So, FAIR essentially replaces one single point of failure (Automattic) by another (Bluesky)?

[dertuxmalwieder]

I might just grossly misunderstand the protocol('s documentation) here, so please enlighten me:

Generally, FAIR packages use PLC DIDs, as they are more powerful and more secure than Web DIDs.

However (emphasized by me):

PLC DIDs are IDs that are registered in a public ledger, operated as a free, open service by Bluesky.

So, if Bluesky (which is basically a company, run and funded by millionaires, not that different from Automattic) decides to just shut down that public ledger for any reason, what would be the benefit of FAIR?

[rmccue]

This is a great question!

PLC is a Public Ledger of Credentials, so the ledger is fully public, and can be mirrored to other sources (see the Audit Log docs in the PLC spec; this is also available for any individual DID on /:did/log/audit). There are already existing mirror projects for it (for example, here's a mirrored FAIR DID, and we may run one of those in the future as well. In the event Bluesky went out of business and the PLC directory was at risk, we could switch to a mirror.

Bluesky have stated publicly (and in direct discussions with them) that they intend on moving this into independent operation in the future as an web infrastructure component. Blueksy are also a Public Benefit Corporation rather than a regular for-profit corp, so there's a degree of difference between them and a pure for-profit.

Additionally, we plan on supporting additional DID methods; in particular, we're planning on supporting Web DIDs which only use DNS/HTTP. These have downsides and don't solve the data portability issue, so it's up to vendors to decide on that tradeoff. Certainly, we don't want to have a hard dependence on PLC DIDs only.