If we continue to think in terms of social media platforms instead of protocols we’ll end up back at square one in enclosed communities with locked in social graphs at the whim of singular, fickle entities driven by profit or ideology.
For all the people who say ”I had to select a server, it was so complicated to sign up to Mastodon. On Bluesky it was easy” - how are these two different logic-wise? I mean what the hell?
Yea.. like I can read your toots here right now in the public. And bc of that there is less difference to #Bluesky if you think about it, but regarding the #atprotocol accessing them by firehose endpoint allows much more like building your own feed (algorithm) that extracts network-wide skeets for you... combine this with AI and you can read the things you're interested in... your own personal bubble. This and the global FTS is a killer feature for me over #Mastadon.
Question about webfinger: if I give it an actor @foo, it's going to query directly to bar.social, right? There's no way to somehow, register a proxy for something? I guess that would be a security risk, no?
My goal is to try to figure out whether I can make some kind of bridge for ActivityPub and ATProto, without touching either codepase directly, since making either one be able to talk to the other is a political issue right now. While code-wise I think it's possible to translate between them, at least with follows, likes, and posts, the only way I can think for people to actually interact with the other would be to have the bridge be itself a server, so you would follow a bsky account by following @username@bridge.net which would then forward to @username which is super cumbersome
> Le 12 septembre, le dérivé de Twitter construit sur un protocole décentralisé a annoncé avoir atteint le million d’utilisateurs. Afin de voir exactement ce que propose Bluesky, nous l’avons testé via son application mobile.
Sur les commentaires qui sont rigolos. Clochers, guerre, toussa toussa.
It describes it as:
“Autistic burnout is a syndrome conceptualized as resulting from chronic life stress and a mismatch of expectations and abilities without adequate supports. It is characterized by pervasive, long-term (typically 3+ months) exhaustion, loss of function, and reduced tolerance to stimulus.” ~Raymaker et al, 2020
Typically the Autistic person in question will still have multiple demands in their life that require cognitive resources, despite having little to no resources left to give. Life goes on, as they say.
I've been tracking how the number of posts per day on #Bluesky is going - looks like it's now at about ~400k posts per day from 70-75k users, which is around 4-5x what it was when I started recording data 2 months ago 📈 #ATProto#ATProtocol
(*) I think it's mostly furries though because my people aren't posting much 🙃
#Prediction: if #Threads follows through with #Fediverse support by using #ActivityPub, and especially if #Wordpress and #Tumblr join the party, #ATProtocol will be all but dead. AP will be the standard for social online, regardless of any benefits of AT. #BlueSky will effectively die with it, unless they switch to AP (unlikely as then they’d just be one instance amongst many). Threads joining the Fediverse would hurt BlueSky (and to some degree #Twitter) a lot more than it would hurt #Mastodon.
We don't usually think of it this way, but it's true: #ActivityPub does entrust identity to the server, not the the user. There were technical reasons for this at the time, & those continue, but it remains a fact which uncomfortably highlight how non-simple these questions are.
#ATProtocol, for example, THEORETICALLY places identity in the hands of users - but in practical terms, it's still in BlueSky's hands.
& NONE of these have truly portable identity. @ricmac]
@ricmac "There’s no way Meta would’ve wanted to join the AT Protocol or Solid, because in both cases they would potentially be handing over control of identity and at least some data to their users."
To me, this translates into "Meta chose #ActivityPub over the #ATProtocol or #Solid because it wants less freedom (as in independence/autonomy) for ist users"?
So I'm looking at playing with the data on #BlueSky's data fire-hose. They claim to have crossed 1 million app downloads, and supposedly have 290k users.
I was plugged in to the matrix for 48 hours now, and have seen interactions from 69083 unique users. Not bad, that's quite a good number of #DailyActiveUsers. I see activity from about 500-2000 users in five minute chunks, but on average it's somewhere below 1000.
On the code end. I've filtered out the language of each post, and it's an array for some reason. :blobfoxthink:
Sometimes it's
undefined (it still has text, so it's not a picture-only post)
["en"] for a user who only ever posted in Japanese. Is it a default?
Multiple languages at once. So is it a per-user parameter instead of a per-post parameter?
Angika, com, ckb, papafish, mai, ... and other non-ISO 639-1 codes. Are they not sanitizing this value at all? I would expect them to at least only allow correct codes, and null-out wrong ones. Is that why there are undefined language labels?
They could run some form of simple language detection code for this, but it's probably too computationally expensive. :ablobfoxhyperthinking:
I've found 1000+ users with posts about #anime and #manga and followed them. My feed is significantly improved. Because I used their #API class to programmatically follow them, I have accidentally followed myself, and can't unfollow me. #BugRequestSent.
I have found my people :fox_x3: , and managed to find their posts. But oof, the engagement of the anime/manga community on Bluesky is ridiculously low compared to other types of posts (lewds, begging, cats). They'd be lucky to get >10 likes/reposts/quote-posts. There were 2212 posts and 2122 engagements. The top posts >12 likes are mostly lewds of anime girls.
