http://irvfactcheck.blogspot.com/p/there-is-lot-of-miss-information.html
<http://irvfactcheck.blogspot.com/p/there-is-lot-of-miss-information.html>Myths
and Facts about IRV
There is a lot of misinformation bouncing around on the Internet
about instant runoff voting (IRV) in the guise of "researched
analysis" that claims to show it has serious flaws. By and large,
such "analysis" is faulty, selective, agenda-driven manipulation. We
have attempted to gather together as many of these false claims and
myths as we could on this FAQ page to provide factual information.
IRV has been known for much more than a century to have "serious
flaws." FairVote has been known among students of voting systems, and
experts, as a source of serious misinformtion about the topic of IRV
and other voting systems. They are political activists, dedicated to
a narrow goal, the "agenda" of the critics is simply the truth about
voting, they are not paid for "success," almost all of them are
utterly disinterested -- except for enthusiasm for improved voting
systems. Some are partisans of this or that system, true, but not at
the cost of fact and sound analysis. The people at Rangevoting.org,
in particular, would rather fall into a deep pit than put up
misleading information, and you will find criticism of Range voting
there. But also election science, serious analysis, and a massive
collection of information about IRV that you will never find at
FairVote, an organization dedicated to its narrow goals that is
utterly unscrupulous about presenting misleading and deceptive
information, that will use every rhetorical device in the book to
convince voters and decision-makers to decide what they want and
promote. Let's see how many of these trick they pull out of the hat this time.
VOTER EXPERIENCE UNDER IRV
1.1 Is instant runoff voting simple for voters to use?
Yes. The voters' task is very simple. Voters can simply mark their
ballots in exactly the same way as they always have in the past.
However, the voter has the option of ranking alternate choices, in
case there is no majority winner and the voter's favorite candidate
doesn't make it into the final runoff count. Since a vote for a
minor candidate won't be wasted, as long as the voter ranks other
choices, the voter can generally avoid the conundrum of voting for a
favorite or a lesser evil. For many voters this makes voting with
IRV easier than under the current plurality method.
U.S. cities that recently adopted IRV have had
<http://www.fairvote.org/?page=2170>exit polls to assess voter
acceptance of the new system. Without exception, in every city
voters have overwhelmingly favored IRV over the old method. Also,
studies of the San Francisco and Burlington IRV elections have
proven that there was no increase in uncountable ballots (spoiled or
skipping the IRV race) with the adoption of IRV. In the Burlington
IRV election in 1996, for example, 99.9% of ballots cast in the IRV
race for mayor were valid. People had no difficulty voting, and
there were almost no "spoiled" ballots.
Two nations with the highest voter participation rates in the world,
Australia and Malta, both use instant runoff voting. The only
"complicated" aspect of instant runoff voting is the tabulation that
occurs if there is no initial majority winner. But the voters don't
need to absorb these details. A voter can dial a telephone without
understanding the intricacies of the internal electronics or vote
for president, without understanding the intricacies of the
Electoral College, which has a provision for election by the House
of Representatives when no candidate has a majority.
For *some* voters, IRV is simpler than under "the current plurality
method." But what FairVote won't tell you is that in most cases, IRV
is not replacing plurality, it is replacing "top two runoff." No,
they don't sensibly vote "exactly the same way as they always have in
the past," especially if the "past" was with top two runoff. One of
the FairVote tricks is to conflate multiple issues, picking and
choosing from each what they can present that makes IRV look good. If
there are facts in a poll that would indicate IRV is easy to use,
they will report it. If there is a fact that looks bad for IRV, they
don't mention it. IRV effectively treats as "spoiled" all ballots not
containing a vote for the remaining candidates, usually the top two.
That is, the result and the final count, including the phony IRV
"majority" is exactly as if the voter simply scribbled on the ballot.
Voting under IRV without understanding how it works can be quite
hazardous to your satisfaction with the result, and, indeed, are
these "facts" or are they simply polemic, made-up arguments intended
to convince the reader without actually informing him or her?
"No increase in uncountable ballots?" Lots of IRV votes are never
counted. Suppose a voter voted A>B>. As it happens, B is eliminated,
but if the B vote had been counted, B would not have been eliminated.
