[EM] IRV, the turkey problem, and everything

[Dave Ketchum]

On Sun, 6 Jul 2003 11:52:41 -0700 (PDT) Alex Small wrote:

> It seems to me that much of the discussion over "which method elects the
> best/worst/most mediocre candidate?" misses a larger problem:
> 
> In a system founded on the premise that the citizens control the
> government, the goal should be to present the voters with a broad range of
> choices in each election.  OK, not so many that you get a 4-page ballot
> like San Francisco had last year (they voted on what type of coffee to
> allow, if I'm not mistaken), but enough so that it isn't just "same old
> vs. same old, with those other kooks that have no chance (aka Third
> Parties)".

The method used needs the ABILITY to handle a broad range of candidates, 
and to evaluate the voters' opinions about them.

While Plurality does well with two candidates, often there is only a 
single candidate of interest - and I see no reason that a more complex 
method would necessarily change this.
> 
> We know that plurality nurtures a two-party system.  Although there are
> variations within the parties (the Dems are home to Joe Lieberman and
> Dennis Kucinich, while the GOP is home to John Ashcroft and Lincoln
> Chaffee), and although some argue that the two parties are
> indistinguishable on issues that matter (a subject to debate elsewhere)
> one thing is clear:
> 
> In any given plurality election, it is almost certain that there are only
> two viable candidates, and while the viable candidates may differ from
> place to place, they are almost always from the same two parties, and each
> party strives to put aside differences and form a united front in the
> various legislative bodies (national, state, and local).
> 
Likely that few have read this far unless they already agree it is time to 
get past Plurality.
> 
> Now, whether you worry about turkeys, Hitler-Stalin-Washington polarized
> scenarios, lukewarm Al Gore bland inoffensive candidates, or whatever, I
> hope we all agree that it would be better to have more options competing
> in the marketplace of ideas.  How to take steps toward this?
> 
> We know that proportional representation fosters multi-party systems. 
> However, it is likely that single-winner election reform will precede the
> implementation of proportional representation.  Supporters of each method
> say "Ours is the best for increasing competition in the marketplace of
> ideas."  Let's evaluate them:

Whatever merits PR may have for electing legislatures, we like 
single-winner elections for mayors and governors.
> 
> IRV-  IRV has failed to achieve significant multi-party competition in
> Australia.  There's been some progress, but PR has done a much better job.
>  I think it is reasonable to ask whether other methods can do better,
> because IRV has proven to be a modest improvement but nothing to crow
> about.
> 
See Ranked Ballots below.
> 
> Approval-  Approval is largely untested in the real world.  At best, it
> will lead us to a Utopia of multi-party competition.  More likely, it will
> lead to a moderate level of competition, hopefully a little better than
> what IRV has given Australia.  At worst, we'll still have a duopoly, sans
> spoilers (in the sense of people with 2% support tipping elections one way
> or the other).  Moreover, Approval is cheaper than IRV in terms of voting
> equipment.  Approval is therefore worth trying in public elections,
> preferably partisan elections and primaries rather than non-partisan local
> elections.
> 
True that equipment for Approval can be cheaper than for Ranked Ballots, 
BUT this happens to be a season for buying new equipment, for which 
fully-capable equipment should cost little more than semi-capable 
equipment.

More difficult than Ranked Ballots for the voter debating whether to 
classify a marginal candidate as acceptable:
     Approving is as strong as the most the voter can say for a truly 
desirable candidate - and could result in marginal defeating desirable.
     Rejecting is as strong as the most the voter can say against a truly 
UNdesirable candidate - and could result in marginal losing to UNdesirable.

Puzzling to group primaries with partisan elections, for primaries are 
non-partisan just as are many local elections.
> 
> Condorcet-  Condorcet is largely untested in the real world.  The
> prospects are probably the same as Approval, but Condorcet is more
> expensive and complicated.  Condorcet should therefore not be seriously
> considered for public elections right now.
> 
I group Condorcet with IRV in the following Ranked Ballot discussion.

Ranked Ballots- both methods referred here from above share voting rules 
and let each voter indicate the most desirable candidate, followed in 
order of desirability by as many more as the voter chooses - the intent 
being that the counting shall pick the one most desired collectively.

IRV does not look at whole ballots, and thus sometimes does not see the 
most desired - resulting in horror stories.

Condorcet does look at whole ballots and does see the best liked - and 
sometimes raises talks of "turkeys" when it recognizes as rejects those 
WELL-liked by a few but DISliked by many.
     Also produces a matrix identifying how well the candidates did in 
relation to each other.  This should help in letting minor parties 
participate, for they can get in with a few votes without disturbing 
the major competition - and produce a record indicating how well the 
voters accept their platforms.
     Can group near ties in a cycle requiring further analysis - 
annoying, but the cycle members are those preferred over the remaining 
candidates.
> 
> Borda-  Don't even go there ;)
> 
> 
> 
> So, if your goal is competition rather than "filtering out the
> turkeys/polarizers/etc." then the basic question is "Can we do better than
> IRV?"  Approval is an excellent method for examining that question, and
> the question is worth asking in light of IRV's under-performance in
> regards to competition.  It could be that PR is the only way to really get
> multi-party competition going, but that's another topic.

Turkeys are discussed elsewhere, beyond my claim that what gets described as 
turkeys do not look that ugly to me.

I see Condorcet as noticeably better than Approval or IRV.
> 
> Anyway, just my $0.02 worth.
> 
> 
> 
> Alex

-- 
davek at clarityconnect.com  http://www.clarityconnect.com/webpages3/davek
  Dave Ketchum   108 Halstead Ave, Owego, NY  13827-1708   607-687-5026
            Do to no one what you would not want done to you.
                  If you want peace, work for justice.