Thanks to async discussions with Stephan[^1], we have clarity on two pieces of the puzzle:
-
deleting a signer is no longer meant to invalidate old messages [proof] [discussion]
-
Key Type 2 is auth addresses, a relatively newer addition which allows Farcaster client to allow sign-in from addresses created by other clients such as TBA
Side note: my tone of writing is for my own amusement only. I appreciate everyone who feeds my intellectual curiosity, despite the snark
More snark, can't help it: it is no wonder this is confusing. You've heard of the common tech retort, "Feature, not a bug". This one appears to be one of those rare, "A bug, which we fix but revisit 5 months later and justify as a feature"
https://github.com/farcasterxyz/snapchain/commit/d2eaa289517ed7f1b899cea4eca9e7eab67381ce
Again, this is totally fine and normal with live software. Just amusing...
[^1]: creator of OpenCast, Yo, signers.steer.fun, and many useful FOSS projects
