Also for the protocol a non-Bluesky client can handle new account creation, but then that client either has to target users that host their own PDSs already or host them for them, and seeing as a PDS includes all the media (not just text like a nostr relay) that’ll add up. As such most clients currently in dev will ask users to create the account on Bluesky (the client) so Bluesky (the company) will in effect subsidise the cloud storage and networking. The vision is for third-party hosting to become cheap and easy so other clients can integrate with these third parties and offer account creation for users who have no interested in begin associated with Bluesky the client.
For example on the DID side they say they'll spin out the directory into an ICANN like non-profit outside the orbit of their entity. Will they or won't they, time will tell. (For many concerned ICANN level decentralisation is fine, though of course not for everyone, hence the emergence of handshake.org domains and all that.)
Also they envision a protocol ecosystem with many clients, each with its own moderation policies (or none). Right now Bluesky (as in bsky.app) is really all there is to talk about client-wise, but there are others clients being built. A couple examples:
- https://ucho-ten.net (Japanese)
There is nothing stopping someone today from developing a client on ATprotocol with zero moderation--so Bluesky the client could be moderating quite heavily, have a full-time team for that, but for some other client in the wider protocol ecosystem anything goes.
And they envision their god-relay being deployed by others with the means, perhaps AWS, Google Cloud, and Azure deploy and offer access as a managed cloud service) or perhaps a few non-profits of the Internet Archive variety. But some others, whoever they turn out to be. In theory that keeps everyone honest, but it comes back to the first point of do they spin out the directory or not.
So right now the directory is in-house, no other client besides their flagship-demo (as they see it) has traction to speak of, nobody else has deployed the god relay or announced plans to. Though the stated vision is supposedly for all that to change.
Discussion
see that ... subsidise... they have no notion in mind that the users will fund it
this is the root of why i keep on saying that nostr needs to have clients with full auth support and smart access controls that let people find each other and probably some of the paid relays want to also actively gather data from other nodes to keep things churning