[numpad0]

Because 2 out of top 3 servers and half of top 10 were in Japan and filled with content that they couldn't politically handle.

Cutting that off and "just walk into the brightest place" couldn't happen at the same time, so the core devs and ops switched to the "Mastodon's strength is in small servers, pick any of the right one" narrative.

[wkat4242]

Can you tell more about the Japan thing? It's the first I've heard of it. What kind of content? I can't think of anything political about Japan that's super hard to handle.

[zimpenfish]

I believe there's a fair amount of content on Japanese Mastodon servers that would land me, as an admin in the UK, in extremely hot water[0] if I was letting it be stored on my servers (whether I looked at it or not.)

[0] https://en.wikipedia.org/wiki/Child_pornography_laws_in_the_...

[wkat4242]

Ah I see the age of consent in UK is 18 and in Japan it appears to be 16 (the same page links to the Japanese one). Weird.

I thought they were super strict there (after all they even require that blocky censor thing on the functional parts in normal adult pornos making it basically useless).

[numpad0]

Strict cuts many ways. Doesn't involve kids, not a CP in Japanese laws. Doesn't depict details, not a porn. Two high schoolers having an affair, uhh not in penal code, so violations. Adults buying kids, that's prostitution and/or assault per code. Age from looks and/or workplace fitness factor, it's subjective so doesn't matter.

IMO, human aging vary too much to solve this "looks kids to me" problem. Age by appearance is clearly regional; there are plenty reports of East Asians abroad barred entries at morning and beer at evening. So any substantial East Asian content feed is destined to include tons of "kids/age unreliable" content, unfortunate opposite being "middle aged/unknown". Other factors as cultural norms and effect of modes of consumption compounds on it.

Long term, the solution to this has to be algorithmic: the mix of content must be artificially manipulated so to suit the need of consumers. The fraction of first(primary) language English speaker is below 9%[2], less than third of Mandarin Chinese at 21% or about same as Spanish at 11%. So a viewer application for a truly global feed of all contents will have to mark and remove ~90% of content for it to be palatable to an English speaker. "Endless Japanese CP" problem is just but an early sampler.

1: https://en.wikipedia.org/wiki/List_of_languages_by_total_num...

[wkat4242]

Algorithmic? I think simply checking like the US law "ID 18 U.S.C. 2257" already does is a much better solution than having algorithms guess the age based on a photo or video.

Though I have to say I sometimes wonder how sites like Onlyfans do this with male actors. Often the only thing you see is their dick and I can't remember having to submit a dickpic for my passport :D So how do they know a submitted passport matches the dick in question?

But really I think it's better to know for sure than to guess.

[numpad0]

Laws just don't matter. People physically can't look adults/kids if they're from "wrong" cultures whether they were 60 or 16 and put GP into hot water by appearance.

Sites like Onlyfans, or any website for that matter, just delete contents based on appearances and user metrics. Otherwise escalations won't stop; users would leave, protest will occur, new laws will be created, state committees will be formed and blacklists are made, and credit card processing suspiciously starts failing. Recordkeeping is for honest people to be honest if they want to be. Some numbers on a driver's license do not stop the horde from banging the walls.

So a truly global feed has to hide inconvenient contents from users until users are ready for it. Ideally user editable. Which is what everyone is fat-finger doing at server side anyway.