[aspenmayer]

Is there some kind of automated NSFW/nudity detector that runs against Wikipedia/Wikimedia uploads? One would think there would be at their scale, I just don’t actually know. I saw your user page the other day while looking at a proposal to have a .onion URL for Wikipedia, and I thought that I’d seen you around here, and figure you’re as good a person as any to ask.

[bawolff]

Not as far as i know, i think things are just manual review.

NSFW images on Wikimedia tends to be a very hot button issue when it comes to Wikimedia politics. There obviously some cases where such images are needed, and there is a lot of debate on where the line should be drawn (or if it should be drawn at all). Wikipedia is traditionally very anti-censorship in any form. Fun fact - Wikipedia founder Jimmy Wales lost a bunch of his user rights when he went on a deletion spree of classical art work containing nudity.

There is an automated filter for child sexual abuse images. Its not public what the procedures related to it is, but i assume if it goes off the fbi gets called.

[aspenmayer]

I think I heard about Jimmy getting his privs revoked, but had no idea what precipitated that to occur. Some kind of misguided heading-off-at-the-pass of future naysayers and potential lawfare, I guess. Heavy hangs the head that wears the crown, and I guess he’s a figurehead as much as he’s a lead of operations. It makes sense to tie his hands so he can say he tried I guess. Interesting interplay of authority, both individual and collective.

No means or methods are necessary. I am familiar with the work of the NW3C and other groups who do important work in that area. My sympathies to the janitors. Truly difficult and important work.

[bawolff]

[NSFW] https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2... has a summary of the issue. From what i understand Fox news was running a bunch of stories at the time on how Wikipedia was full of porn.

In full fairness to jimmy, i think only a handful were risque art from the 1800s. Many of the other images would probably be classed as sex educational if you're being sympathetic, and exhibitionist if you are not. However the ultimate issue was not what the files contained but that he acted alone without agreement to delete outside of proper procedure.

[aspenmayer]

Thanks for the inside baseball play-by-play, coach! Much appreciated.

> However the ultimate issue was not what the files contained but that he acted alone without agreement to delete outside of proper procedure.

Were any changes to Wikipedia policies implemented as a result of this, and if so, do you know which ones?

I found the link to the discussion I referred to upthread since the discussion was from a bit ago:

https://meta.wikimedia.org/wiki/Grants_talk:IdeaLab/A_Tor_On...

Did this ever go anywhere? I found a grant proposal, but I need to sign up for an account or reactivate my existing one, or whatever.

https://meta.wikimedia.org/wiki/Grants:IdeaLab/A_Tor_Onion_S...

[bawolff]

> Were any changes to Wikipedia policies implemented as a result of this, and if so, do you know which ones?

Not that i'm aware but it was a long time ago so i might just not be aware. Note that the majority of the files he deleted were undeleted and are still present to this day. The list is at https://commons.wikimedia.org/w/index.php?title=Special:Log&... and most of the links are blue.

> I found the link to the discussion I referred to upthread since the discussion was from a bit ago:

Not as far as i am aware. People are really nervous that without accurate IP addresses it will be difficult to do anti-abuse stuff.

The new thing that is happening right now is we are killing public display of IP addresses of anonoymous users. Perhaps once that rolls out people will be less attached to ip address tracking and more open to something like tor.

That said, if the tor hidden service is read but not write as the grant proposals, that solves that concern. I don't think anyone would object to a read-only service. There just isn't that much interest from the people who could make it happen, and regular concerns about added complexity for limited gains.

As an aside, the grants process is really disconected from wikimedia tech stuff. Grants might give some money to someone to make an unofficial mirror, but it wont be helpful for making an actual official tor hidden service. There is a 0% chance that a grant will lead to an official hidden service. If this ever happens the discussion threads will be on phabricator and not grant pages.

The way in theory to make this happen is one of:

- convince wikimedia community this is super important. (Unlikely to happen as this is too niche). Wikipedians have some influence over WMF priorities but really only when they start a riot.

- convince wmf senior leadership it is super important (also pretty unlikely)

- lobby the idea with individual developers who work on SRE stuff. Maybe if you convince them, they convince their boss, and it eventually happens when the team is having a sliw sprint. (This is the part where its open source so external contributions are in theory possible to a certain extent, but to effectively do this you basically already need to be an insider and know all the right people to talk to)

[aspenmayer]

Appreciate the thorough concise reply.

I did hear about the hiding of anonymous users’ IPs in passing as I was looking into the .onion URL for Wikipedia concept. I was looking at stuff like TorBlock and thinking that they’re so close and yet so far from having a Tor accessible view or subsite. It could be a special view like the .m mobile views, or something. It could be made to work, but it would take some doing.

https://en.wikipedia.org/wiki/MediaWiki:Torblock-blocked

https://www.mediawiki.org/wiki/Extension:TorBlock

If one can implement that, a read-only mode for Tor users would seem fairly straightforward.

Once while randomly walking around in the Sunset district of SF, near the Internet Archive iirc, I bumped into someone who claimed to be a legal counsel of Wikipedia/Wikimedia with a business card to match. I don’t have many Wikipedia contacts besides, alas, but I am already a user, though I don’t post much.

https://en.wikipedia.org/wiki/User:AspenMayer

Found my old account, as it were. Then I tried to edit my user page and discovered I have an IP block due to using iCloud private relay and/or a vpn. And that block prevents me from editing my own user page. That’s where this policy is going too far, imo. It’s my user page! I can see how even that feature could be abused, but come on. I know it’s not your personal policy, it’s just frustrating.

The issue seems one of institutional support and momentum more than a technically difficult problem to solve, especially if the Tor version were read only. I think it is just a shame that the issue has low awareness. Then again, I bet more folks associate Tor with bad actors than good or merely desperate ones, which is an earned reputation when it comes to bad actors on Tor. I don’t know how fair that is, but that’s the way it is currently. If only we could have a profile option to enable Tor access and then folks could get a unique .onion URL that proxies to a backend Tor read only connection to Wikipedia/Wikimedia? I think that could be done, but impetus is lacking, I suppose.

In a lot of ways, Tor can seem like a solution looking for a problem, at least for legitimate use cases that don’t involve law violating and/or antisocial behavior. It’s a shame that a bad reputation can hold Tor back from doing more good. I think Wikipedia is the single most important and impactful site that could benefit the world and all Wikipedia users simply by adding a .onion version, even if it has the same policy on blocking open proxy access and Tor access from edits. Read only is better than the status quo where Tor use is risky but necessary to access the site, as the alternative to that is no access at all.

I’m a working journalist, so I’ll have to make an effort to edit this Wiki user page while maintaining opsec, because I’m not turning off my vpn or private relay, or any other security features on my end of the connection. I respect Wikipedia/Wikimedia, but I have standards, and sources, to protect.