YouTube videos and see when they go down.
[07:23.950 --> 07:26.090]  And oh shit, none of us can program.
[07:27.010 --> 07:27.330]  Oops.
[07:28.930 --> 07:29.310]  Yeah.
[07:29.310 --> 07:43.750]  So that is when I decided to take a pilgrimage down Massab about a mile down to the friendly tech institute down the street which had recently started their own free culture chapter and were in very close cohorts with us.
[07:44.190 --> 07:46.510]  And I was basically like, Guys, we have this awesome project.
[07:46.790 --> 07:48.550]  We can't code because we suck.
[07:48.810 --> 07:50.170]  Can you guys code?
[07:50.330 --> 07:51.770]  And they were like, Yeah, this sounds really awesome.
[07:52.070 --> 07:53.070]  Let's make a project.
[07:53.070 --> 07:54.630]  And so we started throwing out ideas.
[07:54.950 --> 08:01.530]  And before the end of the day, we had come up with like a reasonable plan and the project YouTube was born.
[08:03.930 --> 08:07.350]  And so these are some of the folks that were working on said project.
[08:08.250 --> 08:12.470]  And so really we just waited for about maybe two weeks or something like that.
[08:12.730 --> 08:15.650]  And Quentin is going to be talking about the back end.
[08:15.730 --> 08:17.050]  This is sort of just a crappy photo.
[08:17.210 --> 08:19.270]  But all of a sudden we had a working back end.
[08:19.350 --> 08:20.010]  There was a scanner.
[08:20.170 --> 08:23.050]  There was a database of videos that we were watching.
[08:23.050 --> 08:25.410]  And so we were all really psyched about that.
[08:27.350 --> 08:29.850]  YouTube doesn't actually scan all the videos on YouTube.
[08:30.370 --> 08:32.250]  We're not that Herculean right now.
[08:33.050 --> 08:35.230]  So we don't have a representative data set.
[08:35.370 --> 08:37.430]  So we felt like we weren't just...
[08:37.430 --> 08:39.690]  We didn't want this just to be a research project.
[08:39.810 --> 08:40.590]  We wanted it to be more.
[08:40.670 --> 08:49.810]  We wanted it to be also a form of activism and a form of reaching out to people and saying, Hey, you know when you click the video and it says, This video is no longer available.
[08:49.870 --> 08:50.750]  Why is that?
[08:50.750 --> 08:52.810]  And we wanted people to sort of see...
[08:52.810 --> 08:55.430]  Get a little more information about the video.
[08:55.630 --> 09:03.390]  So we put together kind of a colorful front end that we thought was a good representation of how things are getting taken down.
[09:03.610 --> 09:05.430]  And that there's lots of...
[09:05.430 --> 09:10.970]  We sort of made some design choices that we thought made the project more accessible to more people.
[09:10.970 --> 09:22.510]  So we could have put up more data, but we chose to make it sort of simple and keep it kind of, I don't know, not brief, but make it very digestible and put sort of the mission in a very focused way on the front page.
[09:24.790 --> 09:31.790]  And I just wanted to kind of like push on that for a second because I think that's something that's important.
[09:31.790 --> 09:34.030]  Like having...
[09:34.510 --> 09:37.250]  There are so many projects that are so awesome out there.
[09:37.530 --> 09:39.290]  Probably much cooler than what we're doing.
[09:39.770 --> 09:45.550]  But because of the simplicity and the way that we sort of pitch this to people, we were able to get...
[09:45.550 --> 09:47.650]  We got a lot of press coverage on this.
[09:47.790 --> 09:54.590]  And it's something that a lot of projects don't get, but that they should get and that they could get if they were maybe a little bit more...
[09:54.590 --> 09:55.890]  If they were...
[09:56.790 --> 10:01.210]  If they pitched their mission to people and they pitched what they were doing in a little bit more concise way.
[10:01.350 --> 10:09.690]  So I just wanted to kind of highlight that as something that we really consciously focused on from the front side of things.
[10:10.210 --> 10:17.330]  Yeah, basically the magical thing about this slide is that we didn't send press releases to anyone.
[10:17.730 --> 10:18.690]  We actually like...
[10:18.690 --> 10:26.990]  I don't really know how we got picked up, but basically Gizmodo I think found us and then everyone else found us and like almost we...
[10:26.990 --> 10:30.510]  It did crash our servers for about five minutes and it was bad.
[10:30.670 --> 10:33.770]  It was on like CNN, Wolf Blitzer, and you know, it was pretty ridiculous.
[10:34.090 --> 10:35.390]  But yeah, the sort of...
[10:35.390 --> 10:48.690]  The thing that we really want to get across is we think the reason why we were picked up is not just because this is a cool project, because there's millions of cool projects like Dean said, but it's because it's a cool project that's really easy for like mainstream audiences to understand.
[10:49.010 --> 11:02.190]  It's a project that, you know, like the name is cutesy or whatever, but you know, it's very apparent immediately like what it is and therefore newspaper people are always looking for easy stories like that to tell people.
[11:02.370 --> 11:08.010]  So if you can make your project accessible enough for reporters, they're going to help you get the audience that you want.
[11:09.830 --> 11:10.790]  Okay, so...
[11:11.290 --> 11:16.310]  And now we're going to hand things over to Quentin, who is one of the folks who put together the back end.
[11:20.320 --> 11:23.020]  All right, so I am a programmer at MIT.
[11:23.420 --> 11:25.180]  I'm an EECS major.
[11:25.320 --> 11:26.720]  I'm a rising junior now.
[11:26.980 --> 11:30.740]  And so I was part of the MIT free culture movement.
[11:30.900 --> 11:37.380]  And these guys from Harvard came over and said, you know, we have this really great idea, but we don't know how to accomplish this.
[11:38.300 --> 11:40.220]  Can you help us come up with a plan?
[11:40.520 --> 11:43.960]  And so we came up with some objectives.
[11:43.960 --> 11:49.820]  What we wanted to do was we wanted to keep track of every video on YouTube and know instantly when they become inaccessible.
[11:50.200 --> 11:51.820]  And this is really hard.
[11:53.480 --> 11:57.080]  YouTube really doesn't even want people to know how many videos are on YouTube.
[11:57.300 --> 12:02.260]  Not even, you know, information about whether they're up or down or anything like that.
[12:02.700 --> 12:04.700]  So we can't really do that.
[12:06.540 --> 12:11.760]  And YouTube doesn't provide a feed of saying, you know, these are all of the takedowns that happened today.
[12:12.320 --> 12:20.120]  They do for Chilling Effects now, which is a site that aggregates the things that are taken down from Google search results.
