00:14:56lunik1 quits [Client Quit]
00:19:36lunik1 joins
00:21:36lunik1 quits [Client Quit]
00:22:01lunik1 joins
00:25:13lunik1 quits [Client Quit]
00:25:30lunik1 joins
01:06:56<lennier1>Visa and Mastercard have suspended payment processing for Pornhub. https://www.theverge.com/2020/12/10/22168240/mastercard-ending-pornhub-payment-processing-unlawful-videos
01:08:02<lennier1>PH already announced policy changes like ending anonymous uploading. https://www.foxnews.com/us/pornhub-rules-sex-trafficking-backlash
01:09:38<lennier1>It's not clear to me that users uploading illegal content is any worse than on more mainstream sites like Twitter, but easier to demonize an adult site if I may editorialize.
01:10:58<lennier1>I'm curious what they'll do about all the content already uploaded by users they can't identify. Most of it is presumably fine, but how can you prove it?
01:12:26<@JAA>I'm surprised this didn't happen years ago. It's not like these problems haven't been publicly known for years.
01:29:14<OrIdow6>Do you have to pay to use PornHub, or does it get its money from ads?
01:31:08<lennier1>It's mostly free and ad-supported. You can get a premium subscription with higher resolution videos and some premium-only videos. Users can also have paid only videos or accept tips.
01:33:29<flashfire42>The Problem is that there are far to many stories of pornhub refusing to take down content unless you get a lawyer involved plus for a platform that claims to check every video for illegal content (Which would be impossible in itself) there is a lot of illegal content there
01:33:50<Ryz>I'm not sure of the viability of archiving Pornhub since a ton of it is videos; like if WBM/IA can't handle hosting the content overall, then it may be unlikely it will handle the flood of YouTube content if what Jason said will happen in the future
01:34:25<flashfire42>I am against archiving pornhub for the simple reason that there are too many non consensual videos up there and I dont wanna be the one that helps accidentally push CSEM into the wayback machine
01:34:47<flashfire42>if we do a few captures of the front page maybe the categories page to show how many videos were up
01:35:14<flashfire42>It gains money from ads but it puts ads in front of videos of very questionable origin
01:36:06<Ryz>I'm talking from the assumption there was no vast amounts of illegal material (compared to I guess say YouTube or Dailymotion), which I did not realize Pornhub harbors that kind and amount of content oo;
01:36:30<Ryz>...There's still countless porn websites though
01:36:37<flashfire42>Yeah Youtube and dailymotion do have a problem with illegal content but not nearly on the scale of Pornhub.
01:37:13<@JAA>Also, on other platforms it's usually copyright infringement. Here, it's rape videos and CP.
01:37:15<flashfire42>Mindgeek and WGCZ Holding are the 2 biggest porn companies in the world and both of them dont give a shit about CSEM on their platform if they can make money off of it
01:37:18<lennier1>I mean, literally any website that allows uploading of adult material without strict verification is going to have some illegal material. I'm sure that was true of Tumblr, is true of Twitter, etc.
01:37:42<@JAA>Yep
01:37:51<Ryz>Huh, what do they mention by 'anonymous uploading'? Like you don't have to make an account and just upload it?
01:38:05<@JAA>The difference is that, from what I've heard, Twitter et al. remove that content if reported. PornHub doesn't.
01:38:16<flashfire42>Twitter needs it pointed out to them to have it taken down which is a fucking annoyance but they usually take it down when reported. Tumblr was a nightmare from the start let us be honest here. But pornhub doesnt report it unless a lawyer gets involved
01:38:32<lennier1>You need an account, but they don't verify your real name (i.e. by requiring a credit card or driver's license).
01:38:34<flashfire42>Take the case of Rose Kalemba who had to impersonate a lawyer to get her assault videos taken down
01:39:15<lennier1>I'm not sure if they've even finalized the new policies, but I think the idea if if someone uploads illegal material, they'll know who did it.
