Recent searches
Search options
Administered by:
It's interesting to hear directly from the CEO of #BlueSky why they have chosen to reinvent the wheel instead of building on top of ActivityPub. The gaps in ActivityPub, according to them, are:
1. Account portability
2. Global discoverability
3. Composable/customizable curation/moderation
I'm not sure if (1) is a big issue. I mean, the only problems I see with the current approach are:
1.1. If the admin turns off the instance or suspends you and deletes your data, then you don't have a way of migrating your account. But how would BlueSky solve this problem?
1.2. Your followers/followed accounts won't be migrated with you upon migration. And this is definitely something that could be worked on and IMHO shouldn't be that hard to implement. Once that gap is closed, are there any gaps on (1) that I'm still missing?
On global discoverability, I strongly agree. Of course there are reasons why many people on the Fedi don't want global discoverability, but I also believe that the whole Fediverse shouldn't be held back by the needs of a few, and a customizable level of discoverability should be possible for those who want their accounts and posts to be indexed - preferably built-in in the protocol and in the clients instead of relying on external services. Open directories of accounts are trying to fix that gap, some instances have implemented their own search, but things are still in their infancy - however, I definitely see something moving compared to a year ago or so.
Composable/customizable curation and moderation...I'm not really sure if I understand this one. Customizable curation hasn't even been implemented on BlueSky - even though they are promising to provide a "marketplace of algorithms", nothing is there yet. And I have no idea of what they mean by customizable moderation - did I miss some insights from the protocol specs?
It's also interesting how she omitted answering the question "will BlueSky/ATProto be compatible with ActivityPub?". That's really a deal-breaker for me. If there's already an open standard that has been worked on for years which runs a distributed platform with >10M accounts, and you're just creating something new, then back-compatibility should be a no-brainer, or we end up creating yet another competing standard within its own walled garden.
@blacklight I know this is not a popular opinion, but I actually feel like the lack of actual account portability is one of my biggest annoyances right now. And I think it is holding back progress. Like, I feel like the ability to switch hosts was one of the selling points for me on the fediverse. But, at least as far as Mastodon stands, people can't even move their posts. Which for a lot of people is important.
It's been a while since I looked into the specs, but I seem to recall #BlueSky was talking about keeping a synced local copy of all your data so if your server goes down, you can just upload your copy to a new one. I'd rather we look at what #Zot has done with nomadic identity. It seems to have been working fine for like a decade already.