Range voting encourages honesty (executive summary)
Theorem:
In essentially every voting system based on rank-order ballots, there is always a
3-candidate election situation (example)
where casting a vote that dishonestly says A>B when you honestly feel B>A,
is strategically better than casting an honest vote.
Contrasting Theorem:
But in range voting, that never happens – being dishonest in this way
never pays in 3-candidate elections.
Another theorem:
In range voting, it also never pays to "betray your favorite" by giving him anything
other than the top score. (Sounds simple, but this unfortunately is
not true in many other voting systems,
including
plurality
and
IRV.)
So range voting encourages honesty in 3-candidate elections better than every
system based on rank-order ballots.
Experimental fact:
In the USA's 2000 presidential
election, voters who thought Nader best were urged not to be strategically
stupid and instead to vote (dishonestly) for either Gore or Bush, and
polls showed that about 90% of the voters who thought
Nader was best in fact did vote
for somebody else.
[Another, different sort, of example – from p.65 of
Lakeman & Lambert's book:
a 1950 Gallup poll showed 38% of British voters wanted to vote Liberal but
only 9% did.]
Experimental fact:
In the USA's 2004 presidential
election,
about 3/4 of range voters (in a range voting exit poll of random voters)
chose to vote in a style which did not
award the max (99), min (0), or X (intentional blank)
score to every candidate. In other words the fraction of range voters who choose
to sacrifice some strategic oomph in order to be more honest,
experimentally is enormous.
The consequences of greater voter honesty are huge – check out the
enormous experimental differences
between the vote totals for the third-party candidates
under range, versus under approval and plurality voting.
There is a good reason to believe that
range voters will be both honest and strategic (both in large fractions) as far
as the eye can see 100 years into the future with range voting.
A 50-50 mix is probably a good rule of thumb.