[EM] Voting reform pitch

Jameson Quinn jameson.quinn at gmail.com
Sat Jan 7 07:06:47 PST 2012


I'm working on a blog post which makes a pitch for voting reform. The
current draft is pasted below, and also up on Google
Docs<https://docs.google.com/document/d/1sPlGQEPTsoWq-VYEW_P6YfRl4rT5W_yA2Auq4nyqCug/edit?pli=1&hl=en_US>.
It's meant to express my opinions, not always a consensus, so I've set it
so that anyone can make comments (Insert -> Comment) but only I can
directly edit it. It has the following sections:

The Problem (what's wrong with plurality, using an extended metaphor
between parties and carmakers)
The Ideal (help open people's minds about how much better things could be)
The Solution(s) (explain that there are many good options for reform, and
give my reasons for choosing two: SODA for single-winner and PAL for
multi-winner)
Detailed description (Explain the process for SODA and PAL; needs work)
How to get there (Activist strategy; currently just an outline, needs
filling out)

I'd appreciate any comments people have (either favorable or unfavorable,
and either here by email or directly on the google docs page).

Jameson

Here it is:

* The Problem
Imagine that there were a law that only two competing models of car were
allowed. Probably one of the two would be smaller and cheaper to start out
with. But as the two companies fought for market share, their models would
grow to look more and more like each other. After all, each company could
take their "base" customers for granted, and the only important fight would
be over the "swing" customers who want a medium-sized car. Meanwhile,
neither company would bother to spend money to develop new features like
antilock brakes and GPS - or even air conditioning and windshield wipers.
Why bother, when even without innovation, they were basically guaranteed
half of the market?

The end result is obvious. Consumers would end up forced to pay too much
money for a choice between two all-too-similar cars of stagnant, outdated
design.

Does that result remind you of the political situation? That's no
coincidence; it's because there is, in fact, a law that only two parties
are allowed. No, not a statute; an empirical law, like the scientific law
of gravity. "Duverger's law" states that, as long as we use the plurality
voting system in which each voter can only support one candidate, two-party
domination is guaranteed. It's a self-fulfilling prophecy: as long as the
two parties always win, a vote for a third party is always wasted; and as
long as voters don't want to waste their vote, the two parties will always
win. But remember: unlike the law of gravity, "Duverger's law" has a way
out: it only applies as long as we use plurality voting.

But until we find that way out, we're stuck with two parties that often
strive to minimize their differences, and entirely neglect the opinion of
majorities on multiple issues, just as in the imaginary car example. Want
to end the war in Afghanistan[1], reduce free trade[2], depose executives
at bailed-out banks[3], or end subsidies for big farms [4]? You're out of
luck; neither party will speak for you (at least, not the mainstream of
either party). The two parties can use apocalyptic rhetoric to battle over
whether health care is provided by slightly-more-regulated market
mechanisms or slightly-more-market-driven regulations, or over whether
immigration enforcement should be slightly tighter or slightly looser; but
anyone who proposes a root solution to these problems --- from either side
--- is not even worth taking seriously.

And in fact, the political duopoly is in many ways even worse than the
hypothetical automotive duopoly. Sure, on the bright side, individual
politicians aren't mass-produced; the two political models can vary a bit
for regional tastes. But consider the downsides:

   - While you can do without a car, you can't opt out of having a
   representative or a government.
   - Up to half, or with "vote splitting" even more, of voters can end up
   not even getting their preferred model; it's as if entire neighborhoods
   were forced to get whichever car was more popular locally.
   - Gerrymandering can make that problem worse; in the car metaphor, a
   gerrymandered system would have a shortage of one of the models and cars
   would choose who drove them, making it hard to get the oldest clunkers off
   the road.
   - Rich political donors, whose early infusions of money can influence
   the makeup of the two options voters are left with, can distort the very
   agenda to their advantage.
   - And finally, any voters who are seduced by the siren's song of third
   parties tend to effectively help elect exactly the candidate they like the
   least.


