Re: Norman Petry and I (Ossipoff) recommended CSSD, but Schwartz Woodall is a better voting system for Debian
On Tue, May 14, 2013 at 8:01 PM, Steve Langasek <vorlon@debian.org> wrote:
> (With apologies to the non-Americans on -vote... :)
>
> On Tue, May 14, 2013 at 11:56:49AM -0400, Michael Ossipoff wrote:
>> One reason why I've been advocating Schwartz Woodal (along with
>> Woodall and Benham) is because, in official public government
>> elections, the chicken dilemma _would_ be a problem, or at least a
>> serious nuisance that would demand special strategy, and would take
>> away the freedom from strategy-need that Condorcet methods could
>> ideally have.
>
> I have seen this argument advanced by advocates of IRV, but I have not seen
> any real proof.
Well, the fractiousness and dis-unity of the U.S. progressive
political parties, and of the voting-system reform community count as
proof. Various people have cited actual instances of the chicken
dilemma in American politics, both early and recent. I don't have
those examples handy, but they could be found on the Internet, when
searching under "chicken dilemma".
Of course the chicken dilemma is so common in the animal kingdom as to
be universal. The mixed strategy that is the game-theorists solution',
and the animals' solution, to the chicken dilemma is similar to the
strategic fractional ratings solution that I suggested, If humans
were an exception to the chicken dilemma, then that exception would be
what needs explanation.
> Most IRV advocates seem to claim IRV is better than CSSD
> because voters are too stupid to cast a ballot non-strategically
It depends on the electorate. I don't think that Debian members betray
eachother by chicken dilemma defection, but I don't have that
confidence in the other groups that I've mentioned above.
,> and
> therefore we're better off with a system that gets closest to the Condorcet
> winner given such broken inputs.
I emphasize that Woodall, Schwartz Woodall, and Benham always elect
the CW when there is one--and also always choose from the sincere
Smith set, meaning that they also meet the Mutual Majority Criterion
and the Condorcet Loser Criterion.
> But while I don't hold the electorate in
> particularly high esteem, I reject this particular premise: most voters
> aren't going to grasp the *proof* of Condorcet, but they are certainly
> capable of grasping the *principle* that their best strategy when voting is
> to vote honestly.
Rivalry can make people act un-cooperatively. Look at the behavior of
the voting-system advocacy community, or the American progressive
political parties.
>> I consider official public government elections to be where we
>> seriously need a better voting system. I believe that it would make
>> all the difference.
>
>
>> The Green Party U.S. (GPUS) offers Instant Runoff (IRV) in its
>> platform. I know that Debian is an international organization, but of
>> course my main goal has been reforms in the U,.S. It seems to me that
>> any reform in the U.S. must start with the election of Greens to
>> office.
>
> However, coming from a Debian background I find the flaws of IRV so much
> worse than those of CSSD in terms of enacting the will of the people that I
> consider endorsement of IRV a mark against the Green Party candidates in my
> jurisdiction.
IRV would be a poor choice _for current conditions_. By "current
conditions", I refer to dishonest, disinformation media, and a public
who believe those media when they say that only the Democrat and the
Republican can win, and that the Democrats are the party of the
ordinary people, and the opposite of the Republicans, and in some way
progressive.
Under those sad conditions, we need a method that is _foolproof_
against favorite-burial to protect a supposed "lesser-evil". We need a
method that meets the favorite-betrayal criterion (FBC). ... a method
in which a voter never has even the least strategic reason, under any
conditions, to vote someone ovet their favorite.
IRV fails that criterion. CSSD and most other Condorcet methods fail
it too. So do Woodall, Schwartz Woodall, and Benham. For current
conditions, I advocate Approval and Score (and a rank method that I
call Sequential ICT, defined at electowiki.
But conditions would be entirely different if the Greens were elected
to the presidency and most of Congress. Current conditions would no
longer exist.
For one thing, any electorate who could elect the Greens in Plurality
would be an electorate who are no longer deceived by our dishonest,
disinformational media.
Additionally, in a Green America, the media would be entirely open,
participatory, honest, and agenda-free.
So both ingredients for "current conditions" would be quite absent in
the "Green scenario".
So there would no longer be a need for FBC. Then, there would be
nothing wrong with IRV, except for the fact that, failing the
Condorcet Criterion (CC), IRV can thereby fail to respect some
majorities that aren't mutual majorities. That could result in IRV
being replaced by Benham or Woodall, when a dis-satisfied majority
insist.
But my point is that, as inadequate as IRV is for current conditions,
it would be fine for the Green scenario, and so the Greens aren't
wrong to offer it in their platform. I'd suggest to the Greens that
Benham or (especially) Woodall would be better. But IRV is okay enough
that it's fine if, in the Green scenario, we start with IRV, and then
change to an IRV Condorcet hybrid if some majority insist on doing so.