01:39:36<flashfire42>also AFAIK twitter doesnt claim to screen every tweet before its posted Pornhub has claimed on multiple occassions they check every video and use fingerprinting despite the fact that illegal content that is the same video has been uploaded multiple times there
01:40:14<@JAA>Let's split off the 'why PornHub sucks' discussion into -ot and deal with any potential archival only here.
01:40:17<flashfire42>sorry
01:40:48<@JAA>No worries, some context is fine here.
01:41:40<flashfire42>Taking aside that Pornhub is a shitty company that hosts non consensual content the size of it is not viable for an archive when we consider that a lot of porn content doesnt often have a lot of variety
01:42:17<flashfire42>If we were to grab maybe the top 10 videos in each category that might be viabkle
01:42:37<flashfire42>But I doubt IA has or wants to provide that level of bandwidth to pornography
01:43:33<lennier1>Is PH (potentially) removing content that different from Tumblr removing adult content?
01:43:51<flashfire42>In a perfect world I would love to archive the internet but we need to pick our battles and I dont think Pornhub is in the scope of archiveteam. Hell there was a debate to eroshare
01:44:09<@JAA>Tumblr was mostly hosting shitty GIFs, PornHub is hosting high-resolution videos. So there's a massive size difference.
01:44:57<flashfire42>Not to mention the fact that a fair amount of tumblr was also text posts. and small images. Pornhub doesnt have a functioning forum or large text based area
01:45:58<lennier1>Tumblr didn't remove text posts, only images they auto-IDd as nudity. I forget if Tumblr had videos.
01:46:25<Ryz>Tumblr does have videos uploaded, and also audio too
01:46:25<flashfire42>Not like xnxx which has sexstories(dot)com which could be a viable thing to save if it comes to it. We are talking lots of videos some of which are of assaults and I dont know about you but I dont wanna be the one to contribute to someone suing IA because their rape is part of history. Of course that would be an edge case but I am Just putting out
01:46:25<flashfire42>I am against this archival for the first time in my life I dont wanna see something gobbled up by IA
01:46:36<@JAA>Well, if they did, it wasn't a very large part I think. It was mostly known for its poor-quality GIFs.
01:46:49<lennier1>But I'll grant there could be a big data size difference, no real ideas of actual numbers.
01:47:43<flashfire42>I would wager there may even be up to a petabyte of porn from pornhub. Not to mention this will most likely affect MindGeeks other sites when people catch on that 2 companies hold a monopoly on the porn tube sites
01:47:57<lennier1>PH does have a lot of images as well. Plus blog-type posts and comments, though obviously that's not the site's primary draw.
01:48:28<flashfire42>Anyway I have said my bit I may say more but I think I have made my stance clear 😓
01:51:53<Ryz>How many websites does Mindgeek have? I'd imagine there'd be a pornado sprouting out from that hell of a mess
01:54:50<@JAA>RedTube, YouPorn, SpankWire, etc. Dozens of sites like this. They also own various porn production companies.
02:00:02Dallas quits [Client Quit]
02:00:15<lennier1>They also own Xtube, which doesn't seem to be as popular anymore, but has lots of older user-generated content.
02:03:45Dallas (Dallas) joins
02:06:13<VerifiedJ>On a unrelated note, I've finished discovery for archiviolastampa.it and have started downloading. 63GB so far. Should be done in a day or two.
02:07:50<jodizzle>VerifiedJ: Excellent! I was worried about that one.
02:08:00<jodizzle>Are you planning on uploading to IA when done?
02:08:28<VerifiedJ>yes that's the plan.
02:09:01<@JAA>Nice
02:09:49<jodizzle>Would there be any value in processing a list of asset URLs in archivebot as well?
02:14:07<OrIdow6>Looks like it's a privately-held company, which makes it harder to tell how this will affect them
02:14:57<@JAA>VerifiedJ: WARCs or plain files?
02:17:13<VerifiedJ> plain files. I'm extracting the images and creating _images.zip.
02:19:38<VerifiedJ>All the important content is loaded via flash with custom session ids and cookies.
