Egregoros

Signal feed

Timeline

Post

Remote status

Context

2
@phnt i think they really just mean "we want egirls on here with a million followers" and i guess, yeah ok, but why? as bao said, the fediverse successfully kept normie npcs away by having two @s instead of one.

i think that's one of the fundmental split personality issues in the fediverse. people say they don't want normies on here, but they also say that normies are retarded for not coming here because it's clearly objectively better. it's either / or. is this a secret club or is this a 200 million user network?

Replies

50
@lain @phnt

artificial virality and fake trend manipulation are how a subset of creators make a ton of money on centralized platforms. thinking you could be recipient of that drives a bunch of the lesser creators who toil on the edge of quitting. the platform shapes the output completely for its profit. we don't have that mechanism so why would a creator come here instead. every time they try they find out their organic audience is a lot smaller.
@sun @phnt i think that's a sort of adversarial framing already, tons of people have millions of real fans and real communities that care about them. why would they switch to a network that's hostile to them? and if you tell people 'well the advantage of this vs twitter is that nobody you care about is here, but you can run your own server with zero users', that's a hard sell as well.

coming back to bao, she wrote that the fediverse is 'just twitter', and i think she's very wrong about that. not even pleroma is 'just mastodon'. i can detect the software of the user on the other end of a fedi thread with high certainty just from their posting style. all these seemingly inconsequential technological and sociological differences create different communities and we should probably not try to social engineer them too much.

i think we should keep building, but i don't think we should keep building for audiences that don't want to be here and that we don't want here. and i think we should completely ignore all foundations, standards bodies and other weirdo commies who want to steal our precious bodily fluids
@lain @phnt they have real fans but the platform is actually a star maker system whether or not you deserved that audience. it works the other way too, if you use youtube and you rack up a million subscribers but they don't like you, they'll just suppress your trending and won't show your video popping up even to your subscribers. my point is its all fucking fake. its not fake here but its a hard sell for creators that are benefiting from the fakeness which is every creator you've heard of
@sun @lain Mastodon has trending hashtags, but I think they are network wide, for the "known network" of that instance. Hashtags could have been really great for trending and tagging posts, but Mastodon ruined it by making the search utterly unusable. Ideally there would also be "local trending tags" and "global trending tags", so an instance has more of a community feeling besides it just being a way to communicate with the whole network. The "instance" being a generic thing instead of a community building block hurts this network I think.
@i @lain @sun I fundamentally disagree with how groups currently work and it's probably a good idea barely anybody knows that Pleroma has them. They are a prime target for uncontrollable spam. Kinda like relays, but the effects can get delivered directly to a user's sometimes even home timelines.

If I were to design groups for AP, I would make a Group Actor with a posts OrderedCollection and a list of Actors who can manually Add and Remove Objects to the Collection. With the ability to set certain instances/Actors as automatic approval. You POST to an inbox, if you are in the auto approve list, the instance automatically adds that Object to the Collection. You aren't in the auto approve list, you get put in a mod queue. Viewing that group would be simply fetching the first X number of pages/posts in it, which can remote instances keep in some cache.
@i @lain @sun Because the posts aren't from the Actor, it isn't their Objects, then why should have an Actor have Objects of someone else in their outbox. AP is extensible to stupid levels, why abuse an endpoint with data it shouldn't have just for the sake of reusing it.
@i @lain @sun Here's even a use case for the outbox: The Group Actors outbox has Add/Remove to it's own OrderedCollection to indicate moderation decisions for that Group.

Because if something should be removed and the instances of the admin making that decision and the receivers aren't in a follow relationship, how else would you federate removal of posts from the Collection without re-fetching it.
@phnt @lain @sun all i'm saying is that a shared /outbox is better, also fuck the concept of a moderated per server invite only group, it's never going to happen and authorized fetch is a thousand years of needless suffering
@i @lain @sun Who said invite only? I agree, a requires_approval true/false field would also work and would need to be exclusive with the manual/automatic approval arrays in the JSON. Want a public group open to anyone, fine, but you can at least still moderate it unlike an auto-repeater.
@phnt @i @lain @sun honk understand pleroma chats tho. I don't think we should cooperate, mastodon sure won't support jack shit (see: emoji reacts); lemmy/peertube/writeas have completely different use case, gotosocial/iceshrimp etc lack features that need to be implemented first; misskey just like mastodon, live in their own to timezone (don't support multiple emoji reacts per user) + language barrier.

