The basic building block of the Fediverse is the instance, right? Every instance is its own self-contained, centralized social media service that optionally interacts with other instances. EG Trump’s Truth Social is a Mastodon instance that does not federate.
ATProto takes a more radical approach. Everything is modular. There is no instance or anything that is complete in itself. It’s more like the WWW. You can make websites in different ways. These are made findable through search engines like Google or Bing, which are not affiliated with companies offering web hosting.
ATProto takes everything apart. It tries to avoid choke points or lock-in as far as possible to thwart monopolies. You have a server that stores your data (posts, etc …), called a PDS. You can move your data to a different server. An identity provider tells others where your account is at any moment. A relay collects all the posts that people make and makes them available for further processing. This can be used to create algorithmic feeds, or moderation (aka labelling). These things are independent of each other and can be independently offered by different parties. You can pick and chose which to use, though there isn’t a whole lot of choice yet.
ETA: No idea what W wants to offer in that regard.
Follow-up question : doesn’t ATProto run into the same issue as Bluesky for example, where even though the protocol allows for decentralization, the vast majority of users are on the same servers, thus negating the pros of decentralization?
I feel like this would be the case here where one server will contain all the modules.
The company Bluesky Social PBC created the microblogging service Bluesky and the ATProtocol. It’s like the company Mastodon created the microblogging service Mastodon. There are other services built on ATproto that are EG like Reddit/Lemmy. But these have not taken off significantly.
The ecosystem is mature enough so that you can participate in Bluesky without using services offered by Bluesky Social PBC and without making sacrifices. For the most part, you can move without abandoning your account.
It is true that the servers are mostly run by the Bluesky company, but so what? Email is mostly run by Google and no one seems to think that’s a problem.
So, my main takeaway is that ATProto basically splits user authentication and content storage and feed algorithm? Like, these are 3 separate things?
You can have your own authentication server at home that basically just says that you are you (similar to a cryptographic identity), then you can store your content on google cloud and then use a 3rd-party feed algorithm?
The option to self-host your identity piggybacks on the DNS-system. A certain domain name resolves to the server where you store your ATProto identity. As long as you control the domain name, you control your identity. The ATProto identity is simply a pointer to where you currently store your data that your followers/contacts can use to find your content (IIRC).
The non-selfhosted alternative is a central identity service run by Bluesky. Unfortunately, the identity cannot be moved for obvious reasons. It would be good if there were some more options there. In principle, if the ATProto identity was tied to the government identity, that would make it moveable and non-hijackable. For some people, celebrities and such, that would be a good option.
Regardless of whether you selfhost your identity, you can selfhost a Personal Data Server (PDS), which stores your data and makes it available to the network. The PDS can move, cause that’s what the identity is for.
For non-believers in that format of social media, is ATProto good or bad?
The basic building block of the Fediverse is the instance, right? Every instance is its own self-contained, centralized social media service that optionally interacts with other instances. EG Trump’s Truth Social is a Mastodon instance that does not federate.
ATProto takes a more radical approach. Everything is modular. There is no instance or anything that is complete in itself. It’s more like the WWW. You can make websites in different ways. These are made findable through search engines like Google or Bing, which are not affiliated with companies offering web hosting.
ATProto takes everything apart. It tries to avoid choke points or lock-in as far as possible to thwart monopolies. You have a server that stores your data (posts, etc …), called a PDS. You can move your data to a different server. An identity provider tells others where your account is at any moment. A relay collects all the posts that people make and makes them available for further processing. This can be used to create algorithmic feeds, or moderation (aka labelling). These things are independent of each other and can be independently offered by different parties. You can pick and chose which to use, though there isn’t a whole lot of choice yet.
ETA: No idea what W wants to offer in that regard.
Thanks for taking the time to explain.
Follow-up question : doesn’t ATProto run into the same issue as Bluesky for example, where even though the protocol allows for decentralization, the vast majority of users are on the same servers, thus negating the pros of decentralization?
I feel like this would be the case here where one server will contain all the modules.
The company Bluesky Social PBC created the microblogging service Bluesky and the ATProtocol. It’s like the company Mastodon created the microblogging service Mastodon. There are other services built on ATproto that are EG like Reddit/Lemmy. But these have not taken off significantly.
The ecosystem is mature enough so that you can participate in Bluesky without using services offered by Bluesky Social PBC and without making sacrifices. For the most part, you can move without abandoning your account.
It is true that the servers are mostly run by the Bluesky company, but so what? Email is mostly run by Google and no one seems to think that’s a problem.
Bluesky itself is behind ATProto
Make of that what you will
I didn’t know that.
So, my main takeaway is that ATProto basically splits user authentication and content storage and feed algorithm? Like, these are 3 separate things?
You can have your own authentication server at home that basically just says that you are you (similar to a cryptographic identity), then you can store your content on google cloud and then use a 3rd-party feed algorithm?
The option to self-host your identity piggybacks on the DNS-system. A certain domain name resolves to the server where you store your ATProto identity. As long as you control the domain name, you control your identity. The ATProto identity is simply a pointer to where you currently store your data that your followers/contacts can use to find your content (IIRC).
The non-selfhosted alternative is a central identity service run by Bluesky. Unfortunately, the identity cannot be moved for obvious reasons. It would be good if there were some more options there. In principle, if the ATProto identity was tied to the government identity, that would make it moveable and non-hijackable. For some people, celebrities and such, that would be a good option.
Regardless of whether you selfhost your identity, you can selfhost a Personal Data Server (PDS), which stores your data and makes it available to the network. The PDS can move, cause that’s what the identity is for.
Feeds and other stuff is again independent.
thank you. that’s indeed fascinating to contemplate about.