The reason for this, to me, is simple, because of the way BlueSky's DIY non-#Algorithm works, there's no proper Anime Feed. :blobfoxthinkanime: The list of discoverable feeds in-app or in-website literally stops at the top fifty. The API also does not support querying more.
So I have taken it upon myself to work on it. #MoreWork. :meowcrying:
@randulo From my understanding of the AT-protocol that is the idea. It seems to work similarly to ActivityPub with some design differences and a lot of missing features not implemented yet.
One thing that has me skeptical is how they aren't testing multiple servers or even distributing alpha/beta versions of server software for people to self-host to try and find bugs. If the project is intended to be open source, it doesn't really follow in that #foss spirit.
(CORRECTION: There is a sandbox open source version that anyone can try from their GitHub.)
I just read another news article about why mastodon didn’t make it and is dying. It’s very sad to hear and probably explains why I have to keep adding more server capacity to handle all the people quitting mastodon.
@joelmartinez#atprotocol is just #BueSky right? Or there are other implementations? I think there's still some question of whether blue sky even complies with it yet. I saw some saying it might happen by the end of the summer.
From a preliminary ~12h sample of #bluesky / #atprotocol , a small number of accounts receive most interactions. This is an obvious byproduct of the way the default algorithmic feed prioritizes posts.
5% of accounts received 72% of likes, 1% of accounts received 41%
The top 5% of accounts make 48% of posts
37% of accounts receive no interaction
The median account received 1 like.
Just a quick, incomplete look, but ya looks like an engagement farm
alright, I think it's time to translate the #NWB spec language to RDF and start on the first leg of getting this p2p thing off the ground and mirror all the neurophysiology archives. got a big stack of hard drives and bandwidth going to waste
the next stage is to get to full read/write interop where the servers can basically behave transparently like peers. the tricky part is how to make a spec for interacting with other servers that might use some complex mixture of auth methods and protocols without requiring an extremely complex client.
I had been trying to design something like #ATProtocol's lexicons for this, and seeing their implementation makes me understand just how much of a mess you can make that way - theirs is a difficult to understand blend of portable specs with code (yes you can spec that you can request posts from a server, but how do you actually implement the server logic without making the lexicon superfluous).
Their lexicon framework seems super slapped together though and basically ignored all prior art, like the rest of the protocol, so I have some ideas about how to make that actually work.
side note - I think it's super funny that #ATProtocol runs on IPLD CIDs but they didn't want to just use IPFS or some other p2p technology that actually makes sense to use there because, uh, reasons.
alright I'm going to boot up one of these #atprotocol graph indexers and start sampling the firehose. even doing that feels like an invasion of privacy but I gotta know what ya can do with that
I'm probably talking to a brick wall, but I want the #bluesky / #atprotocol devs to have to publicly answer for their design decisions that make the network either just another de-facto centralized attention market or a sprawling vector of abuse.
The sandbox environment has now been launched, which means you can set up your own AT protocol-enabled server.
It's possible that Bluesky will finally start federating by the end of summer. However, knowing how most dev teams work, this launch is not a certainty. It always takes time to iron out bugs.
How successful do you think Bluesky's federation will be? And how will it impact the current userbase using Bluesky?
Don't be naive, people. #Facebook is only interested in collecting more data and to connect that data to profiles to feed to the ad machinery. If #ActivityPub can be used for that, they will give it a try. That's it. That's all. It's still Facebook.
OK even funnier than #bluesky / #ATProtocol 's plc DID method relying on a single domain is that the hardcoded DID for bsky.app isn't registered, they don't even use it for the default algorithms and just hardcode a workaround.
WHICH MEANS that whoever finds a (truncated) hash collision for the bsky.app DID will cause an absolute fucking mess hahhahaahhaah
@blaise
but ya, here's another example where the #Bluesky / #ATProtocol devs seem to not really have planned for federation. talking about thread-level moderation, which would be really cool, but that is really hard to do in a federated context. ultimately to make that work you have to defer functionality to the feed generators (making federation pointless) or else just keep having a single server.
jo.nny.rip . 1h @jo.nny.rip this is another thing that is easy with single server but extremely hard in a federated context. OP deletes a harmful post. can you make the other server delete the post? should you be able to? OP locks replies. other servers ignore, issue replies. should people who follow those ppl not see them?
jo.nny.rip @jo.nny.rip . 1h you can make all sorts of features if you assume every server behaves nicely. what happens when they don't? ATProto's federation architecture is still largely unspecd, but can you do anything at the level of a server? what about the feed constructors ignoring your deletion? jo.nny.rip @jo.nny.rip⚫ 1h this thread seems to show another thing about the Algo feeds - they don't seem to work for getting new accounts integrated into any sort of social space. who is going to read a feed specifically of posts with no engagement? this is what instances do.