We can spend years deliberating how to do the thing and waiting for others to cooperate or we could do some basic MVP, iterate on it and hope others will follow suit.

I don't think Pleroma Chats are bad, they are just underimplemented and lack groupchat support because no groups I guess.
@hj @i @lain @sun I disagree.

Look, Pleroma is in no position to bake a feature, and say "take it, or leave it" and expect everyone interested to take it, or pull a Mastodon and force it on everyone through, the shear number of users. If you want to do groups properly, for which there is interest in the FEP community and also in the Mastodon community which used guppe quite a lot. I think it would best not to half-ass it and deal with the consequences later.

This mentality is how we got to the point where there are 3 different ways to federate emoji reactions (Pleroma EmojiReact, Misskey's _misskey-emoji and someone else already did Like with an emoji field in JSON). Same issue with quotes, Akkoma/Pleroma quoteURI, Misskey _misskey-quote and now Mastodon's Quote Activity. How did we get here, nothing was ever documented and no specification how it works written so someone that lived in the Mastodon vacuum had no idea that Pleroma does EmojiReact.

That's the whole point of FEPs. It's a place for discussion about AP extensions for people that care. There's no process to get in besides write a spec and implement later. If we want to fix this messed protocol and do it properly, collaboratively, and not waster our time, this is the way to go. It doesn't have take years, because the window for a FEP to be "finalized" is a year after the last update or there abouts. It won't take literal years.

I don't think Pleroma Chats are dead, nor do I think they are bad, but there is no documentation on how they are supposed to work in AP, so very few implemented them. And how will you notify others that they exist if not by making a FEP and a somewhat high frequency issue tracker and announce them on activitypub.rocks.

cc @silverpill just because I've seen you talk about how good groups could work and I think my post above isn't completely stupid. And also sice FEPs got mentioned.
@phnt @i @lain @silverpill @hj I thought FEPs was servers implementing things and then telling you in a document how you did it, for example Smithereen has a mechanism for doing a myspace-style wall. It's not great and I'll be doing it in a different way completely if I ever do it for egregoros (but will be following a different better fep like public-appended collections)
@sun @i @lain @silverpill @hj They way I've seen them work is that you have a draft spec, a draft implementation and publish that as a draft. If someone is interested, they will comment on it and eventually a discussion will form about how it should be done. Basically W3C that isn't gatekeeping everyone and isn't actively hostile.

You can do it by implementing first, publishing later but then you are in the same place in my first two paragraphs. You failed to force it on everyone and someone doesn't like it? They will write their own FEP and now you are in the same mess like before. Competing over the same thing, because you didn't talk before.
@phnt @i @lain @silverpill @sun I'll discover them on fediverse. Every proposal needs something to back up, otherwise it's pure speculation. Implement groups - see results, propose it as standard - adjust as discussion goes or trash it in favour of another.

It's like "hypothetically if we had a public transport submarines what speed limit on them should be?" Well just make the damn submarine first then we can talk speed limits.
@hj @i @lain @silverpill @sun
1. I just thought how groups could work better literally in this thread
2. There are more pressing issue to resolve than reimplement Groups at least for me (like install docs being largely broken, still no substitute for OTP releases, releases CI still not rewritten, BBCode parser having a bug that annoys me greatly)
3. Not enough time and motivation at least from my part
@hj @i @lain @silverpill @sun You don't see much wrong with that many ways of doing the same thing, because you don't have to deal with it. Do you think it's great to write 4 different ways to normalize one thing from multiple ways it has been done.

Look at Pleroma's Transmogrifier if you want to see the shear amount of stupidity on this network.
@phnt @i @lain @silverpill @hj there was already a way to do federated badges before I implemented it for egregoros but the way it was done before was bad and I did it right. I want to believe this is different than usually someone does it bad and then someone else also just does it bad in a different way because they just plain don't care. I considered, then rejected the previous implementation (mine are 100% compatible with the w3c spec for openbadges while the previous one was hacked up partial compat with the old open badges spec and mutilated to make mastodon happy)