[12:20.980 --> 12:22.220]  But they don't for YouTube.
[12:22.700 --> 12:27.680]  And part of the reason is that YouTube is currently embroiled in a lawsuit with Viacom, as you may know.
[12:28.420 --> 12:30.400]  And so they're kind of secretive about this.
[12:30.480 --> 12:34.680]  So we couldn't exactly do that, but I'll talk in a minute about what we can do.
[12:35.840 --> 12:41.840]  So as a reminder, the only way you find out about a video being down is that you get a page that looks like this.
[12:41.980 --> 12:45.140]  It says the video is unavailable due to a copyright claim by Foo.
[12:45.700 --> 12:48.360]  And you only get that if you look at the video page.
[12:48.960 --> 12:50.600]  And there's no feed.
[12:50.820 --> 12:53.120]  There's no, like, list of the takedowns that happened today.
[12:53.540 --> 12:58.460]  And in fact, even takedown... videos that have been taken down even show up on other pages on YouTube.
[12:58.940 --> 13:01.740]  You can still see them in search results for a little while.
[13:01.820 --> 13:04.140]  You can still see them on a homepage if they were popular.
[13:05.160 --> 13:11.680]  The data feeds still show the videos for minutes or hours after the video is taken down.
[13:12.180 --> 13:17.500]  So the only way you see it is on the actual page when you try to look at the video, you get, you know, this video has been taken down.
[13:19.280 --> 13:23.740]  So given that requirement, we sort of had to modify our objectives.
[13:24.300 --> 13:28.260]  And we had to say, well, let's keep track of all of the popular and interesting videos.
[13:28.260 --> 13:33.540]  And when we say interesting, we don't necessarily mean, you know, what people like.
[13:33.740 --> 13:36.580]  We mean what is currently being discussed.
[13:36.820 --> 13:40.680]  And we sort of tried to predict what we think people...
[13:40.680 --> 13:43.380]  what we think would be likely to be taken down.
[13:43.540 --> 13:45.360]  So we think what would be likely to be noticed.
[13:46.160 --> 13:48.180]  And we can't know immediately when they're taken down.
[13:48.380 --> 13:54.260]  But what we can do is we can poll periodically and find out as soon as we can when they've been taken down.
[13:55.320 --> 13:57.060]  So how did we accomplish this?
[13:57.060 --> 14:01.120]  We wrote a whole bunch of modular components that can run on any machine.
[14:01.320 --> 14:03.320]  They currently all run on one machine, but they don't have to.
[14:03.880 --> 14:12.680]  We designed this and are approaching the point where we can just, you know, have someone run a Python script on their machine and help with the scanning load for YouTube.
[14:13.040 --> 14:16.080]  So we have these three different types of modular components.
[14:16.260 --> 14:20.360]  We have explorers which go out on the web and look for the interesting videos.
[14:20.560 --> 14:23.680]  And they use various heuristics for determining what are the interesting videos.
[14:24.520 --> 14:34.240]  We have a set of scanners which are responsible for looking on YouTube site and finding out the status of the videos, the metadata, number of views, title, you know, when things change.
[14:34.820 --> 14:41.040]  And we also have a set of archivers which cache the media related to those videos.
[14:41.280 --> 14:51.300]  And it's strikethrough, and I'll explain why in a minute, but it's strikethrough because that isn't all quite implemented and available to the public yet, and I'll explain why.
[14:53.020 --> 14:59.540]  So we wrote a whole bunch of explorers that look all over the web for things that are interesting.
[14:59.880 --> 15:03.580]  We wrote a dig explorer which watches every video that anyone posts to dig ever.
[15:04.720 --> 15:14.900]  We don't... we watch the feed that dig provides that you can refresh up to once a minute, I think, that just gives you every single post that someone puts up regardless of how many digs it has.
[15:15.120 --> 15:16.260]  So that's a really cool feed.
[15:16.260 --> 15:18.040]  We get a lot of interesting videos off of there.
[15:18.600 --> 15:23.820]  We watch the Technorati video page, which is sort of the videos that are being talked about in the blogosphere.
[15:24.120 --> 15:29.480]  They watch a whole bunch of RSS feeds and find out, you know, what videos are popular, et cetera.
[15:30.460 --> 15:41.540]  And we wrote a whole bunch of different YouTube explorers which look at the various lists on YouTube of most popular, most viewed today, most, you know, top rated, all of those different lists.
[15:42.060 --> 15:46.360]  And so these explorers collect a whole bunch of lists of...
[15:46.360 --> 15:49.360]  Basically, we just take the video ID out of, you know, the URL.
[15:49.360 --> 15:55.440]  If you ever see a YouTube URL, it's youtube.com slash watch question V equals and then this like... what is it?
[15:55.500 --> 15:56.340]  12 character string.
[15:56.460 --> 15:57.940]  That's the video ID.
[15:58.640 --> 16:00.800]  And we just collect a whole bunch of video IDs.
[16:01.180 --> 16:06.980]  And then we have a set of scanner modules that look at this list of video IDs.
[16:06.980 --> 16:09.680]  We call these artifacts, the different videos we're looking at.
[16:09.680 --> 16:15.940]  So we have the least recently scanned scanner, which just sequentially scans all of the videos that aren't down.
[16:16.880 --> 16:19.220]  And we just keep going through and through the list.
[16:19.400 --> 16:24.620]  And if you look on our website on the statistics page, we actually show you how long it takes us to get through the list of videos.
[16:24.900 --> 16:31.280]  We have a whole bunch of threads, and we try and just, you know, get information about every video as quickly as we can.
[16:32.480 --> 16:36.880]  Unfortunately, the only YouTube GData API is get info about one video.
[16:36.880 --> 16:38.840]  So we do that a bunch of times.
[16:40.560 --> 16:49.960]  And then we also didn't originally think we would need this, but it turned out that we actually needed to write a dead scanner, a scanner that would scan videos that we had seen as dead.
[16:50.460 --> 16:54.640]  And it turns out that in more cases than you would think, those videos come back.
[16:54.840 --> 16:57.720]  And this is copyright takedowns that have been put back.
[16:58.020 --> 17:04.160]  But it's not all copyright takedowns, and we don't quite know why videos that are down suddenly become available again.
[17:05.400 --> 17:10.200]  And we'll talk a little bit more about some of the takedowns a little bit later.
[17:11.560 --> 17:18.080]  And so then we have some archivers, and the archivers are responsible for saving the large media chunks related to these things.
[17:18.080 --> 17:19.840]  So we have a thumbnail archiver...
[17:24.220 --> 17:26.080]  Let me just get...
[17:26.620 --> 17:26.920]  What?
