At 08:38 AM 1/17/2010, Kathy Dopp wrote:
> Also of course if the A supporters had not ranked B then A would
> have won, a big violation of Later-no-Harm.
Later-no-harm is a very bad feature of IRV that prevents IRV from
finding majority-favorite compromise candidates and tends to elect
extreme right or extreme left candidates. In any negotiation, it is
necessary to reveal the 2nd choices of all parties early on, not hide
2nd choices entirely of some voters and never consider them at all
like IRV does to most voters in many elections.
Chris is Australian, and is one of a rare breed: someone who actually
understands STV and supports it for single-winner because of LNH
satisfaction. Of course, LNH is a criterion disliked by many voting
system experts, and it's based on a political concept which is, quite
as you say, contrary to sensible negotiation process.
In rational process, people start by advocating their first
preference, and will only disclose lower preferences if the
preference strength of the favorite over the lower ones is low or zero.
Then, as they become aware of the first preferences of others, they
will start to disclose lower preferences. The more eager they are to
find consensus, the sooner they will do this. In Range, they *may* do
it immediately, or they may conceal those preferences, it is their option.
My problem with straight Range voting is precisely because it
too-quickly bypasses the normal strategy. And it will be less
effective because of this.
However, Bucklin does imitate the rational process, and Range/Bucklin
would do it even more accurately. The voter discloses to the method,
as it were, as if the ballot becomes the voter's agent, true relative
utilities. That becomes the optimum strategy! The system then slides
down the voter's established utilities, starting out by bullet voting
for the favorite or any equally rated candidates. Not finding a
majority, it then clicks down the scale for all voters and checks
again, and it keeps doing this until it finds a majority winner. If
it's a deterministic election, it will go all the way down to the
lowest non-zero rating. But I greatly prefer only setting a winner if
a true majority approval is found, thus the method sticks with the
tradition, a tradition which has centuries of experience behind it,
and which is only abandoned in public elections in the name of
efficiency over democracy.
There is a name for that choice. Fascism. But I'm not claiming that
plurality is fascist, just that some aspects of some of the arguments
for it are.
What I'd do with Range/Bucklin is to use it to discover if there is a
majority winner with the Bucklin process. Then I'd count all the
votes for all candidates. If sum of votes indicates a different
winner, I'd hold a runoff. If not, that's it.
If there is no majority-approved winner, there is a runoff. I'll
defer, for a moment, the question of candidate eliminations.
Ultimately, I prefer no eliminations, or, at most, only ballot
listing eliminations, write-ins still allowed. Or perhaps the
candidates are listed on the second ballot with their Range ratings
from the first election, so voters have that information handy and
can make reasonable compromises right there. And the second election
might be pure Bucklin (i.e., Approval/Bucklin).
And what happens if there is no majority in the runoff? In democratic
organizations, they just keep at it until they get a majority. If
there is a harm from the office not being filled, well, that's
pressure to make a compromise. If the majority of people don't want
to fill the office by finding a majority compromise, we have a loss
of democracy, I call it the "tyranny of the past." It was decided,
that, in the future, decisions would be made with less than a
majority. However, there is an option. If there is a question asked
on the ballot, "Shall this election be awarded to the candidate with
the most votes, even if there is no majority," the majority will have
determined on an election unconditionally if more voters vote yes on
that question than vote no. But under current conditions, we wouldn't
ask that question. We might be afraid of the answer we get.
Those who are in power fear that the power might evaporate, and often
come to not trust the people. Systemic problem, I'm not blaming anyone.
Yes. *Any* system that guarantees the principle of
"one-person/one-vote" as some US judges insist on, has the feature
that it encourages bullet voting.
Actually, there are few decisions that insist on a narrow
interpretation of one-person, one-vote. Minnesota is in an odd
situation now, but I should really read the recent decision. Brown v.
Smallwood was an anomaly, not confirmed anywhere else.
What is important is that, in the end, everyone gets equal voting
power, without discrimination. Taking this principle to the end
conclusions, though, plurality has some severe problems, so does IRV.
Approval, in my view, clearly satisfies the principle.
I've really come to like Bucklin, because it allows voters to
exercise full power for one candidate at the outset, then add, *if
they choose to do so*, alternative approved candidates.
The biggest problem with Bucklin was that it was oversold as finding
majorities without a runoff, which, of course, no method can do with
a guarantee! I'm sorry, but my vote Genghis Khan > Adolf Hitler is
not a vote for Genghis Khan. Unless that's a Bucklin vote, and it's
deterministic, and those are the only two choices, in which case the
system is actually coercive and, in fact, in a runoff between the
two, I'd be trying to find a way to get out of the country and not
waste my time voting.
Consider the Lizard v. Wizard election, and if you don't know the
reference, read Gaming the Vote, Poundstone. What if write-in votes
were allowed? As they are by default in California and some other
places? The Roemer supporters, sensible, could have polled to
determine the chances of a write-in campaign succeeding. If it
failed, no big deal, because the Wizard wouldn't win anyway, it was
just the Center Squeeze effect hitting Roemer. The Lizard (Edwards)
would win if the write-in fails, not the Wizard, only a certain
largely racist faction would have voted for the Wizard.
But, hey, that runoff sure got people out to vote! That, in fact, is
part of my point about runoff elections. If the people care, they
turn out. Low turnout in runoff elections isn't any kind of pathology
in itself.
Use Bucklin methods with a majority requirement, and use simple
Bucklin in the runoff, and there is no spoiler effect in either
election. It really should be, I suspect, a no-brainer once all the
cards are on the table and seen by all players.
> If IRV doesn't satisfy the US courts, then how come IRV is used in the US?
Because the judges have been cleverly and grossly misled on the facts,
as they were in the most recent MN court case by FairVote. Also, the
attorney who spoke for the plaintiff was not very educated on all the
facts about IRV/STV, having also been misled by defendants, and should
have let the attorney who did all the research for the firm speak
instead.
The plaintiff was not an organization truly knowledgeable on the
issues, they had axes to grind. I argued long ago that voting systems
activists should provide friend-of-the-court briefs in that case.
Kathy attempted to do so, I know. But I haven't read the actual
decision, I did read some of the arguments.
This is a place where having some organizational structure advocating
voting system reform, overall, more than FairVote, which is stuck on
one defective approach, would have helped. The FotC brief should have
been written by the best available constitutional attorney, and that
would properly take some money, unless a good one were available
pro-bono or willing to accept whatever donations could be collected.
My worry was not that the MN court would allow IRV (Brown v.
Smallwood was defective), but that it would allow IRV and continue to
disallow other, better methods, based on the LNH problem. That would
be a damaging outcome, truly. Winning on the constitutional arguments
against IRV is probably politically impossible. The people may make
bad choices! Legislative bodies may make bad choices!
> Why does Kathy elsewhere defend Top Two Runoff which isn't monotonic?
Another highly misleading statement. In Top-two runoff each election
is monotonic and voters thus retain the right to cast a vote that
increases, rather than decreases, the chance of their candidates'
winning, unlike IRV/STV which removes this right, among others.
And where were voting systems activists when the California Supreme
Court was deciding, based on arguments from the City of San
Francisco, that the runoff wasn't an election in itself, but merely
an appendage to the primary, part of the same election? One of the
worst decisions I've seen in a long time, preventing the public from
fixing mistakes in the primary, and thus frustrating the purpose of
runoff elections and of the constitutional requirement for allowing
write-ins in any election.
Asleep at the switch, I'd say. How about we take steps to make sure
that doesn't happen again?
----
Election-Methods mailing list - see http://electorama.com/em for list info