02:21:42<@JAA>Right, so having it in the WBM (via AB) is probably of little value.
02:21:54<@JAA>Unless IA adds some site-specific session ID removal.
02:44:22<lennier1>To the extent that video archiving is a question of storage size, it could be cool to use fingerprinting to not save duplicate videos. There's definitely a mix of people uploading self-produced videos and people uploading videos they saved elsewhere.
02:57:18<OrIdow6>Yes, could work
02:57:58<@arkiver>so
02:58:02<@arkiver>pornhub going away?
02:58:03<@arkiver>or?
02:58:04<OrIdow6>Though of course video is subject to generation loss just as much as video
02:58:11<@arkiver>what is this all about
02:58:27<OrIdow6>And that's more of ahypothetical; don't think IA has enough money to host all video ever put online :)
02:58:43<@arkiver>no we dont
02:58:46<@arkiver>selected video yes
02:58:48<@arkiver>low res maybe
02:59:04<OrIdow6>arkiver: No, they just may be losing payments from sources
02:59:25<OrIdow6>Which means long-term threat, if consumers paying them directly, as opposed to advertisments, are a major source of their money
02:59:49<@arkiver>right
03:00:00<OrIdow6>*from some sources
03:00:33<@JAA>And potential loss of very problematic content if they finally take their role seriously and take down rape videos etc.
03:00:54<@JAA>Nothing we'd want to archive except possibly metadata.
03:03:44<OrIdow6>Unless IA wants it for whatever reason
03:07:17<lennier1>They've announced changes including requiring verification to who can upload videos moving forward. Then the payment processing decision came soon afterwards, so apparently Visa and Mastercard felt their changes so far were inadequate. Anything beyond that is speculation, but I do think there's some risk of mass removal of videos with no obvious illegal content, but of unknown provenance.
03:09:25<lennier1>Not sure what percentage of their revenue comes from credit card payments vs. ads, but obviously there's potential risk to their whole business as well.
03:10:42<OrIdow6>On fingerprinting for deduplication of large sets of video: one thing I can see working, given natural and artificial "generation loss", is doing it on the level of segments of the video, vs. the whole thing
04:54:20qw3rty_ joins
04:57:44qw3rty quits [Ping timeout: 240 seconds]
05:00:01treora quits [Quit: blub blub.]
05:01:16treora joins
05:13:19Stilett0 quits [Remote host closed the connection]
05:50:37rsn_ joins
05:50:37rsn_ quits [Remote host closed the connection]
06:09:45<OrIdow6>Running a very crude script that OCRs the country code from the .eu registry, if one exists
06:10:01<OrIdow6>Fairly strict rate limit
06:16:16Craigle quits [Ping timeout: 244 seconds]
06:17:34Stiletto joins
06:20:14<OrIdow6>Now checking the language of the registrant if country is not present
06:32:32<thuban>what's the rate limit?
06:33:38<OrIdow6>Don't know
06:33:48<thuban>(& is this just for testing, or can we not take an ocr dependency in the warrior?)
06:33:53<OrIdow6>Strict rate-limiting, I should say
06:34:15<thuban>?
06:34:29<OrIdow6>Somewhere in 1-10 seconds, but once it limits you it takes a lot of exponential backoff before it lets you back
06:34:42<thuban>ah
06:34:43<OrIdow6>Safe, no-429 rate limit is somewhere in that range
06:35:20<OrIdow6>I'd rather not have Tesseract on the warrior - presumably there are a finite number of country code PNGs, so we can look for those, maybe
06:35:32<thuban>mm, good thought
06:36:27<OrIdow6>Though curiously, I'm not seeing "UK" in any country code, and hardly any no-address domains except for those with a language listed as English
06:37:30<thuban>"GB" is the iso 3166-1 for the uk. don't ask me
06:37:53<Ryz>!ig 4efdbcgenejtrjbusqs03fug9 ^https?://paydayloansbits\.com/
06:37:55<Ryz>Oops
06:38:15<OrIdow6>Oh, I see one for GB of ~30
06:38:34<OrIdow6>Now it's 429ing me, 15 seconds was too fast, apparently
06:38:36<thuban>i checked two i turned up at random and they're bit-identical
06:38:42<OrIdow6>Good
06:38:52<thuban>(https://whois.eurid.eu/en/search/?domain=eldridgehouse.eu, https://whois.eurid.eu/en/search/?domain=coleridge.eu)
06:39:30Craigle (Craigle) joins
06:40:38<thuban>what about fallbacks, then? english-language isn't very specific; emails are better (.co.uk, .uk.com, etc) but bring back the ocr question
06:41:36<OrIdow6>Seems like most of the emails I've looked at are Gmail, Hotmail, etc.