[17:28.560 --> 17:29.100]  There.
[17:30.100 --> 17:32.380]  Someone was trying to reach my firewall.
[17:33.460 --> 17:34.480]  So we had...
[17:34.480 --> 17:37.900]  So the little snitch said, do you want to allow this?
[17:37.980 --> 17:38.460]  And I said no.
[17:39.400 --> 17:42.240]  So we have a thumbnail archiver that saves video thumbnails.
[17:42.920 --> 17:49.620]  And the reason we do this is that the thumbnails disappear, you know, a few minutes or hours after the video gets taken down, like the rest of the content on YouTube.
[17:50.400 --> 17:54.700]  And so we need to save them so that we can show them on our site so people know sort of what we're talking about.
[17:54.980 --> 18:03.680]  Our original site didn't have thumbnails, and people were like, well, why is there this video that was one of the most popular videos on YouTube called XXX porn, and it's taken down now?
[18:04.560 --> 18:10.340]  By the way, that was because it was actually a clip of My Little Pony, which was a TV show from like 10 years ago.
[18:10.460 --> 18:11.800]  Has anyone seen that video on YouTube?
[18:12.820 --> 18:13.900]  Yeah, exactly.
[18:14.620 --> 18:16.820]  We were like, why is this video taken down?
[18:17.560 --> 18:20.180]  But yeah, so that was one of the most popular videos.
[18:20.300 --> 18:22.120]  It was on the YouTube most popular list.
[18:22.220 --> 18:24.480]  It had, you know, 100 million views or something like that.
[18:25.220 --> 18:36.020]  So we do that, and we save these thumbnails, and we show them under fair use, because we say this is, you know, this is commentary about YouTube and about this video.
[18:36.220 --> 18:41.100]  We need to give people, you know, something that's representative of the video, but not confused with the video.
[18:41.100 --> 18:41.860]  It's just a thumbnail.
[18:42.220 --> 18:47.580]  And then we also haven't fully implemented a video archiver, which would actually save the video files for later review.
[18:48.220 --> 18:56.900]  And we're still working with lawyers right now to try and figure out just how far we can go there, whether we can actually show videos that YouTube says are copyright infringements.
[18:58.420 --> 19:10.120]  And it's important to note with the DMCA that you're only required to take something down if you actually get a legitimate DMCA request that follows the requirements in the DMCA law, and then you are required to take it down.
[19:10.340 --> 19:15.140]  You are not required to take it down just because someone says that video is a copyright violation.
[19:15.320 --> 19:17.200]  They need to follow a very specific format.
[19:17.360 --> 19:21.520]  So it's not clear whether we can actually say...
[19:21.520 --> 19:27.340]  whether we can show videos when some other site has taken it down because they say that they received a takedown notice.
[19:27.340 --> 19:29.700]  So we're still working with lawyers on that.
[19:30.440 --> 19:36.740]  But anyway, so all of these modules, they collect a whole bunch of information and they stuff it into a MySQL database.
[19:37.580 --> 19:42.040]  Not because MySQL is the best database, but because it's what we had freely available at MIT.
[19:42.840 --> 19:45.120]  So we stuff everything in a MySQL database.
[19:45.320 --> 19:49.860]  As of last night, we had about 282,000 rows in the artifacts table.
[19:50.020 --> 19:55.700]  So we're watching, you know, a quarter million videos, which is not all of YouTube by any stretch of the imagination.
[19:55.700 --> 20:04.420]  But we get a nice wide swath of the videos that are on YouTube and hopefully a somewhat representative sample, although it's not statistically a representative sample.
[20:04.640 --> 20:07.620]  So the artifacts take up, you know, a couple hundred megabytes.
[20:07.820 --> 20:12.280]  And then we save every piece of information we have ever collected from YouTube, ever.
[20:12.980 --> 20:14.860]  We currently have about 70,000...
[20:15.260 --> 20:17.820]  70 million rows in 32 gigs.
[20:18.040 --> 20:21.840]  And this is like 5 or 10 snapshots per second of different videos.
[20:22.100 --> 20:24.760]  We keep every piece of information we've ever collected.
[20:24.760 --> 20:30.180]  The idea being that it is going to be useful for someone to be able to go and mine that information.
[20:30.480 --> 20:37.960]  We don't know for sure now what's going to be interesting, but we need to collect it now because if the video gets taken down, we can't see all that information.
[20:38.640 --> 20:45.440]  If you want to look at this, we have a public MySQL database that's just set aside for anonymous access.
[20:46.280 --> 20:52.180]  And you can just check out our blog there for information about how to access our MySQL database and perform queries against our data.
[20:53.040 --> 20:55.940]  If you feel too constricted by the web interface.
[20:56.600 --> 20:58.740]  So that sort of is how we do this.
[20:59.220 --> 21:00.420]  So it's all Python.
[21:01.500 --> 21:06.680]  It's, you know, we have a whole bunch of different modules that are completely independent except for sharing a database.
[21:08.200 --> 21:12.020]  And the web interface is a Python cherry pie app that sits on top of it.
[21:12.120 --> 21:13.760]  And it's all, you know, pretty modular.
[21:13.760 --> 21:17.440]  And it runs on a cluster of five servers at MIT.
[21:17.980 --> 21:20.400]  So that sort of is our technical architecture.
[21:20.680 --> 21:30.160]  I'm now going to turn it over to Oliver, who is another Harvard person, who sort of did some research on, you know, what does this data say?
[21:30.300 --> 21:33.000]  And one of the things we found was geolocational filtering.
[21:33.000 --> 21:36.940]  And he's sort of affiliated with, you know, ONI and stuff like that.
[21:37.140 --> 21:37.760]  So, Oliver.
[21:38.080 --> 21:38.660]  Thank you.
[21:40.480 --> 21:40.900]  Hello.
[21:40.900 --> 21:42.260]  My name's Oliver Day.
[21:42.600 --> 21:47.920]  And I got interested in this project because an email was passed to me.
[21:48.860 --> 21:54.660]  And while they were figuring out the XML format for the API, they found this really interesting tag.
[21:56.080 --> 22:06.320]  And if you look where the little arrow is pointing, you can see media restriction type equal country, relationship equal deny.
[22:07.200 --> 22:09.700]  And then it's usually a list of country codes.
[22:10.300 --> 22:13.780]  So, if you think about, like, the way this project is working, we're studying takedowns.
[22:13.800 --> 22:15.800]  This is, in effect, a soft takedown.
[22:15.880 --> 22:21.420]  This is saying, we are going to let everyone in the world see this video except for this list.
[22:21.900 --> 22:25.820]  And the first one we saw was actually China, and that's actually why I got so interested.