It doesn't have to be that way. Only plurality --- essentially the worst
known voting system --- binds you to vote for a single candidate, forcing
you to make the lose/lose choice between a meaningless vote for your
favorite or a compromised one for the lesser evil. Almost any other voting
system solves this basic problem (as with Approval Voting, SODA voting,
Majority Judgment, Range Voting, and Condorcet systems) or at least
mitigates it (as with Instant Runoff Voting).

The Ideal
So, what would the world look like without a political duopoly?

First and foremost, you could always vote for your true favorite candidate,
and still have your vote count against your political opponents. There
would be no need to vote the lesser evil to avoid losing the election to
vote-splitting (as happened to Democrats in 2000) or eliminating moderates
in the primary (as arguably happened to Republicans in several Senate races
in 2010). That individual freedom has crucial implications for the broader
political atmosphere. It means that there would be no more need for purists
and compromisers on either side to be desperately fighting. Progressives
and blue dogs, or tea partiers and republicans, would still have
disagreements; but their arguments would no longer be fueled the desperate,
insecure knowledge that one false step could cause the worst possible
result in an otherwise-winnable election.

Second, it would be harder for special interests to twist the government to
their ends. In plurality, where your vote is wasted unless it went to one
of the two most popular candidates, candidates are forced to use
advertising to essentially buy their way into one of those top slots. As a
plurality voter, even if I hate all the political ads I see, I have to vote
for one of the people making those ads, because I know everyone else will
--- and they are doing so because they know I will. But in a better system,
that viciously circular logic would be broken; I'd be free to give those
ads exactly as much credit as they deserved. Furthermore, if the electorate
really favored policy X, a special interest which favored Y would have to
corrupt not just two politicians and/or party primaries (or, in
uncompetitive partisan areas, just one); but a potentially infinite number.
That is, no matter how many politician and/or party primaries they
corrupted to support Y, there would always be room for another honest one
who favored X - and a fair chance for that candidate to win on that basis.

Yes, it's hard to imagine a politics where cooperation had a chance against
infighting, where the majority had a chance to set the agenda without being
vetoed by special interests. It would truly be a case of "may the best
ideas win", or at least the best wisdom that the democratic crowd can
offer. Of course, I think my own ideas are wise, and I think that in a
fairer democracy, my side would win more often. Since I happen to be a
progressive, I think that eventually more and more people would realize how
well progressive ideas work to create prosperity for all; if you're a
libertarian, or a Christian conservative, or a centrist, I'm sure you feel
the same way about your ideas. But it's important to realize that voting
reform is not a sneaky recipe for a third-party takeover. A Ralph Nader or
a Ron Paul will have a fair chance to raise their issues; but as long as
most Americans agree that the Democrat and the Republican are the
most-qualified candidates, then one of those two will win. In the
automotive analogy, an even playing field just means that smaller makers
like Mitsubishi and Mazda are allowed to exist, not that they'll sell as
many cars as GM and Toyota. In any foreseeable future, the Democrats and
Republicans will continue to be the biggest two parties; it's just that
they'll have to work harder and be more responsive in order to maintain
that position. And as a voter, that's just how I want it.

The very best voting systems can provide even more benefits. Some systems
provide optimal expressivity, allowing voters to clearly express various
levels of support for candidates. Some offer the possibility of choosing
the truly best candidates, if voters use that expressivity honestly. Some
provide assurances of fairness and robustness, that results tend not to
give an unfair advantage to voters who follow the polls and use strategic
calculations to determine their vote (nor an unfair disadvantage to those
who don't). Some provide optimum simplicity for voters, allowing easy
choices to lead to good results. Some help smooth the process of ballot
counting and reduce spoiled ballots. Some provide reasonable safeguards
against a surprise win by an unknown candidate who's only been
superficially evaluated by the voters. And some provide a fair role for
strong leaders of minority factions, allowing them to negotiate some
compromises even if they don't end up winning.