06:42:37<thuban>do we want to have one warrior round for discovery with web-whois and another for actual archiving?
06:45:03<OrIdow6>Just looking at the results I have so far, fewer than 1/5 domains are from the UK, and presumably even fewer of those are going to be something other than a parking page etc.
06:45:47<OrIdow6>So that could get the warriors blocked, considering those tend to have a smaller number of IPs, than would be possible for a dedicated discovery project
06:46:11<OrIdow6>But I suppose they could just have a delay
06:46:55<thuban>what's the dedicated-discovery setup look like
06:51:34<OrIdow6>So I don't think there's been a separate-discovery warrior project in the time I've been here, but looking at what's happened in the past, it's like a normal one, except wget-lua (or for some, a specialized Python script) writes to a text file instead of warcs
06:51:50<OrIdow6>However, I presume backfeed might be used now
06:52:52<thuban>oh--i thought you meant that there was some alternative arrangement with a larger number of ips that we've used/could use
06:55:32<OrIdow6>I get the general impression that it is easier to run a small number of high-bandwidth workers, or a bunch of many ips, than do both
06:55:42<OrIdow6>But maybe that's not the case
06:56:27<OrIdow6>So I suppose this is really a question for the subset of AT that does the running
06:56:38<thuban>aye
07:46:41Craigle quits [Ping timeout: 244 seconds]
07:59:27Craigle (Craigle) joins
08:12:44<tech234a>Orldow6: I don’t think you need to do OCR for each image... it seems that each country code is it’s own base64 encoded image separate from the rest of the address and consistent for each country code and you could probably just collect a list of domains with each base64 country code and then just look at the final list
08:13:03<tech234a>Are you scraping the similar domains list also?
08:13:15<OrIdow6><OrIdow6> I'd rather not have Tesseract on the warrior - presumably there are a finite number of country code PNGs, so we can look for those, maybe
08:13:51<OrIdow6><thuban> i checked two i turned up at random and they're bit-identical
08:14:08<tech234a>Whoops missed that but yeah that’s what I also found
08:15:35<Ryz>!concurrency aod8v7tn4csv9ys3xam351jg4 1
08:15:48<Ryz>Oops
08:16:20<OrIdow6>No, I am not; the point of this scrape is just to gather statistics on how many domains are in the UK
08:17:54<tech234a>Yeah I thought that might help with finding more domains but it probably adds complexity
08:18:34<OrIdow6>Speaking of that, looks like .ευ and .ею might be having their registrations revoked, too
08:18:46<OrIdow6>"The same registration requirements as for .EU and .ею apply, that is to say that a presence in the European Union, Iceland, Liechtenstein or Norway is required, or EU citizenship." - https://www.lexology.com/library/detail.aspx?g=bc0b9ac7-1284-4884-91d9-ffabf39d9724
08:24:02<tech234a>Their WHOIS policy actually seems kind of reasonable, perhaps we could request a disclosure instead of scraping?: https://eurid.eu/d/205797/whois_policy_en.pdf
08:24:32<OrIdow6>Appears to be one domain each in the WBM
08:26:34<thuban>^ hmm, maybe
08:26:56<OrIdow6>Looks to me like that process pertains to "thedisclosureofpersonaldatanotpublishedintheweb-basedWHOISbutprocessedbytheRegistry" (add the spaces yourself)
08:28:00<OrIdow6>Doesn't explain why some of them don't have addresses listed, considering the registry supposedly "publishes" that
08:28:14<thuban>...which would include whether any given uk-owned domain is noncomplying and will be removed; we know they have that information
08:28:20<OrIdow6>Well, city and country
08:28:37<thuban>that's only for legal (as opposed to natural) persons
08:29:05<OrIdow6>Oh, you're right
08:29:32<thuban>there are some i've seen that look like they're noatural persons, but do include addresses; dunno whose end that happens on
08:31:07<thuban>i guess the question is, if we ask nicely, are we likely to get btfod in a way that would compromise any subsequent scraping operation. i don't have a lot of experience with stuff like that. is this something we should rope jason in on?