[22:26.020 --> 22:31.800]  The work at ONI that I do, we study the Great Firewall and what things are really pissing off the Chinese government.
[22:33.240 --> 22:35.360]  And I'll get to that in a few minutes.
[22:36.460 --> 22:39.580]  So, what it looks like if you're in one of these countries, right?
[22:39.680 --> 22:41.240]  So, this is geo IP, right?
[22:41.380 --> 22:43.360]  So, it will say your IP is from this country.
[22:44.080 --> 22:51.880]  And if your country equals that code, you will get this little pink bar that says this video is not available in your country.
[22:52.180 --> 22:57.900]  Now, obviously, anyone that has Tor or a proxy or anything can bypass this pretty quickly.
[22:59.200 --> 23:09.780]  But, anyway, YouTube has done this probably because it's keeping certain countries from black-holing them in BGP routing tables and turning it off for the entire world.
[23:10.140 --> 23:11.640]  Yeah, like Pakistan did.
[23:12.700 --> 23:23.320]  And so, you know, once we started collecting data on this, I went through the database and started trying to analyze what type of groupings there were as far as the country codes.
[23:23.320 --> 23:26.420]  And what we found was actually pretty interesting.
[23:26.600 --> 23:29.700]  Not all of the stuff you find makes sense at first.
[23:30.300 --> 23:33.640]  This video is about Tibet, right?
[23:33.780 --> 23:35.120]  This is the riots in Tibet.
[23:35.380 --> 23:37.740]  It was from someone in Japan.
[23:38.540 --> 23:44.960]  But it was blocked in Germany, France, and a bunch of stuff that's owned by France.
[23:45.820 --> 23:51.020]  And if you look at that middle thumbnail, you'll see there's actually a tiny swastika.
[23:51.020 --> 23:55.040]  And so, what had happened was someone had made a reference to the Nazis.
[23:55.780 --> 24:07.100]  And because hate speech is banned in Germany, France, Poland, and apparently all of France's territories, this is kind of the deal that was struck between YouTube and these countries.
[24:07.100 --> 24:14.740]  Is that we will go ahead and block these things from showing up in your country so you don't disable all of your citizens from being able to see YouTube.
[24:16.540 --> 24:17.980]  Some of it is ridiculous, right?
[24:18.100 --> 24:24.500]  This is an old Walt Disney cartoon of Donald Duck where he apparently turns into a Nazi for a few frames.
[24:25.840 --> 24:28.880]  And again, banned in Germany, France, Poland.
[24:29.120 --> 24:31.000]  And so this is what I've been calling the hate speech block.
[24:31.160 --> 24:33.880]  And now I can go through the database and look for this block.
[24:34.000 --> 24:36.740]  And every time I see that, I know it's some type of hate speech.
[24:36.740 --> 24:41.560]  Although, since they're not talking about it, we have to make all our own assumptions.
[24:41.900 --> 24:45.080]  And sometimes, I mean, it doesn't seem to make as much sense.
[24:45.220 --> 24:50.900]  I included this in case anyone wanted to know what the hell half of those two-letter country codes were.
[24:51.220 --> 24:52.720]  I had to look most of them up.
[24:52.760 --> 24:56.560]  And the Wikipedia articles are fascinating if you ever get around to it.
[24:57.860 --> 25:01.220]  So, the first really interesting one we found was in Thailand.
[25:01.660 --> 25:03.540]  Is anyone familiar with Les Majestes?
[25:03.640 --> 25:05.900]  I can't speak French, so I'm probably butchering that.
[25:06.720 --> 25:09.660]  So, Les Majestes is a law, right?
[25:09.780 --> 25:11.880]  And it says you cannot impugn the king.
[25:12.460 --> 25:16.580]  And if you could think about how ridiculous that would be if you went to jail for making fun of Bush.
[25:17.740 --> 25:24.900]  This law says, and there are reporters that have gone into jail for this, if you make fun of the king, you can go to jail for up to 15 years.
[25:25.260 --> 25:29.880]  And this is for any type of criticism of his policy or what he's wearing, anything.
[25:31.140 --> 25:38.560]  These videos basically caused Thailand to say, we're not allowing YouTube in our country anymore.
[25:39.780 --> 25:46.020]  And so, again, a deal was struck probably behind closed doors where they said, well, we have this thing in place.
[25:46.040 --> 25:49.060]  We can make sure that videos do not show up in Thailand.
[25:49.640 --> 25:57.740]  And it's interesting because we started watching the Gert Wilder videos when Pakistan blew up and said, screw you guys, we're turning off YouTube.
[25:58.240 --> 25:59.920]  And they haven't been banning.
[26:00.100 --> 26:04.560]  So, it's, again, very unclear why they make these decisions, when they make them.
[26:07.700 --> 26:09.380]  And then we have our country.
[26:09.980 --> 26:12.800]  So, is anyone familiar with Soldier Boy?
[26:13.400 --> 26:15.020]  A couple people.
[26:15.260 --> 26:15.380]  Okay.
[26:16.360 --> 26:19.720]  So, Soldier Boy signed recently, like a while ago, I guess.
[26:20.580 --> 26:22.860]  And so, we do things differently here, right?
[26:23.020 --> 26:28.600]  We can make fun of Bush, but you just cannot put shit up if somebody else has said it first, apparently.
[26:29.120 --> 26:34.060]  And so, these are all of the banned countries for Soldier Boy's video.
[26:34.460 --> 26:37.940]  Now, we don't think these countries are pissed off at Soldier Boy.
[26:37.940 --> 26:39.280]  We think that...
[26:39.280 --> 26:48.260]  I mean, our current theory is that since a lot of music videos are optioned by country, that's how they're doing this.
[26:48.340 --> 26:51.000]  And so, it's kind of like region encoding for DVDs.
[26:51.240 --> 26:58.380]  And I actually had to, like, create a really quick script that would tell me what countries you're actually allowed to see the video in.
[26:58.380 --> 27:04.120]  Because there's a little over 200 country codes available in the ISO standard.
[27:04.320 --> 27:06.640]  And I think that list is about 200.
[27:08.340 --> 27:10.360]  So, earlier I was talking about China.
[27:10.720 --> 27:15.720]  And interestingly, the only videos that are only blocked in China are from the NBA.
[27:18.040 --> 27:20.700]  We're really not sure what's going on with the NBA.
[27:21.420 --> 27:31.020]  We did find this really interesting story that the guy that used to work for Microsoft and was in charge of licensing in China was snagged by the NBA.
[27:31.840 --> 27:33.360]  And so, he works for them now.
[27:33.620 --> 27:35.860]  So, we think that might have something to do with this.