Then, if A is eliminated, perhaps C wins. I'd call this an "increase
in an uncountable ballot." It was actually a clear vote for B over C,
and it wasn't counted.
By the way, what's the "myth" here? Critics of IRV, claiming that
there have been problems as to voter education, have cited studies
and specific results. What FairVote has done here is to simply pick
up on other alleged facts (really mostly just arguments) that they
can cite. As usual, it's possible to come up with lists of arguments
on both sides, and you really do have to compare all the arguments,
together, to come up with sound conclusions. I'm not dredging up the
problems, here, so much as pointing out how the arguments presented
here aren't the kind of facts that check other facts, as implied, but
simply cherry-picked individual facts and general polemic. This is
what you get from political activists: lots of argument for their
position, claims that any opposition is deceptive, but no actual
balanced examination.
From someone initially friendly to IRV -- this was true of a lot of
anti-IRV writers and certain activists -- I gradually became so aware
of the problems that I would now vote to retain plurality (probably)
over IRV, and the reason is that with plurality, the problems are
very easy for voters to understand and recognize. IRV makes it very
complex, it only is made to *seem* simple by using false analogies
and other misleading evidence.
With the assertion about "voter participation rates" in Australia,
FairVote has completely outdone itself. Voting is mandatory in
Australia, it is a legal offense not to vote. What they aren't
telling you is that "donkey voting," where a voter simply marks all
the candidates in sequence on the ballot, down the ranks, is quite
common there, so common that the use of "Robson rotation," which
randomizes the sequence of candidates on the ballot, is absolutely
necessary. They also, note, don't mention ballot spoilage rates. If
you don't rank *all* the candidates, in most places, it makes the
ballot "informal," i.e., spoiled. This is a truly beautiful example
of misleading polemic. It is unwise to trust any argument at all from
them. For a long time, I accepted as fact much of what was on the
FairVote web site. And then I decided to check sources and look at details.
Perhaps they aren't any worse than other political activists. But
with IRV and election science and history, it's often possible to
check. They tell such whoppers.
1.2 Is instant runoff voting prone to strategic voting?
No. There is another side to voting that makes IRV simpler than most
other voitng methods, such as Plurality, Approval, Range, etc. While
every voting method is potentially subject to manipulation by
strategic voting in some situations, IRV is uniquely resistant to
such strategy (see
<http://econ.ucsb.edu/graduate/PhDResearch/electionstrategy10b.pdf>James
Green-Armytage analysis for more) Under most voting methods, a
potentially beneficial voting strategy can be recognized by at least
some voters (who may gain an advantage over other voters). Thus,
voters may face a dilemma deciding whether to engage in strategic voting.
For example, in a simple plurality election there is the
"lesser-of-two-evils" problem where voters often realize that voting
for their favorite candidate, who is likely to get very few votes,
may deny their second choice candidate enough votes to beat out that
voter's least preferred choice. Such a voter might find the
plurality election decision to be extremely difficult. Under IRV,
this voter's ballot will automatically transfer to that second
choice candidate if the first choice is at the bottom, and the
second choice candidate can win with that transferred vote. This
makes the voter's task much simpler with IRV. The default strategy
is always to honestly rank candidates in the order of choice. Of
course, voters may still face strategic dilemmas with IRV in some
rare situations, but these are far less common than under most other
voting methods. For example, under Range and Approval voting, giving
any support to a second choice may cause that voter's first choice
to lose (violation of the Later-No-Harm Criterion), causing some
voters to strategically truncate their true preferences. IRV
complies with the Later-No-Harm Criterion, and is thus immune to
such strategic calculation.
What they are doing here is framing questions to create the desired
impression. In certain ways (and not in others), IRV is less affected
by strategic voting, in actual practice. The link to James
Green-Armytage is dead. Perhaps this is what they had in mind:
http://fc.antioch.edu/~james_green-armytage/vm/svn.pdf The question J
G-A studied should be considered carefully. "Strategy" refers to a
voter affecting the result favorably (from the point of view of the
voter) by voting insincerely, which means different things with
different voting systems, so it's tricky to compare. With IRV, this
would be, for example, a Republican in Burlington obtaining a better
result by voting for the Democrat than voting the true preference,
for the Republican over the Democrat. James G-A was studying, with a
simulation method, *how often* it would happen that a voter could
have improved the result by voting differently. If it's possible less
often, he concluded that the method was "more resistant" to strategy.