08:32:46<OrIdow6>From the way it's written, it sounds to me like it's intended for information on individual domains, rather than a list of everything
08:34:07<OrIdow6>It's not like the registry has control over the contents of the websites, and it's unlikely they'll suspend it early if someone asks
08:35:11<OrIdow6>But I can see there being a possibility - how likely, I don't know - for adverse legal effects
08:38:49<OrIdow6>Here's the form - https://eurid.eu/d/5294809/request_form_disclosure_personal_data_en.docx - clear that it's for a single domain
08:39:08<thuban>ah, disappointing
08:39:26<thuban>i suppose we _could_ ask anyway
08:42:42<OrIdow6>Very unlikely, I think, that they'd give it up
08:43:07<OrIdow6>Well, maybe
08:43:41<OrIdow6>More likely, but still unlikely, if you just asked for a list of domains being revoked
08:45:03<OrIdow6>But the questions are - how willing would they be willing to do that (given that it would be used for web scraping)? And from how high does the decision have to be made?
08:46:35<OrIdow6>(Normally we make a distinction between scraping and what happens here, but to many website owners, and I'd guess in legal contexts, they're the same thing)
08:57:13mutantmonkey quits [Remote host closed the connection]
08:57:31mutantmonkey (mutantmonkey) joins
09:26:47godane1 quits [Read error: Connection reset by peer]
09:26:56qw3rty_ quits [Read error: Connection reset by peer]
09:27:29godane1 joins
09:28:45qw3rty joins
09:31:00godane2 joins
09:31:29godane1 quits [Read error: Connection reset by peer]
09:55:08Atom joins
09:56:11Atom-- quits [Ping timeout: 252 seconds]
10:15:58godane1 joins
10:18:32godane2 quits [Ping timeout: 252 seconds]
10:37:04Barto quits [Ping timeout: 240 seconds]
10:50:06hooway joins
10:53:00HP_Archivist quits [Read error: Connection reset by peer]
10:55:15Barto (Barto) joins
13:06:05godane1 quits [Client Quit]
13:20:39<@JAA>The .eu CDX API scan should be done in a few hours I think.
14:42:04Arcorann quits [Ping timeout: 240 seconds]
14:52:52Ryz quits [Remote host closed the connection]
14:53:36Ryz (Ryz) joins
16:42:07AeonG joins
16:42:18AeonG_ joins
19:47:30lukash quits [Client Quit]
19:52:18mupf joins
19:59:24lukash (lukash) joins
20:40:04Barto quits [Ping timeout: 240 seconds]
20:44:34Barto (Barto) joins
21:02:15AeonG_ quits [Read error: Connection reset by peer]
21:02:16AeonG quits [Read error: Connection reset by peer]
21:04:38spirit quits [Remote host closed the connection]
21:09:00Barto quits [Ping timeout: 252 seconds]
21:52:27Barto (Barto) joins
23:10:45lennier2 joins
23:14:14lennier1 quits [Ping timeout: 252 seconds]
23:14:15lennier2 is now known as lennier1
23:20:53Arcorann (Arcorann) joins
23:36:16hooway quits [Client Quit]