[27:36.060 --> 27:40.160]  There's a little over 450 videos right now that we're tracking that are all banned.
[27:40.260 --> 27:41.480]  And it's just China, right?
[27:41.560 --> 27:45.260]  Like, it's not like this where it's like a super set, right?
[27:45.380 --> 27:47.620]  Like, China's part of this huge list.
[27:47.740 --> 27:50.560]  By the way, you cannot see Soldier Boy in China.
[27:51.300 --> 27:55.680]  But in this case, it's literally just China that cannot view NBA videos.
[27:59.220 --> 28:03.420]  So, we're also on this list sometimes.
[28:04.280 --> 28:07.040]  These actually make... this is kind of our WTF list.
[28:07.180 --> 28:09.700]  We really don't understand what's on here.
[28:09.920 --> 28:12.760]  This stuff is banned in just the U.S.
[28:13.680 --> 28:15.380]  Brokeback Mountain, okay, maybe.
[28:15.620 --> 28:18.040]  But the Fresh Prince of Bel Air, we don't really get.
[28:18.720 --> 28:22.060]  Rick Rolling is back up now, but it was blocked in the U.S.
[28:22.180 --> 28:22.660]  for a while.
[28:24.920 --> 28:25.560]  You know...
[28:25.560 --> 28:25.980]  That's one of them.
[28:26.040 --> 28:26.200]  Quiet!
[28:26.460 --> 28:27.200]  Yeah, one of them.
[28:27.360 --> 28:27.680]  Yeah, yeah.
[28:27.780 --> 28:31.840]  I was going to Rick Roll the entire audience, but I didn't want to be like hit or pelted with things.
[28:34.500 --> 28:36.660]  So, again, you know, sometimes it's...
[28:37.200 --> 28:44.400]  And we haven't really worked on this yet, but for us, if we try and look at it from the U.S., it will come up as down, colon, other.
[28:44.400 --> 28:49.140]  So, some of those videos might still be up and viewable from other countries.
[28:49.300 --> 28:49.740]  Again, Tor.
[28:50.240 --> 28:51.060]  Yeah, we don't know.
[28:51.180 --> 28:55.280]  Tor is really great for this, because you can pop out of another country and check that out.
[28:55.440 --> 28:59.300]  So, I've been doing this all through a database interface, which is now available to you guys.
[29:00.060 --> 29:01.160]  And you can look at this.
[29:01.440 --> 29:03.120]  There's a column called banned countries.
[29:03.800 --> 29:07.680]  And if it's not null, then it means it's banned somewhere, and you can check that out.
[29:08.280 --> 29:13.120]  We've also kind of opened up some interface ideas where...
[29:13.120 --> 29:14.100]  This is just a mock-up.
[29:14.200 --> 29:15.520]  We haven't actually implemented this yet.
[29:16.700 --> 29:17.820]  Right now you've got a map.
[29:18.300 --> 29:18.660]  Right.
[29:18.820 --> 29:24.980]  Yeah, there's a map right now that will show you little red countries that will be, like, where they're banned in.
[29:25.600 --> 29:32.560]  But this is going to be kind of the web interface that we show at some point in the near future, as soon as Dean gets to work on that.
[29:34.480 --> 29:36.840]  So, anyway, we really wanted this to be interactive.
[29:36.840 --> 29:38.400]  We wanted to open it up to questions.
[29:39.520 --> 29:43.640]  And before we do that, you know, we just wanted to let you guys know where our website is.
[29:44.920 --> 29:46.300]  So, you guys can check this stuff out.
[29:46.380 --> 29:50.840]  And if you find anything interesting, you know, please send us email.
[29:51.160 --> 29:52.640]  We follow up on a lot of this stuff.
[29:53.140 --> 29:55.840]  And so, just opened it up to you guys then.
[29:56.980 --> 29:57.320]  Yes.
[29:57.600 --> 30:06.610]  The question was, are you...
[30:07.890 --> 30:11.210]  The question was, are you just interested in YouTube or the whole web?
[30:11.650 --> 30:15.130]  The short answer is, we're obviously interested in the whole web.
[30:15.250 --> 30:17.750]  The long answer is, we're obviously interested in the whole web.
[30:17.970 --> 30:22.350]  And we're all full-time students or, you know, have jobs and all that kind of stuff.
[30:22.350 --> 30:23.830]  So, if it...
[30:23.830 --> 30:26.530]  Like, we would like to implement other stuff.
[30:26.710 --> 30:28.870]  And if you guys want to help us implement other stuff, let us know.
[30:28.870 --> 30:29.630]  How are you?
[30:29.970 --> 30:35.110]  So, I had 21-hour...
[30:40.460 --> 30:40.980]  Hello.
[30:41.920 --> 30:44.160]  I've got 21 hour videos on Google.
[30:46.180 --> 30:46.700]  But...
[30:46.700 --> 30:48.420]  Which have put up in the last year and a half.
[30:48.980 --> 30:51.680]  And, about, I don't know...
[30:54.000 --> 31:03.580]  Until a few months ago, their counter to see hits and downloads for each video was working fine.
[31:04.400 --> 31:05.320]  And then it stopped.
[31:05.560 --> 31:06.280]  I don't know what happened.
[31:06.540 --> 31:09.580]  I've also had trouble uploading any more one-hour videos.
[31:09.860 --> 31:10.740]  I don't know what's happening.
[31:10.960 --> 31:12.840]  So do you have any idea what could be going on?
[31:13.680 --> 31:14.560]  That's not a clue.
[31:14.700 --> 31:17.660]  But if you guys have interesting stories like that, please email us.
[31:17.740 --> 31:18.800]  We do read all of our emails.
[31:18.800 --> 31:19.720]  Right, right.
[31:19.780 --> 31:24.140]  And if you can send us the links to the video as well, we can at least try and do a little digging ourselves.
[31:28.140 --> 31:29.140]  This thing... Okay, good.
[31:29.280 --> 31:29.560]  Kind of works.
[31:29.800 --> 31:35.880]  I want to ask if the swastikas were auto-blocked in France, Germany, and other countries.
[31:36.060 --> 31:46.440]  Because presumably, if they're not being auto-blocked, this means someone in Germany or whatever manually said, oh, I find any reference to the Nazis so bad that it must be eradicated.
[31:46.440 --> 31:47.940]  Well, that's what their boss said.
[31:48.080 --> 31:48.420]  So, I mean...
[31:48.420 --> 31:50.000]  I mean, this is a great question, right?
[31:50.160 --> 31:54.580]  And part of the problem, part of the motivation for this project is that it's such an opaque process.
[31:54.580 --> 31:55.520]  We have no idea.
