arcanicanis on Nostr: Firefish is unusual in that it can import posts exported from Mastodon. You can ...

Firefish is unusual in that it can import posts exported from Mastodon.

You can export/import posts under the Pleroma lineage of server implementations too, I believe even from an export from Mastodon.

A DID is securely associated with a document describing the subject (similar to a user or actor) identified by the DID, but without requiring a single, centralised authority.

I guess I view the DID specification as more of an extensible URL-like format, whereas the DID method dictates the general meaning of the remainder of the identifier. There are some very unusual DID methods listed in the registry that sound very centralized, just like Mastercard's did:id method: https://github.com/Mastercard/did-methods/blob/master/id.md#retrieve

Just like how HTTP is used/abused as an application protocol far beyond the scope it was intended for (hypertext media) just because of the versatility of URLs, I assume DIDs are intended to be a more broadly versatile/extensible format of URI-like identifiers, which could be for anything from identity, down to individual resources associated under that identity.

for example if the instance is offline or the user has been banned. This feature was cited as one of the main reasons Bluesky did not adopt ActivityPub.

I honestly would argue it's just an excuse on their part. I still can't find anywhere in ActivityPub spaces where they even tried talking to developers in the community. Also, you could actually Move your account to another service, while your original server is down, if you had 'your' private key that's used in HTTP Signatures. But because of the present state of how HTTP Signatures are sometimes implemented, servers actually keep you from exporting your key: https://github.com/mastodon/mastodon/discussions/22315

suggested a proposal to implement ActivityPub on top of the AT Protocol

Last I remember reading of that proposal, it felt like an insane idea of just stacking things, and not thinking carefully. I believe much of the same can be achieved far more simply with portable object signatures, and a duality of cryptographic identifiers that can be resolved by from conventional mechanisms (WebFinger-resolved acct: IDs, URLs, DNS, etc that point to a cryptographic ID) rather than one type of identifier only (only cryptographic or only conventional).