He used a random issue space simulation to come up with expectations.
And, unfortunately, this does not at all consider the effects of
partisan elections. James G-A is also aware of the problem of
defining a "sincere vote" in Range and with Approval voting. He used
a method of deciding what the "sincere" vote would be that is
radically different from what real voters will use in real elections.
To make a long story short, "strategy" means different things with
different voting systems, and the comparison that J G-A makes that
shows, with his methods, "alternative vote" (IRV) being superior to
other methods is an artifact of how he defined voting practice. In
particular, the "sincere vote" that he set up -- and he is then
looking to see if the Range or Approval voter can improve the outcome
by voting differently, and if the voter can, then the method is
"vulnerable" to strategic voting in that election. Since voters will,
in real elections, vote far more sensibly than the vastly
oversimplified algorithm J G-A sets up, this result was garbage-in,
garbage-out, as far as comparisons between methods are concerned.
Part of the problem here is the definition of "strategic voting."
When Approval voting was introduced as a serious modern proposal by
Steve Brams, he announced it as a "strategy-free" system. But here is
J G-A proclaiming that it is highly "vulnerable" to "strategy." What
happened? Was Brams lying or seriously deluded? No, the definition of
the word was pulled out from under him by the proponents of ranked
voting systems. With pure ranked systems -- like IRV -- the only way
to vote insincerely, "strategically," is to *reverse preference*.
Reversing preference is never advantageous in Range Voting or
Approval, but strategic voting under IRV requires it; this is also
called "Favorite Betrayal." In order to hang the tag "vulnerable to
strategy" on Approval, it was necessary to change the definition.
James G-A is quite aware of the difficulty, he discusses it in the
paper, that there is no single "sincere vote" in Approval. Rather, in
approval, the voter, with full sincerity, groups candidates into two
classes, approved and not approved, and there is never any advantage
to putting a candidate in the disapproved category who is equally
preferred to or more preferred to a candidate in the approved
category and vice-versa. One would *always* put into the approved
category the favorite, and always put into the disapproved category
the least-liked. But suppose there are three candidates, and one is
"in the middle." Where does one put this candidate?
In ordinary language, this is a "strategic" decision. How one makes
it is quite the same as one makes *any* kind of compromise. If I ask
you which you prefer, A: $100, B: $500, or C: $1000, the choice is
pretty easy. But what if you can vote for one of these or two (or
even all three), and the result with the most votes wins. But the
"value" of each choice is different for the other voters. The choice
with the most votes wins. If it's Plurality, how would you vote? In
repeated ballot, which is what Robert's Rules of Order actually
suggests as the default voting method, you'd vote for $1000, very
simple. In Top-two runoff, *usually* you would vote for $1000, the
same. How would you vote in Approval? Suppose you know which of the
three choices, A, B, or C, are the "top two" choices, most likely to
win this election. Your strategy is obvious, then. Vote for your
favorite among the top two and any candidate better than that. Your
vote depends on your perception of what choices are realistic! If
$1000 is a frontrunner, you "bullet vote" for $1000. If not, you vote
for both $500 and $1000.
James G-A sets up, with his simulations, is considering 3-6
candidates, and he sets up a "value" for each candidate, and assumes,
then, that you will "approve" any candidate better than the average.
This is an *atrocious* Approval voting strategy, *unless you have no
knowledge at all of what is likely.* So if you are totally ignorant
of the social situation, sure, you might vote this way. In real
elections, voters either have better knowledge, or have some idea of
how close to popular opinion their own perceptions are. They will
vote better than this lousy strategy, *normally.* So, of course, big
surprise, we then find that they could improve the outcome by ...
voting more sanely.
And then FairVote cites this as some kind of proof about IRV. The
facts are that real elections are of two kinds: partisan and
nonpartisan. Most IRV implementations in the U.S. have been in
jurisdictions holding nonpartisan elections, San Francisco city
council elections, for example, where RCV (the limited 3-rank version
of IRV) replaced top-two runoff. In these elections, it is extremely
rare for the result from IRV to be any different from Plurality; it
is quite likely that every election held in San Francisco, for
example, would have turned out the same way if they had just gone
back to Plurality instead of using IRV, and the only serious
difference is that IRV *pretends* to find a majority. I'm sure we'll
come to that!
