[EM] Is IRV just a costly version of Candidate Proxy?

[Forest Simmons]

My Aussie acquaintances claim that in Australia when IRV (aka STV) is used
in single winner elections "most" of the voters fill out their ballots by
copying "candidate cards."

What does "most" mean here?  90% ?  99% ?

If everybody parroted candidate cards, then IRV would be equivalent to
Candidate Proxy, in which the voters just vote for one candidate and let
the candidate act as their proxy for the vote transfers.

If only one percent deviated from the candidate cards, that probably
wouldn't make much difference in the outcome. Would it be different enough
to justify the costs associated with ranked ballots?

What if 25% deviated from the candidate cards?

Let's think about when it might make a difference:

Suppose that you are voting under Candidate Proxy and your preferences are
   A>B>C
but A's preferences are A>C>B.

Do you vote for A or for B ?

If A has a better chance of winning than B, then you will definitely vote
for A, even though you disagree with A on the relative merits of B and C.

If B has a better chance of winning than A, then you still vote for A if
the B/C race is not close.

But even if the B/C race is close you might still vote for A if (in your
opinion) the difference between A and B is much greater than the
difference between B and C.

This leaves the case (***) where the race is close between B and C, and A
has no appreciable chance of winning, and you consider the difference
between B and C to be important, but you disagree with A on which of these
two is better.  In this (relatively rare) case you might hold your nose
and vote for your compromise B instead of favorite A.

Of course, under IRV you would have the freedom of expressing your true
preference order, but in the case of a close race between B and C, even
under IRV it would make no difference in the outcome if you voted
                A>B>C  or  B>A>C .

In other words, by use of a small amount of strategic voting by the few
that are compelled by their conscience to abandon their favorite in favor
of compromise (i.e. only some of the minority who would not use the
candidate card under IRV as in case *** above) Candidate Proxy gives the
same outcome as IRV.

So, is the added expressivity of IRV worth the cost of the more complex
ballots when most voters just copy candidate cards, and the outcome of IRV
and Candidate Proxy is the same anyway?

I believe the answer is "no," but for some folks the answer might depend
on the meaning of "most" in the above sentence.  For their sake it would
be nice if someone would carefully collect some data in countries where
IRV is used.

However, keep in mind that one of the most frequent objections to IRV is
the cost (including the effort) of (filling out) less simple ballots.

This may seem quaint to those of us who study election systems and take
ranked preference ballots for granted, but remember that we are dealing
with Joe Q. Public.

Forest