The Solution(s)
So, which voting system is best? There are a number of systems which would
do the basics: end spoilers, reduce infighting, curb the power of special
interests, and broaden the political agenda to include more of the issues
people care about. STV-PR, Approval Voting, Condorcet systems, Majority
Judgment, and Range Voting would all accomplish these important goals. Even
Instant Runoff Voting (IRV), though it does not fully fix the spoiler
problem which is at the heart of Plurality's pathology, would in most cases
be a significant improvement (and, as the only reform which has gotten
traction in the US so far, it can prove it.)

But in the end, we can't choose all the good systems. We need to choose one
best system for single-winner races (president, governor, mayor, or
senator) and one for multi-winner races (congressional representative,
state representative, city council). This is a lively debate among voting
theorists like myself. There seems to be no end to the mathematical
criteria, monte-carlo experiments, practical considerations, and election
scenarios which can be used to support or argue against one or another
voting system, and there's certainly no end to the arguments about which
system is best. So I'll tell you what I think, and I can assure you that
I'm not the only one who feels this way; but if you're interested in
finding out more and developing your own opinion on these matters, I invite
you to join the election methods mailing list where these matters are
discussed.

Despite the sometimes-heated arguments, there are certain matters on which
voting theorists can agree. Everything I've said so far --- the glaring
flaws of plurality, the serious consequences of those flaws, the shared
advantages of possible reforms, and the specific advantages offered by some
reforms --- is the consensus of most or all of the community of theorists.
What follows, though, is not necessarily the view of more than an important
minority.

Which voting system you think is best, depends on what you're looking for
in a voting system. Here are the values which drive my decision. A good
system should...:

   1. ...be simple for voters, even when that means reducing expressivity.
   While I personally would love to be able to grade a dozen candidates on a
   scale that let me finely distinguish whether my favorite candidate merited
   an A+ or just an A, I know that many more voters just want to pick their
   favorite and go home. It's more important that the system deal well with
   them than with me; after all, even for a high-information voter like me,
   there are probably more-productive ways to engage with politics than
   researching no-hope candidates enough to make fine distinctions between
   them.
      - An essential component of voter simplicity is reducing the voters'
      need to strategize. It's little use having even the simplest possible
      ballot format if a voter has to pore over poll results in order
to get the
      most out of it. Also, insofar as voting strategies involve dishonest
      switching of preferences, instead of just "semi-honest"
exaggeration, they
      almost inevitably lead to worse results from the overall society's
      perspective.
   2. ...encourage healthy politics, by balancing diversity with incentives
   for cooperation. That is, the system provides a fair, proportional degree
   of power and voice even to minority groups, but avoids deadlock when a
   decisiveness is necessary.
   3. ...be acceptable for honest incumbents. Passing voting reform is a
   heavy enough lift even if our only enemies are lobbyists, party
   bureaucrats, and the most-corrupt politicians. If the new system is such a
   radical upheaval that it threatens even honest incumbents, it's a dead end.
      - In practice, this means that a new system should not allow an
      unknown (and possibly unqualified) upstart to win unexpectedly just by
      taking the most-centrist positions.
      - The system should leave some room for political parties to play a
      role, if voters and candidates want to organize themselves that way. That
      doesn't mean that the system must formally recognize political parties;
      just that it must be compatible with them. See value number 5,
though, for
      a counterbalance to this point.
      - Also, for multi-winner systems, winners should have some connection
      to predefined electoral areas, so that each voter still has
their own local
      representative.
   4. ...allow for counting procedures that provide reliable, fraud-proof
   results.
   5. ...leave the decision in the hands of the voters, reducing the
   influence of external factors such as money and party bureaucrats.


With those values, the outstanding best single-winner system is Simple
Optionally-Delegated Approval, or SODA voting. For multi-winner elections,
there are several possible systems which might serve, but for consistency
I'd choose the multi-winner system which corresponds to SODA voting, known
as Proportional, Accountable, Local (PAL) representation.

