[EM] Condorcet vs Range (Greg Nisbet)

[Aaron Armitage]

I would suggest that majoritarian voting reveals utilities better than
simply asking, because having made the effort to show up at the polls at
all there is no marginal cost to overstating how much utility (or
disutility) you expect from a given outcome. Which is why, as everyone
keeps pointing out, a rational voter will give only maximum and minimum
ratings. But cutting a deal does reveal utilities because there is a real
cost: the strength of your utility in a given preference is the same as
the disutility you would accept in exchange for satisfying it. This is how
the economy works: my utility in having food exceeds the disutility of
having to work.

Every candidate is a bundle of policy positions and other characteristics,
which different voters will more or less strongly like or dislike. Each
candidate, in short, is a potential compromise. Rating one candidate over
another means choosing that trade-off over the other. But Condorcet does
this better than range and approval, because one or more particular set of
voters, while preferring the range/approval package to the average
offering or (if they use optimal strategy) to the top two candidate whom
they prefer less, and who therefore obviously see some good in that
winner, may willingly sacrifice that good for what they consider the
greater good in a different candidate. If there are enough of them to make
the Condorcet winner different from the range/approval winner, then by
definition people are not getting the characteristics they value more, and
overall utility is suffering from the lack of the clarity that a distinct
rating from best to worst of all possibilities would provide.