the collusion protection is limited, and we would be retaining substantial trust in a central server. in an n-of-x model, we would be risking that n of x parties could reconstruct our key and sign events on our behalf not only this, but also if enough of the parties we chose are offline, we can't sign like ourselves. and if they don't come back and we don't have a separate backup, we risk losing access to our private keys indefinitely in addition to this, there's the latency issue. it may not appear important, but nip46 for me is a pain in the ass, and i've created my own local network bridge to handle these things without latency any millisecond that we add to these processes makes things more complex, more when we have apps that require actions like, idk, decrypting or sending big groups of messages. i understand that the cases where we do these may be limited, but an app is always as weak as the weakest link, right? using this model for login and recovering the private key to use in a local signer would solve the latency issue. but i feel at this point that's minor compared to the first issue: we'd still be risking that n of x parties could reconstruct our key, even if we're using it locally ?!