Of course, if you have different values, you might choose differently.
Among single-winner systems, SODA is unrivaled on aspects 1 and 2 above,
and among the best systems on aspect 3, 4, and 5. But if instead you value
simplicity of describing the system, you'd choose Approval Voting; if you
value optimal results, you'd choose Range Voting; if you value
expressivity, you'd choose Majority Judgment; if you're willing to do
without the secret ballot, you might choose Asset Voting; and there are
several values and mathematical criteria which would point you towards a
Condorcet system such as Schulze voting. And then there's IRV, which is the
best-known reform and by far the most-successful one so far (perhaps
because it is the only rival to SODA on aspect 3 above); although it fails
to fully resolve the spoiler problem, and has a counting process which can
be expensive and insecure.

Detailed process
These procedures are outstanding for the ease of the choices they leave to
the voters. Although that does not mean that they are the simplest systems
to describe, every step in the process is justifiable in reference to the
values above.

SODA voting
This system consists of a three-step process for finding how many voters
approve of each candidate --- either directly, or indirectly through
delegation. At the end of the process, the candidate with the highest
approval wins.
*

   *
   1. Candidates publicly declare their rankings of the other candidates. This
   satisfies value 5 by allowing voters to make an informed choice in step 2.
      - 1A) Equal rankings and truncation are allowed. The candidates'
      rankings are all made public. Later, in the "candidate voting" step,
      candidates may only approve other candidates in a way that is consistent
      with their rankings. This helps reduce the possibility of corrupt
      vote-selling or "smoke filled rooms".
      - 1B) It is presumed that there is some mechanism in place so that no
      candidate is "caught off-guard" by another candidate's rankings. For
      instance, there could be an option for a candidate to change
their rankings
      after another candidate has announced their own rankings. That
is, if A and
      B agree to rank each other for delegation, but then after A ranks B, B
      reneges on the agreement by ranking A last, A could retaliate by moving B
      to the last ranking. The details of such a mechanism are unimportant,
      because its very existence would prevent it from being used; candidates,
      realizing that surprises don't work and only cause resentment, wouldn't
      attempt to use them.
      - 1C) Ballots are formatted in order from candidates who submit the
      most complete rankings of other candidates, to those who submit
the least.
      Where this is a tie, ordering is random. This is a slight incentive for
      candidates to submit a complete ranking in order to appear high on the
      ballot.
   2. The ballot lists all the the candidates, plus a line called "Do not
   delegate". Voters may approve any number of candidates as they wish. Voters
   who approve only one candidate, and do not mark "do not delegate", have
   delegated their vote to that candidate. This means that they may have
   additional approvals (effectively) assigned to their ballot by the
   candidate they chose. This delegation process satisfies value 1, because
   a lazy voter does not need to do anything more than delegate to their
   favorite candidate. Unlike in plurality, such a vote is never wasted; if
   the favored candidate cannot win, they can still make the vote count for or
   against the eventual winner in step 3. This also helps satisfy value 1 by
   relieving the voter from the need to worry about strategy. The ability to
   opt not to delegate satisfies value 5.
      - 2A) Total approvals and delegated ballots for each candidate are
      tallied and announced.
      - 2B) After totals are announced, there's a week for candidates to
      analyze results, negotiate, and plan what to do in step 3.
   3. As long as no candidate has approvals from more than half the voters,
   the candidate with the most approvals may assign further approvals to the
   ballots delegated to him or her, consistent with the rankings declared in
   step one. Candidates with less than 5% approval have their delegated votes
   assigned automatically such that they approve as many as possible without
   approving both of the two highest-approval candidates which are
   distinguished in their ranking.
      - A) There is a brief period - perhaps a week - for candidates to
      analyse and negotiate based on these preliminary results. (Actually, the
      correct strategies for all candidates and the resulting winner
will already
      be obvious. Usually, all candidates except this winner would concede as
      soon as preliminary results are announced. However, for the occasional
      candidate inclined to act irrationally in a way that matters -
say, by not
      delegating to an ally, even though the alternative is to see an enemy
      elected - this interim period would give them a chance to rethink things
      and come into reason.)
      - B) All candidates, in descending order of the number of total votes
      they have, choose how to add approvals to the votes delegated to them. To
      do so, they choose a cutoff candidate, and approve all others
whom they had
      ranked above that cutoff. They may choose the top of their
ranking list as
      the cutoff - that is, not approve any other candidates. They may not
      approve all other candidates, as that would require choosing a cutoff who
      was worse than all candidates including themselves. If they
declared a tie
      in their preferences, they must either approve all candidates whom they
      included in that tie (as well as anyone they ranked above that),
or none of
      them. (Note: Doing this in descending order prevents a weaker candidate
      from making an ultimatum to a stronger candidate, and thus
strengthens the
      strategic equilibrium of any pairwise champion there is.) These
choices are
      announced publicly as they are made. Delegated approvals which are shared
      during this process may change the decision order of succeeding
candidates.
      - When a candidate adds an approval to their delegated votes, their
      delegated vote total is added to the approval total of the candidate they
      approved.
      - B)i) Write-in candidates could not receive delegated votes, but
      could be approved at will by official candidates, regardless of declared
      ranking order.
      - C) The number of delegated votes held by each candidate is added to
      the approval total of everyone they approved.


