WS's example shows the quirkiness of even the best rank methods. They'll do
unintended things that we'd rather they didn't do. Approval has a stability
and behavioral simplicity that the rank methods don't have.
But the best rank methods, such as SSD, MDDA and MAMPO, offer strategic
advantages that are worth it. Well, the matter of whether you agree that
it's worth it determines whether or not you're a Condorcetist. I claim that
it's worth it.
Because MDDA and MAMPO have Condorcet-like critrerion-compliances, I include
their advocacy in Condorcetism.
I often liken Approval to a solid reliable handtool. Rank methods are
automatic laborsaving machines. There are a few top-qualitly automatic
laborsaving machines. Even the best such machines are quirky, but the best
ones work well if you accept that no automatic laborsaving machine is
perfect. But most of the automatic machines are a cheap, shabby piece of
junk like Instant Runoff.