[31:55.920 --> 32:09.020]  One of the tests that I've wanted to do, obviously not under account with my name on it, is get like one of those Buddhas that have kind of the reverse swastika and just film it by a mirror so that it kind of creates, you know, an actual swastika.
[32:09.020 --> 32:15.120]  So I don't seem like a Nazi douchebag, but maybe trip, you know, this little, like...
[32:15.120 --> 32:17.480]  We're assuming there is some type of swastika detector.
[32:17.640 --> 32:18.440]  We have no idea.
[32:18.660 --> 32:27.000]  There certainly is a textual, you know, detector where if you put metadata about a video that is blocked, then the video will be immediately inaccessible.
[32:27.000 --> 32:28.520]  Oh, well, I mean, you can do this very easily.
[32:28.660 --> 32:35.580]  I mean, you just upload a video that has all these various, you know, signs in it, and you just make sure that it has no views.
[32:36.200 --> 32:38.840]  And that actually should be pretty straightforward to test for.
[32:39.100 --> 32:39.460]  Right.
[32:39.860 --> 32:45.960]  So, and I also want to ask about the... So, where the NBA was blocked in China, I can see no reason why...
[32:45.960 --> 32:47.240]  It's not all NBA videos, by the way.
[32:47.260 --> 32:48.780]  It's only some random ones.
[32:49.100 --> 32:49.460]  Oh, I see.
[32:49.480 --> 32:51.600]  And it's not just ones about Yao Ming, because that was my first thing as well.
[32:51.600 --> 32:52.240]  Okay, yeah, no.
[32:52.460 --> 32:55.980]  So, I mean, presumably, China has no opinion on these...
[32:55.980 --> 32:57.480]  on these videos.
[32:58.400 --> 32:59.540]  But I guess they might.
[32:59.720 --> 33:03.860]  So, I guess it seems more likely that the NBA would be more likely to complain about it.
[33:04.020 --> 33:04.320]  Oh, absolutely.
[33:04.380 --> 33:11.420]  We think what it is, is the NBA market in China has jumped, like, hugely after Yao Ming's induction into the United States.
[33:11.560 --> 33:21.100]  And so, because of that, the NBA now goes, oh, there's a huge market over there, and let's try to make money off of them while we can, even though they're the best pirates in the world.
[33:21.780 --> 33:22.360]  Good luck, guys.
[33:22.360 --> 33:31.240]  Right, so, so I was wondering if you could detect if it was the country that complained, or was the rights holder that complained, based on, uh, down other or down copyright.
[33:31.520 --> 33:35.500]  Yeah, with the, the geolocational stuff, it's just a field inside the XML file.
[33:35.620 --> 33:36.900]  There's no reasons that are given at all.
[33:36.960 --> 33:38.360]  Well, but you had the reason there on the slide.
[33:38.680 --> 33:42.320]  So, so that was the, the text string that shows up when you try to view the video.
[33:42.800 --> 33:42.980]  Oh, I see.
[33:43.000 --> 33:45.980]  And down other is just, we didn't recognize what the string was.
[33:45.980 --> 33:49.660]  So that includes the thing saying, this video is inaccessible in your country.
[33:49.920 --> 33:50.040]  Right.
[33:50.240 --> 33:55.520]  Because we're scanning from the U.S., anything that's blocked in the U.S., we can't see the actual status of.
[33:55.920 --> 34:06.100]  And, and obviously another thing to keep in mind is that these are just videos that are blocked with tools that YouTube gives either countries or content creators.
[34:06.420 --> 34:11.000]  China most likely blocks a whole hell of a lot of videos with their own technology.
[34:11.000 --> 34:11.620]  Actually, they do.
[34:11.740 --> 34:13.400]  And I can, I can talk about that a little bit later.
[34:13.880 --> 34:15.440]  I just want to make sure we get to some other questions.
[34:15.680 --> 34:26.400]  But I've, I've actually done some experiments and, and they, China has a very advanced deep packet filtering system that they can detect things that piss them off and just block those videos.
[34:28.140 --> 34:29.100]  I've got a few questions.
[34:29.280 --> 34:33.080]  First of all, have you contacted YouTube at all and see, and what is their reaction to this?
[34:33.320 --> 34:34.040]  I mean, have you...
[34:34.040 --> 34:34.540]  Quiet.
[34:34.980 --> 34:36.220]  Their reaction is quiet.
[34:36.480 --> 34:37.000]  It's quiet.
[34:37.000 --> 34:38.940]  Yeah, we've, we first contacted them.
[34:39.060 --> 34:49.940]  Like the first thing we did before we even coded anything was we, um, we contacted representatives at Google and YouTube and were basically like, you guys want to give us some feeds so we don't have to like do this nasty scraping stuff?
[34:50.160 --> 34:50.780]  No response.
[34:50.780 --> 34:55.000]  And, and we talked to Chilling Effects and, and they said they had also asked Google for this and they had said no.
[34:55.380 --> 34:55.840]  So they said...
[34:55.840 --> 34:58.500]  We think part of the quietness might be the Viacom lawsuit.
[34:58.760 --> 34:59.740]  That's kind of understandable.
[34:59.960 --> 35:00.480]  That makes sense.
[35:00.600 --> 35:01.060]  Stop snitching.
[35:01.740 --> 35:06.980]  I mean, with all its problems, Google does have a history of being, you know, and trying to be open about the way they do.
[35:06.980 --> 35:15.140]  So I would assume they'd have some, or they'd at least want to put some kind of a, you know, API XML or SQL or something you could use.
[35:16.080 --> 35:16.520]  Okay.
[35:18.620 --> 35:28.780]  Now, uh, you guys use the example of, uh, the, like the anonymous, the anonymous video where the guy's files, he gets, uh, he gets, this video gets taken down.
[35:28.780 --> 35:40.180]  Well, I guess, as I understand it, the, the, the owner of that stuff, like theoretically legal owner, would, uh, upload their stuff and have it fingerprinted and all of that.
[35:40.180 --> 35:45.340]  Um, like, uh, not fingerprinted, but watermarked or not watermarked.
[35:45.980 --> 35:46.520]  Yeah, fingerprint.
[35:46.900 --> 35:50.040]  The, the problem was that Huffington Post claimed they owned it when they uploaded it.
[35:50.040 --> 35:50.200]  Yeah, yeah.
[35:50.500 --> 35:51.200]  But they didn't.
[35:51.620 --> 35:56.080]  Well, it's, uh, it's like a video fingerprint or some sort of.
[35:56.080 --> 35:56.480]  Yeah, right.
[35:56.600 --> 35:57.360]  Analyzation like that.