This is what real election results in the U.S. from IRV has shown:
the preference order of the candidates only rarely changes as
candidates are dropped and additional rounds are counted. I've not
seen *one* nonpartisan election in the U.S. that was different from
this. This was, by the way, unexpected, though I later found that
"flipping" was uncommon in Australia. Never does a third ranked
candidate in first preference come up and win, and for second rank to
come up and win is quite unusual. And those are partisan elections.
With nonpartisan elections, popularity seems to be connected with
name recognition; regardless of the specific cause, what seems to be
normal is that those who prefer A to all other candidates will be
divided as to second preference as if they were a sample of the rest
of the population. I.e., they will prefer B to C in about the same
percentage as all the other votes who don't prefer A.
But when there is a partisan affiliation, that changes. Even in a
nonpartisan election, when there is *ethnic affiliation* as an issue,
which happened in San Francisco, transfers do not fit this "sample"
model. Those who voted for an Asian who was eliminated, in first
preference, almost overwhelmingly voted for another Asian. And, then,
we see, in partisan elections, that those who prefer a minor
candidate, on the left or right, say, will largely prefer to vote for
the candidate more toward the center than the one they preferred ...
but not on the other side of the spectrum.
I'm not sure of the effect of these considerations on J G-A's
simulations. The manipulability of IRV, allegedly so highly resistant
in the FairVote report of his work, is almost exactly the same as of
Top Two Runoff, no big surprise.
FairVote treats the 'superiority' of IRV in this respect as a "fact,"
when this is just one study, one that did not simulate real
elections, and we know of a recent real election where IRV showed the
kind of vulnerability that J G-A was predicting would be rare,
Burlington. It is *normally* rare. Think about it. IRV is *close* to
"contingent vote." (James G-A did not actually simulate top-two
runoff, he simulated contingent vote, as if there were a single poll
with, then another immedate vote, based on the immediate preferences
with the same voters and same unchanged preferences. That is
contingent vote, or 2-rank IRV, not top two runoff). Real partisan
elections don't generally behave the same as would "issue" elections.
The serious effect of "center squeeze," which is the only situation I
know of where voters with IRV can improve the outcome by voting
insincerely, only arises when there are three candidates within
possible reach of winning, and it especially arises when the two most
popular parties are at either end of some spectrum, so that there is
a middle party which will be the second choice of nearly everyone. So
in a direct contest between this centrist and either of the other
two, this candidate might win with a margin as high as two to one.
But if the candidate is in third place in first rank, that candidate
is eliminated with IRV (and with top two runoff). Thus to see this
situation where strategic voting benefits the voter with IRV, there
must be three major parties. That was the situation in Burlington. If
that situation is common in a place, IRV is an absolutely lousy
method to use, and the voters of Burlington figured it out and tossed
out IRV as soon as they got it.
And in most other places, the whole issue is moot. There, other
issues loom much larger. Will FairVote even raise the most important
issues? Let's see. Since the three-major-party situation is unlikely
to arise in James G-A's simulation, the simulation makes it seem that
IRV isn't vulnerable.
FairVote tends to assume and promote that IRV is a fix for every
voting problem. Not. In almost all places where it has been
implemented, it has done no good at all, at high cost. Usually the
results match plurality, period, they have in San Francisco, which
has run the most IRV elections in the U.S., so far.
But top two runoff does something quite different, and James G-A is
not going to see it in his simulations, and, almost certainly, it
represents real improvement of results. In one out of three top-two
runoff elections, the winner is a flip from the first round. This is
far more rare with IRV, and probably explains why strong minor
parties are a feature of jurisdictions that use top-two runoff, and
places using IRV are two-party duopolies. They can actually win,
occasionally. All they have to do is to make it into second place,
and then, now that they have shown some clout, convince the voters in
the runoff to vote for them, and their supporters will be highly
motivated to turn out, to get a rare chance at victory. That's
useless with IRV, where a loss is a loss! More about this later, I assume.
To be continued.
----
Election-Methods mailing list - see http://electorama.com/em for list info
