At 10:49 PM 7/23/2008, Kathy Dopp wrote:
Thought this list might be interested in this "real world" example of IRV.
It turns out that Fair Vote Director Rob Richie's home town of Takoma
Park Maryland, the home base for IRV, has Zero (0) zilch NADA minority
representation. And voter turnout flat out sucks.
IRV was a fish bicycle in Takoma Park and it is rather obvious that
it was only implemented there because of probably intense lobbying by
a resident, Mr. Richie.
Takoma Park has, more often than not, unopposed elections. An
election with three serious candidates is rare.
This is the big secret: IRV is being implemented in jurisdictions
with nonpartisan elections. With nonpartisan elections, IRV
duplicates, almost always, the results of Plurality. Voting systems
experts seem to have missed this entirely, probably because they
mostly think in terms of factions, which applies to partisan
elections, not necessarily to nonpartisan ones.
It would be speculative to predict this result from an abstract
consideration of how voters form and vote preferences, though
theoretical justification is possible. But it's an obvious phenomenon
when one studies the actual elections taking place.
There have been 32 IRV elections in the U.S. since 2004. Of these 9
went to Instant Runoff. In every one of the instant runoff elections,
the leader in the first round went on to win the election, and, in
every one, the runner-up remained the runner-up after vote transfers.
That is almost 18 opportunities to see vote transfers cause some
reversal. (In some of the Takoma Park elections, there was only one
candidate on the ballot, so "runner-up" doesn't mean much!)
There has been a lot of hot air generated about runoff voting (real
runoffs). But real runoff voting, about one time out of three,
results in a "comeback election," where the runner up in the primary
wins the runoff. Why isn't IRV doing this, supposedly it "simulates"
runoff voting?
But it does not simulate runoff voting, which includes an opportunity
to reassess the candidates with a narrowed field. In addition,
differential turnout means that the voting population changes. The
kind of voting systems theory that Range Voting is essentially based
on -- which considers preference strength to be important -- would
lead us to conclude that this improves results, as the decision to
vote or not vote is a presumably sincere decision, on average. When
runoffs have involved very high preference on the part of large
groups, such as the Lizard vs. Wizard election in Louisiana, runoff
turnout can exceed that in the primary.
So: the most democratic voting system we have in actual use, top-two
runoff (and it is even more democratic when write-ins are allowed in
the runoff; in Long Beach, the mayor was recently re-elected in a
runoff election in spite of ballot rules -- term limits -- preventing
her name from being on both ballots, the primary and runoff) is being
replaced by a system, that in nonpartisan elections, actually closely
simulates Plurality. Which is a decent election method, more so in
practice than in theory. It probably gets it right about 90% of the time.
The arguments being used for Instant Runoff Voting have been crafted
to be politically effective, not to be cogent and accurate. We see
that arguments which, in debate after debate, have gone down in
flames, continued to be advanced as if nothing happened, and this is
typical for a political campaign. If it works, if it creates support,
use it. Truth? What's that? It's all opinion anyway, right?
This is a very old argument, isn't it?
Who is actually looking at the real election results? Certainly not
FairVote! No, they focus on arguments they can win -- whether or not
the arguments are true or not, it only matters how they arguments
sound to the naive. We've seen, again and again, that legislators
don't take the time to actually understand how voting systems really
work. If they did, they'd happen upon a curious fact: Voting system
in use, all of them, could be drastically improved simply by counting
all the votes, instead of discarding and considering spoiled, ballots
with more than one vote per officer to be elected. All voting
equipment can do this (because it must be able to handle multiple
winner elections). It's just a matter of counting them.
These votes would solve the spoiler effect (because the spoiler
effect takes place when a minority candidate draws off enough votes
from a major party candidate, so that the less-preferred major party
candidate wins by a plurality. Spoiler effect disappears, *mostly*,
with top-two runoff, and probably with Approval just as well as with
IRV. -- those considering this tend to assume that all minor party
voters will vote for their favorite among the top two, but Optional
Preferential Voting in Australia -- much more like what is being
proposed here than the standard Preferential Voting which FairVote
uses to claim is the "Australian system" -- shows this not to be
true. Ballot truncation is common. My expectation is that you would
see roughly the same number of additional Approvals for major party
candidates with Count All the Votes Approval as with IRV. No method
guarantees a majority, not even top-two runoff if write-ins are
allowed. But majority failure is common with IRV (7 out of 9 of those
instant runoff elections) and rare with top-two runoff.
TANSTAAFL. You want a majority winner? You have to pay for it, with
occasional runoffs. My estimate is that with a good election method
(Approval or Bucklin, which I call "instant runoff Approval," it uses
the same ballot as the "IRV" Ranked Choice Voting in San Francisco,
but the counting is much, much easier, simple totalization), about
half of the runoffs, even with many candidates, will be avoided. One
could expect, in such a situation, about one election out of six,
roughly, to require a runoff. And in roughly one election out of
twelve, the first-round result reverses. Yeah, these statistics may
be inconsistent, the whole matter requires further study, and it's about time.
(IRV isn't a good method for this because it does not count all the
votes, it leaves some concealed, underneath votes for the top two.
This is also the reason why it cannot find a "compromise winner,"
i.e, why it experiences Condorcet failure under some conditions.)
----
Election-Methods mailing list - see http://electorama.com/em for list info