[35:57.360 --> 36:05.940]  But I'm wondering if you guys know, uh, or if you've done any tests on how that can be, um, like how close the video image needs to match and stuff like that.
[36:06.060 --> 36:07.340]  Or, you know, if anybody's done that test.
[36:07.780 --> 36:09.100]  It's an interesting set of tests.
[36:09.300 --> 36:12.560]  And if you want to help, like, we're more than happy to, uh, to bring you into the fold.
[36:12.840 --> 36:12.880]  So.
[36:13.280 --> 36:13.420]  Right.
[36:14.880 --> 36:15.880]  We haven't had time yet.
[36:16.260 --> 36:16.320]  Yup.
[36:18.180 --> 36:19.060]  Just more of a...
[36:19.060 --> 36:19.720]  Hello?
[36:20.280 --> 36:20.440]  Yep.
[36:20.440 --> 36:20.480]  You're alive.
[36:20.480 --> 36:21.920]  Just more of a quick clarification for the...
[36:21.920 --> 36:22.160]  Hello.
[36:22.480 --> 36:28.020]  More of a quick clarification for the gentleman in the front row asking about the, uh, one hour, uh, uh, uploading.
[36:28.480 --> 36:35.620]  Uh, I know with Google, uh, Google, uh, YouTube, they normally do like a 10 minute cap on their uploads.
[36:35.960 --> 36:36.020]  Right.
[36:36.020 --> 36:42.880]  So that is probably why you can no longer upload anything like an hour long video because Google and YouTube are the same.
[36:43.020 --> 36:44.980]  They just use YouTube's policy for uploading.
[36:46.080 --> 36:46.500]  No.
[36:46.540 --> 36:49.500]  Uh, no, because Google, uh, YouTube actually does counting and stuff like that.
[36:49.620 --> 36:50.280]  Tracking and all that.
[36:50.340 --> 36:51.340]  How many hits and all that.
[36:51.520 --> 36:54.240]  Has, has the 10 minute limit started being applied to Google Video?
[36:54.240 --> 36:56.760]  It used to only be YouTube that had that limit.
[36:57.300 --> 36:57.360]  Right.
[36:57.600 --> 37:03.120]  I'm just, I'm just, I'm just actually just speculating right now because due to the fact that he is no longer able to upload an hour video.
[37:04.300 --> 37:04.460]  Hmm.
[37:06.160 --> 37:06.480]  Question?
[37:08.060 --> 37:22.820]  Uh, given, uh, given the, uh, relative simplicity of your project and the relative popularity of things like, say, SETI at home and folding at home.
[37:23.040 --> 37:39.520]  I wonder if you might be able to solve some of your bandwidth and disk concerns that stop you from being able to scan and record all of YouTube using something like a distributed project using, say, Freenet as the back end, uploading everything there.
[37:41.840 --> 37:44.680]  That's, I can't get too much closer than I am.
[37:44.680 --> 37:55.880]  So the, the, the question, the question was basically can, can we do what SETI at home and folding at home are doing and distribute some kind of client that, that can crawl, you know, from people's PCs.
[37:56.020 --> 38:01.920]  Having individual PCs do the crawling and upload to a central store on something like Freenet .
[38:01.920 --> 38:03.360]  So we've, we've talked about that.
[38:03.640 --> 38:03.800]  Definitely.
[38:04.280 --> 38:07.800]  Um, right now our infrastructure is not capable of supporting that.
[38:08.020 --> 38:11.840]  Uh, we're not capable of handling the, the, the scan results that would come in.
[38:12.000 --> 38:19.400]  And also we're a little bit concerned about, uh, about people poisoning our data by allowing data submissions like that.
[38:19.760 --> 38:21.720]  But it is an idea that we've tossed around.
[38:21.980 --> 38:22.360]  It is an idea that we've talked about.
[38:22.540 --> 38:22.600]  Yeah.
[38:23.060 --> 38:23.240]  Thanks.
[38:26.300 --> 38:27.260]  Um, questions?
[38:27.960 --> 38:29.980]  Do you want us to show you some takedowns?
[38:31.140 --> 38:31.500]  Yeah.
[38:31.860 --> 38:32.140]  Okay.
[38:32.380 --> 38:34.020]  We can, we can take a look at the live site.
[38:37.260 --> 38:37.820]  All right.
[38:40.420 --> 38:40.780]  Oh.
[38:43.740 --> 38:44.280]  Let's see.
[38:47.320 --> 38:49.300]  No, we are not going to show you recastly.
[38:50.720 --> 38:52.940]  So, uh, so this is what the site looks like.
[38:53.100 --> 38:55.320]  Uh, does anyone know something that might be taken down?
[38:55.480 --> 38:56.080]  We can do a search.
[38:57.000 --> 38:58.820]  Oh yeah, search is live now, by the way.
[38:58.920 --> 38:59.200]  Yeah.
[38:59.200 --> 39:01.180]  We just added search functionality an hour ago.
[39:02.640 --> 39:05.720]  That was like the number one theme of all the emails we got.
[39:05.860 --> 39:07.480]  Like, have you guys thought about a search bar?
[39:07.660 --> 39:08.540]  Like, yeah, we've thought about it.
[39:08.540 --> 39:09.960]  And yes, we've been thinking about it.
[39:10.380 --> 39:12.820]  So, so what do, what do people want to like search for?
[39:13.220 --> 39:13.580]  Anonymous.
[39:13.920 --> 39:14.280]  Anonymous.
[39:14.800 --> 39:16.460]  So I don't know if there are any anonymous takedowns.
[39:16.480 --> 39:19.280]  There's, there's 16, um, copyright takedowns.
[39:19.340 --> 39:25.260]  If you go to browse or whatever, you can browse by, um, the, the copyright, like the copyright holders who have issued the takedown notices.
[39:25.560 --> 39:30.020]  And I think there's like 20 or so, um, that have been issued by the Church of Scientology.
[39:30.020 --> 39:31.880]  So, so it's up there for you.
[39:32.340 --> 39:35.040]  So these are the, uh, takedowns or these are, these are all available.
[39:35.300 --> 39:37.280]  There's one here that was removed by the user.
[39:37.960 --> 39:42.340]  Um, there was one that was removed as a terms of service violation.
[39:42.340 --> 39:44.180]  That actually looks anti-anonymous to me.
[39:44.540 --> 39:45.540]  Anonymous leader named.
[39:45.540 --> 39:52.020]  Um, interestingly enough, some of the copyright violation takedowns are, um, listed under terms of user violation.
[39:52.020 --> 39:58.160]  And we think that's just a faction, a factor of how the violation is reported, but we're not absolutely sure.
