> your favorite artists and photographers on Pixelfed

Pixelfed is more or less an Instagram clone ... which is like photo/video microblogging. Because the audience is largely 'casual' and encourage taking and consuming media from a smart phone it does some things with images that are antithetical to what a _good_ platform would be for artists. Instagram compresses the living hell out of images, it limits you to square and close-to-square dimensions, the color profile is stripped (I would assume you still can't even upload DCI-P3 photos from non-Apple platforms for whatever reason), and all license and metadata is stripped from the image. Most, if not all, of these issues have traveled to Pixelfed too. If it wasn't for the walled status and trying to gain 'casual' followers, I don't think artists would prefer the degradation of image options, quality, and metadata of Pixelfed/Instagram if they could be federated on a different, image-quality-focused platform. It's not without it's shortcomings, but Flickr better at this sort of thing and was talking of federation this week: https://twitter.com/DonMacAskill/status/1594945727255699457. Currently nothing in the Fediverse respects the media in the way a deviantArt, 500px, Flickr, et.al. type platform does.

Presumably a pixelfed server could be configured to support full resolution raw images for their local users. The federated feed could push out the lower resolution images that are fine for viewing on phones.

This would be a way for an ActivityPub service to work on a freemium model.

I don't think you need to ship full-res images, it's that the compression and stripping should be turned down and formats like JPEG XL should be supported to push a high-quality-but-small-size image for consumption. The line between a photo and photography is vague, but folks' daily coffee photo probably doesn't need to have the compression dialed back a lot.

My point was that a pixelfed instance wouldn’t necessarily need to reduce the size or quality of images if those that were uploading them were bearing the cost of storage and data transfers. I said raw because I am far from enough of a photography geek to suggest something else that would be ridiculous for casual viewing on a phone but a reasonable request for someone that will do something interesting with photos they download.

Why is flipping the switch bad in this context? Isn’t that actually what the protocol needs, a major visible platform adopting it with open arms all at once?

Look at how much chaos came from just adding a few million people in a month. I don't see a way that goes better adding tens of millions in the same or shorter period. The best way for Tumblr to do it would be to:

1: Open source it

2: Make it federated with other Tumblrs. Probably using whatever BlueSky cooks up, or some similar effort focused on big platforms.

3: Offer a managed version the same way they do with WordPress for the majority of people who don't want to run their own

4: Provide an optional ActivityPub extension

5: Telegraph that federation is coming for the still probably too big main instance(s) to give administrators time to decide how they'll handle it: mute, block, allow, etc. This gives people who eventually migrate off in 6/7 a clear view of how instances plan to handle it.

6: Add ActivityPub support to the main instance(s). Make it opt-in.

7: Wait to see how it shakes out as people migrate to non-Tumblr instances. By now, probably 2-3 years in, there will be enough diversity in platforms and hosting options that most people will have some place to go that's appealing. We already see minimal AP instance software popping up that focuses on providing people with a Linktree/Caard-like experience where it's only a profile that's exposed to AP instances. Maybe even sites like that will add AP support! Posts from these could replace newsletters. (I know, a common claim for as long as newsletters have existed)

Tumblr already effectively federates inside its own network. Instead of accessing <blog>.tumblr.com, log in at tumblr.com and then visit tumblr.com/<blog>. Each user must have one blog. A user can create more blogs and a blog may have more than one contributor. Only primary blogs can comment, like, etc. What I would personally do is copy Mastodon's approach. This is super easy and pain free and doesn't require messing with Hellsite's architecture and potentially ticking off thousands of users while simultaneously playing better with Mastodon, specifically. How do I accomplish this? By creating a proxy for each Federated entity the first time it is needed. Tumblr's local API continues to just operate against local entities, tagging the proxy when needed. A separate process batches outbound notifications, grouping by target server. Inbound, I just locate or create the proxy for the given federated entity - then use the existing deliver infrastructure to deliver the notification. This also gives me a good way to cache content - and allows local trust and safety to do their thing if they need to. It limits all code changes related to this to the edge of my network, limiting the risk.

i think right now the protocol and the people who run popular mastodon instances are being treated as the same thing - mastodon.social and various affiliated servers are trying to build a community based around a certain set of values and type of content, and feeding all of tumblr into that might cause some drama.

it's unquestionably good for the protocol to have more people using it. but it might not be good for mastodon.

Mastodon.social and other large instances certainly aren't working to build the same focused, higher quality communities that smaller instances thrive on.

This is part of why many instances block the large, poorly moderated instances.

Pivix did exactly that, becoming the majority of the fediverse overnight. The result was a huge concerted effort to block them.

You mean Pawoo created by Pixiv? Welp, yeah, but having Tumblr federated would not raise legal issues like that.

Over a novel, non-standard definition of child pornography. I suspect the fediverse would find something equally offensive to redefine tumblr as responsible for.

You have to be pretty far down the weeb hole to think that anime loli isn't regarded as child porn by anyone outside of that particular cave.

There are reasons child pornography is bad and loli doesn't share any of those reasons. Equating those two is pearl clutching and an insult to victims of actual child pornography.

There are reasons child pornography is bad and loli shares some (but by no means all) of the reasons. Just because loli isn't as bad as 'normal' child pornography doesn't mean that it's not still bad.

Sexualisation of minors is not a good thing, and taking a stance against it is a perfectly reasonable action. For example - Netflix's "Cuties" film should never have been made, except that it did at least draw some attention to the real life pageants that are themselves a problem.

Regardless of the specific distinction between the two, it would be possibly illegal for me, as a UKian, to allow that content[1] - "The Act made it illegal to own any picture depicting under-18s participating in sexual activities, or depictions of sexual activity in the presence of someone under 18 years old."

(I admit that I am not well versed enough in "anime loli" to know whether it is just "sexual activities" or more nuanced but I'm not risking a jail sentence either way, thanks. UKGOV is trigger happy on anything CSAM/fictional CSAM.)

[1] https://en.wikipedia.org/wiki/Legal_status_of_fictional_porn...

I think many people would find it reasonable to draw a line between victimless drawing and abuse material, regardless of whether it contains children, gore, violence, bestiality, rape, etc.

> Over a novel, non-standard definition of child pornography.

The problem is, there are jurisdictions in which loli and similar content can be classified as CSAM [1], and there are jurisdictions where e.g. the operator of a federation server can be held liable for facilitating access or for not blocking access to such material, even if the material in question isn't hosted at that server, such as in Germany (the case in question [2] was about software piracy, but the general legal principles apply just the same).

Legal systems worldwide haven't even begun to catch up with tech developments, and I think it will take at least a decade until regulations adapt.

[1] https://en.wikipedia.org/wiki/Lolicon#Legality_and_censorshi...

[2] https://rsw.beck.de/aktuell/daily/meldung/detail/bgh-kein-un...

So one of the nice properties about the fediverse is that tumblr "flipping the switch" wouldn't really do much of anything. You'd either have to explicitly follow a tumblr account, or have someone you follow boost a tumblr account, for it to show up in your feed.

I quite like how it's much more network-y than regular social media. Keeping your experience sane is much easier because you'll generally only see people you know, or people who people you know know. Reach intentionally has some friction to it. People have to keep vetting posts by boosting to extend the reach it has.

I actually never considered Tumblr supporting ActivityPub reads. Just publishing over the protocol. That is intriguing indeed, as a Tumblrina

By implementing ActivityPub wouldn't Tumblr become the overwhelming majority of the Fediverse?