The problem I run into is that RCV can be nonmonotonic, where increasing a candidate’s ranking can cause them to do worse and vice versa. For most elections this doesn’t matter because the vast majority are uncompetitive, but it’s the tight races where whacky things can happen. Occasionally RCV will fail to elect the Condorcet winner, who (when they exist) is the person who wins every head-to-head matchup.

I would agree that more major expression is better, except we’re seeing evidence that even RCV is complicated enough to disenfranchise poor people at a disproportionate rate, something that doesn’t happen under FPTP. The voting system needs to be simple enough that that doesn’t happen, and we’re lucky that Approval Voting happens to be very good at electing the most popular candidate. It’s essentially a simultaneous approval rating poll, afterall.