An interesting detail on FROST:

"Note that in a distributed setting, each participant P i must be sure to have the same view of C as all other participants. In practice, implementations guarantee consistency of participants’ views by using techniques such as posting commitments to a centralized server that is trusted to provide a single view to all participants, or adding another protocol round where participants compare their received commitment values to ensure they are identical."

(from the 2020 paper).

I know this basically a meme, but in this case, publishing the data to the blockchain is not inconceivable... or possibly an opentimestamps setup, so you xould check your local view agrees with what is published? Not sure.

I'm keen to better understand these tradeoffs; FROST can be used even for N of N, it involves more data transfer in set up than MuSig, but it's a nicer mathematical structure. It's also less robust.