There have been several interesting discussions about what would happen if Bluesky PBC disappeared -- this thread from @phildini, for example, with great perspectives from @ted and @glyph.

Right now, 99.99% of the ATmosphere is on Blueksy PBC-hosted infrastructure, so there would obviously be a huge short-term hit. My take, though, is while things would be very chaotic for a while, there are enoguh independent infrastructure projects in the works that the network be able to start recovering relatively quickly. Blacksky and Spark are two of the independent projects that are farthest along, and I know a lot more about Blacksky than I do about Spark, so that's what I'll focus on. (You can find out more about Blacksky in Rudiy Fraser's Rudy Fraser's recent 🔭🖤🚀 Social media’s next evolution: decentralized, open-source, and scalable, and there are a lot more links here )

• Blacksky currently still has some dependencies on Bluesky's infrastructure; in the thread I linked to above, for example, Ted Han discusses the role of the PLC directory. My guess though is they could come up with workarounds fairly quickly if they have to -- and Blacksky's rsky-relay already uses its own mirror of the PLC directory, so it's not like they'd be starting from scratch. And there's did:web implementation, which doesn't depend on the PLC directory; this might or might not make sense for Blacksky, but other projects might well make more use of it.

• Blacksky currently only hosts around a thousand people on its own infrastructure, so there are likely to be growing pains. But the underlying AT Protocol architecturei is highly scalable, and Blacksky's been designed with a goal of supporting a few million people. That by itself is a lot smaller than Bluesky, but then again Blacksky's only one of many independent infrastructure providers, so they don't have to take the whle load.

• While anybody who hadn't backed up their data from Bluesky's servers before they vaporized might lose a lot of their posting history and collections, a suprising amount could potentially be recovered from services like microcosm and zeppelin as well as relays like atproto.africa (which keeps the last three days of posts)

• Bluesky PBC's platform-level moderation plays a key role in spam and malware filtering, CSAM scanning, and other stuff. Behind the scenes, though, they also outsource all their automated scanning to Hive, so independent implementations could do the same -- or find another solution.

Other projects aren't as far along with a complete parallel infrastructure stack as Blacksky, but in a crisis situation they could probably cobble somethiung together fairly quickly – either deploying stuff at an earlier stage they had planned(Gander, for example, might well get their elbows up and launch a few months ahead of schedule) or by using Blacksky's implementations.

All that being said, the devil is in the details, and there certainly are a lot of details to be worked out.

Then again I've been consistently impressed with the collaborative problem-solving approach of the AT Protocol developer community, and they'd certainly take this as a challenge to address.

#bluesky#ATproto #blacksky

There have been several interesting discussions about what would happen if Bluesky PBC disappeared -- this thread from @phildini, for example, with great perspectives from @ted and @glyph.

Right now, 99.99% of the ATmosphere is on Blueksy PBC-hosted infrastructure, so there would obviously be a huge short-term hit. My take, though, is while things would be very chaotic for a while, there are enoguh independent infrastructure projects in the works that the network be able to start recovering relatively quickly. Blacksky and Spark are two of the independent projects that are farthest along, and I know a lot more about Blacksky than I do about Spark, so that's what I'll focus on. (You can find out more about Blacksky in Rudiy Fraser's Rudy Fraser's recent 🔭🖤🚀 Social media’s next evolution: decentralized, open-source, and scalable, and there are a lot more links here )

• Blacksky currently still has some dependencies on Bluesky's infrastructure; in the thread I linked to above, for example, Ted Han discusses the role of the PLC directory. My guess though is they could come up with workarounds fairly quickly if they have to -- and Blacksky's rsky-relay already uses its own mirror of the PLC directory, so it's not like they'd be starting from scratch. And there's did:web implementation, which doesn't depend on the PLC directory; this might or might not make sense for Blacksky, but other projects might well make more use of it.

• Blacksky currently only hosts around a thousand people on its own infrastructure, so there are likely to be growing pains. But the underlying AT Protocol architecturei is highly scalable, and Blacksky's been designed with a goal of supporting a few million people. That by itself is a lot smaller than Bluesky, but then again Blacksky's only one of many independent infrastructure providers, so they don't have to take the whle load.

• While anybody who hadn't backed up their data from Bluesky's servers before they vaporized might lose a lot of their posting history and collections, a suprising amount could potentially be recovered from services like microcosm and zeppelin as well as relays like atproto.africa (which keeps the last three days of posts)

• Bluesky PBC's platform-level moderation plays a key role in spam and malware filtering, CSAM scanning, and other stuff. Behind the scenes, though, they also outsource all their automated scanning to Hive, so independent implementations could do the same -- or find another solution.

Other projects aren't as far along with a complete parallel infrastructure stack as Blacksky, but in a crisis situation they could probably cobble somethiung together fairly quickly – either deploying stuff at an earlier stage they had planned(Gander, for example, might well get their elbows up and launch a few months ahead of schedule) or by using Blacksky's implementations.

All that being said, the devil is in the details, and there certainly are a lot of details to be worked out.

Then again I've been consistently impressed with the collaborative problem-solving approach of the AT Protocol developer community, and they'd certainly take this as a challenge to address.

#bluesky#ATproto #blacksky

There's a whole thread happening over at Eugen's place re: what even is decentralization maaaan

So the thing is, in practice, AP isn't actually decentralized. What it's done is replace one center with many that are even less scrutable. This is not in any way an improvement.

I used to be a big fan of AP

Not being able to see art anymore on tech dot LGBT due to The Art Instance™️ being run by a tyrant was the first big crack in my liking the AP fediverse. It revealed how centralized it really is. Meanwhile Bluesky builds AT and all the tooling to moot the company and its potential power.

AT still has potential. AP is hard locked to a certain course unless some massive disruption happens internally. Its potential, timeline-wise, was where AT is now, not almost 10 years in. I don't see it happening in a world where it peaked at ~3m and doesn't seem to grow any. The million active users it has are the million it will have for the foreseeable future.

#ActivityPub#ATProto#Bluesky

There's a whole thread happening over at Eugen's place re: what even is decentralization maaaan

So the thing is, in practice, AP isn't actually decentralized. What it's done is replace one center with many that are even less scrutable. This is not in any way an improvement.

I used to be a big fan of AP

Not being able to see art anymore on tech dot LGBT due to The Art Instance™️ being run by a tyrant was the first big crack in my liking the AP fediverse. It revealed how centralized it really is. Meanwhile Bluesky builds AT and all the tooling to moot the company and its potential power.

AT still has potential. AP is hard locked to a certain course unless some massive disruption happens internally. Its potential, timeline-wise, was where AT is now, not almost 10 years in. I don't see it happening in a world where it peaked at ~3m and doesn't seem to grow any. The million active users it has are the million it will have for the foreseeable future.

#ActivityPub#ATProto#Bluesky