[39:59.400 --> 39:59.840]  Because...
[39:59.840 --> 40:00.960]  That's an interesting video.
[40:04.570 --> 40:09.610]  I'm not going to watch it, but, uh, let's, let's look at the things that are taken down by the Church of Scientology.
[40:09.910 --> 40:15.430]  Um, if you click stats, we show all of the, the, the takedowns that we've seen and, uh, who did them.
[40:15.430 --> 40:19.090]  So if you find, uh, I don't know, where's Scientology?
[40:19.370 --> 40:20.230]  It's way down there.
[40:20.970 --> 40:21.610]  It's like 20.
[40:21.890 --> 40:25.150]  The Church of Scientology International has 16 takedowns.
[40:25.870 --> 40:28.750]  Um, so this is the Tom Cruise video there.
[40:29.030 --> 40:32.910]  Does everyone remember the Tom Cruise video where he, he acts like a complete idiot?
[40:37.810 --> 40:38.290]  Um...
[40:38.290 --> 40:39.430]  That's why it says one of six.
[40:41.430 --> 40:45.210]  So, so it actually looks like we haven't seen a Scientology takedown in a long time.
[40:45.430 --> 40:46.470]  Which is kind of surprising.
[40:47.230 --> 40:49.270]  Maybe there's, maybe Anonymous is succeeding or something.
[40:49.750 --> 40:55.370]  But we, we did see a round of Anonymous, sort of like, takedowns and then reinstatements.
[40:55.670 --> 41:00.790]  Uh, one of the things that can happen is if your user account gets suspended, then all of your videos disappear from YouTube.
[41:01.210 --> 41:07.490]  So they were doing, there was some kind of back and forth, and there was, you know, not a lot of transparency in it, but we saw some of it on the site.
[41:08.370 --> 41:10.570]  Why do you use your account to get suspended?
[41:11.050 --> 41:12.890]  Um, for a number of reasons.
[41:13.090 --> 41:23.070]  There's an automated, there's an automated process that if they get a certain number of copyright takedowns or, or individual video terms of service violations, that YouTube will, uh, suspend the account.
[41:23.290 --> 41:30.970]  Uh, and this is due to one of the clauses in the DMCA that says, uh, you know, you have to also take steps to try and prevent this in the future.
[41:31.010 --> 41:36.510]  So you need to have some kind of policy where you will block people that, that have gotten too many takedowns.
[41:36.590 --> 41:42.730]  And it's actually kind of, kind of interesting that you don't actually need to know for sure that these are violating videos.
[41:42.910 --> 41:44.410]  You just have to receive the takedowns.
[41:44.490 --> 41:50.230]  So if you don't like someone, you can go find three, three or four videos that they posted on YouTube and submit takedowns.
[41:50.310 --> 41:54.270]  And in all likelihood, they're not going to actually bother contesting the takedowns.
[41:54.490 --> 42:05.250]  Um, and that's one of the, one of the big things we want to do is make sure that people are We are educated about the fact that, yes, you can actually say, no, this video is not infringing and YouTube is required to put it back up.
[42:05.930 --> 42:14.110]  So if you get three or four takedowns, then YouTube's obliged to suspend your account, regardless of whether they're legitimate or not.
[42:15.890 --> 42:18.190]  Are they really required to put them back up?
[42:18.790 --> 42:19.610]  It's the law.
[42:19.810 --> 42:20.690]  It is the law, yes.
[42:21.170 --> 42:22.070]  The DMCA.
[42:22.090 --> 42:24.010]  The Digital Millennium Copyright Act.
[42:25.250 --> 42:29.470]  Yeah, it says that they're required to take them down within 10 days, I think.
[42:30.250 --> 42:31.230]  I'm not sure about the details.
[42:31.630 --> 42:38.810]  So there's a short number of days that they're required to take the video down in after receiving the legitimate takedown notice.
[42:38.990 --> 42:42.350]  And then you have, I think, 14 days to...
[42:42.350 --> 42:49.930]  No, you can file a counterclaim to YouTube and they're required to put it back up within 14 days.
[42:50.190 --> 42:52.550]  Yeah, our friend Kevin was able to get his video back up.
[42:52.550 --> 42:56.350]  Within 14 days, they have to forward the counterclaim onto the original claimant.
[42:56.770 --> 43:02.190]  And if the original claimant doesn't file a lawsuit against you within 14 days, they're required to put it back up.
[43:02.350 --> 43:11.330]  And if you're interested in more details on the DMCA, you can search for it or in the About section, there's actually a link to a counterclaim form.
[43:11.330 --> 43:11.950]  So...
[43:11.950 --> 43:13.330]  If you look in our FAQ here...
[43:14.230 --> 43:14.350]  Yeah.
[43:15.030 --> 43:15.450]  Where is it?
[43:15.590 --> 43:16.350]  My stuff...
[43:16.350 --> 43:17.290]  My video got taken down.
[43:17.430 --> 43:17.830]  What do I do?
[43:17.950 --> 43:19.890]  You can file a DMCA counterclaim.
[43:20.530 --> 43:21.330]  And this is...
[43:21.890 --> 43:24.030]  It looks like it's big, but it's really just simple.
[43:24.150 --> 43:27.530]  It explains how you can determine if it's actually a legitimate takedown.
[43:27.670 --> 43:31.310]  And if so, how you can file a counterclaim.
[43:31.310 --> 43:32.410]  Can you make this easy?
[43:33.070 --> 43:33.430]  Yeah.
[43:33.550 --> 43:33.990]  We have five.
[43:34.470 --> 43:34.830]  Five.
[43:35.210 --> 43:36.130]  Yeah, we have five of these.
[43:36.310 --> 43:36.590]  Okay.
[43:36.890 --> 43:37.950]  Are there any other questions?
[43:42.350 --> 43:43.330]  Let's just wrap it up.
[43:43.490 --> 43:43.610]  Sure.
[43:43.990 --> 43:48.390]  Before we wrap this up, I just wanted to say, there are some folks from Creative Commons here.
[43:48.570 --> 43:56.850]  They're going to be doing a workshop right after this, down in the second floor, right next to where the Segways were in the Capture the Flag area.
[43:56.850 --> 44:01.510]  So, anyone who's creating audio, video, text, probably all of us are.
[44:01.670 --> 44:06.070]  If you don't know about Creative Commons, it's like an alternative licensing system.
[44:06.210 --> 44:10.730]  It's similar to like the GNU public license, but for culture instead of code.
[44:10.910 --> 44:14.030]  So, they're going to be doing that right after this.
[44:14.270 --> 44:15.170]  Check that out.
[44:16.110 --> 44:17.230]  And thank you all.