I'm not proposing it as a rule within any system, though, and my analysis generally holds so long as it's true a meaningful chunk of the time - it doesn't even need to be a majority of cases.
The last thing we need, seen through that lens, is a complex system where people don’t understand the results or how exactly their vote will affect those results.
And this is one of the best features of approval voting compared to any form of ranked choice!
It is, but 1) the public needs to be sufficiently convinced that all of the bookkeeping reliably determines which candidate that applies to in the face of possible error or malfeasance, and 2) sometimes there is no such candidate (cycles are possible) and what happens then can be complicated.
(And sure, ties are possible in ~any system, but if we treat cycles as ties we've made ties much more likely.)
Where you get into trouble is that not all elections have a Condorcet winner. In particular:
> The problem with Approval is that we're right back at the Faustian bargain of needing to fully support the lesser evil at the expense of your preferred option.
I think the concern raised by PaulHoule is very important. Assuming, for the sake of this particular digression, that it can be sufficiently satisfied by IRV in a context under consideration, BTR-IRV seems like an easy sell over IRV. It's like IRV, but instead of discarding the candidate with the least first place votes, you do a head-to-head runoff (looking at the rank ballots you already have) and discard the loser. Not much more complicated to understand or administer, and you can never discard a Condorcet winner (or a member of the Smith set, except in favor of another member of the Smith set).
RCV, in the most common form of IRV, has a problem that I think is relevant.
If we accept the truism that the appropriate compromise won't be anyone's first choice, then the very first thing IRV does is throw out any appropriate compromise, leaving it to make a decision between inappropriate options.
I think we need to be more able to pick a compromise, not less.