   PAL representation
   (forthcoming)

   How To Get There (rewrite needed)
      - Raise consciousness
         - Make the argument
         - Website where people can use good voting methods for their own
         elections and polls
         - Don't get caught in fighting between methods; there are many
         worthwhile improvements on plurality.
      - Get organized
         - Get support from a larger organization like Common Cause and/or
         LWV.
         - Form alliances with partisan groups like Libertarians, Greens,
         OWS, and Tea Party.
         - Develop a voting-reform-specific mailing list of people who
         might participate in actions to help raise consciousness, to pressure
         representatives, or to promote local and state reform initiatives.
      - Take over
         - Local and state citizen initiatives
         - Push legislatures to form nonpartisan citizen commissions to
         study the issue and make recommendations
         - Eventually, push for an Article V constitutional amendment
         convention, called by the states, to implement this on a federal level
            - Such a convention should also address issues of campaign
            finance
            - Voting reform does not, in principle, require constitutional
            change. However, an Article V convention is a way to avoid
the dysfunction
            and status-quo-bias of the US congress.
            - The delegates to the convention should be a randomly-selected
            sample of citizens. This avoids the pathologies of passing
the job to
            professional partisans.
               - In order to prove this concept, activists should fund and
               organize "deliberative polls" such as the one in California.

   Note: these strategies owe much to Lawrence Lessig's work on
   "rootstrikers".

   [1] 58% opposed Afghan war in an AP-GfK Poll in August 2010 (before Bin
   Laden was killed); 59% opposed it in a USA TODAY/Gallup Poll in May 2011
   (after his death).

   [2] 47% felt that free trade agreements hurt the US, and 69% felt that
   they cost jobs, as opposed to only 23% and 18% who took the opposite
   position, in a NBC polls in Nov 2008 and Sep 2008 respectively.

   [3] 56% of those polled by Bloomberg (Mar 2010) said they would support
   government action to limit compensation of those who helped cause the
   financial crisis, or to ban those people from working in the banking
   industry.
   http://www.bloomberg.com/apps/news?pid=20601087&sid=a4nQoiYaj2ag&pos=1

   [4] 61% oppose large farm subsidies; this proportion is basically the
   same across parties.
   http://www.worldpublicopinion.org/pipa/articles/brunitedstatescanadara/602.php
   *
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120107/cd63e69b/attachment-0003.htm>


More information about the Election-Methods mailing list