And if IRV remains, due to inertia, that' would be fine too, under
Green scenario conditions. If you're in a mutual majority, IRV will
work _fine_ for you. If anyone doesn't think that they'll be in a
mutual majority, then they must think that their policy preferences
won't be popular. Maybe they should re-consider their policy
preferences.
>
>> IRV, like Woodall, Benham, and Schwartz Woodall, meets the Mutual
>> Majority Criterion, and has no chicken dilemma. But of course IRV
>> fails the Condorcet Criterion. IRV's failure to always elect the
>> Condorcet winner compromise makes IRV too uncompromising and inimical
>> for amicable organizations. It also makes IRV vulnerable to
>> replacement by a dis-satisfied majority, when IRV is used in official
>> public government elections.
>
>> So, I feel that, if the GPUS were elected here, and IRV were
>> established as the voting system, there might soon be majority wishes
>> to replace IRV with a Condorcet-complying voting system. A good
>> Condorcet-complying replacement would be Benham, Woodall, or Schwartz
>> Woodall.
>
> I think this significantly underestimates the power of inertia in such
> matters. It's difficult to mobilize people around something as esoteric as
> voting systems; even with all the readily available evidence of how our
> current system betrays the will of the electorate, changing the voting
> system remains a fringe issue throughout the US. Once IRV is adopted, this
> problem will be compounded. Voters will have a knee-jerk reaction to any
> request to change the voting system again after they've already done so
> recently; the practical advantages of Condorcet over IRV will be more
> difficult to explain to the electorate; and there will be no history of
> abuses / wrong election outcomes for advocates to point to in justifying the
> need for a further change.
Well, with IRV in use, the voters who are not in the mutual majority
will soon notice that things aren't going their way, and will see that
they're at a disadvantage. Likewise, when the preferrers of the CW see
their CW getting eliminated, they might not like that. Those CW-plus
the non-MM voters, will add up to a majority. If the CW preferrers
feel strongly about wanting their candidate to win, then there will be
a majority who are motivated to insist on switching to Benham or
Woodall, in order to always elect the CW compromise, and thereby honor
_all_ majorities, however constituted.
Besides, if you're in a mutual majority, then IRV will work fine for
you. I'm confident that I'll be in a mutual majority, because I
believe that my policy preferences are good, and I believe that a
majority of the public are (from what i've heard them say) practical
enough to agree. IRV is not for people who don't have confidence in
the desirability of their policy preferences.
>
> So I think we have (at most) one shot in our generation to fix the voting
> system; and if we manage to get rid of FPTP only to settle for another
> exploitable system, we'll be stuck with it for good.
I claim that, when the Greens enact IRV, and we begin using it, some
people, likely including the Greens themselves, will notice that it
would be better to ensure the election of CWs. Remember that, in a
Green America, the media would be vastly more open and participatory,
and that there would be completely open and free discussion of such
things as voting systems, and dis-satisfaction with IRV results, if
there is any.
If IRV is dis-satisfying a majority (as it of course can do), then
that will be discussed, and that majority will easily have the power
to replace it, via initiative or referendum. GPUS advocates vastly
expanded powers of initiative and referendum at all governmental
levels, in all jurisdictions.
So voting system reform, under the Greens, is definitely not a one-shot thing.
>
>> That gives me incentive to advocate Schwartz Woodall for
>> organizations, because it's the kind of voting system that would
>> likely be eventually adopted in a Green U.S. So I'm just telling my
>> motivation to advocate Schwartz Woodall, even to organizations that
>> don't really have a chicken dilemma.
>
>> Of course (at least if there's a chicken dilemma), Schwartz Woodall's
>> combination of the Mutual Majority Criterion, no chicken dilemma, and
>> the Condorcet Criterion would make it the a good choice (the best
>> choice, I claim) for organizational voting.
>
>> Of course obviously, if Debian doesn't have a chicken dilemma, there's
>> no need for Debian to change its voting system from CSSD to Schwartz
>> Woodall.
>
> Given that Debian is my own standard example of Voting Done Right™, I think
> it's worth evaluating how to improve Debian's system even if we're fixing
> bugs that we think aren't practical problems in Debian - if they are
> practical problems elsewhere. I just have not been convinced that the
> chicken dilemma is the practical problem that IRV advocates argue it is.
Well, I've presented my arguments, and I respect your right to your
own judgment.
But the game-theorists seem to consider the chicken dilemma a problem,
to some degree (I call it only a "nuisance"). The chicken dilemma must
be, and is, dealt with throughout the animal kingdom. As I was saying,
I point to the fractious rivalry among American progressive political
parties, and among the voting-system reform community, as examples of
possible sources of chicken dilemma, and as evidence that humans are
not an exception to the animal kingdom's susceptibility to chicken
dilemma.
Michael Ossipoff
>
> --
> Steve Langasek Give me a lever long enough and a Free OS
> Debian Developer to set it on, and I can move the world.
> Ubuntu Developer http://www.debian.org/
> slangasek@ubuntu.com vorlon@debian.org
Reply to: