Thanks for your response. In an attempt to write something short for once (this piece still went over 1000 words but oh well) I chose not to discuss some of the 'unconventional alternatives' I mentioned at the end.

Range voting, of which STAR is a variant, is one of these, and some very important metrics (see: Bayesian regret; https://rangevoting.org/BayRegDum.html) is superior to ranked choice.

Arrow's theorem only applies to situations where voters rank candidates in their order of preference (FPTP falls under this category as well, it just doesn't consider any preferences past #1). Cardinal voting systems, like range voting (I think STAR voting is a variant of this), can be both spoiler-free and strategy-free.

I agree with you that STAR voting would probably be more difficult to get people to agree to than ranked choice. The more convoluted the voting system, the harder it will be to implement.

Still, anything is better than FPTP!