[EM] serious strategy problem in Condorcet, but not in IRV?

[Dave Ketchum]

On Tue, 19 Aug 2003 12:31:13 -0400 James Green-Armytage wrote in part:

>>James, 
>>your own example 
>>Sincere preferences
>>46: A>B
>>44: B>A
>>5: C>A
>>5: C>B
>  
> The problem with plurality (and to a lesser extent, with IRV, though not
> in this example) is that the C voters are forced to choose between a
> sincere vote and the lesser of two evils. In the terminology stated by
> Markus in a recent posting, this is the "compromising" strategy. 
> 
> I agree that it is a very serious problem to force voters to make this
> kind of compromising choices on a regular basis, as plurality clearly
> does. But what Monroe is concerned with is more so the "burying" strategy,
> and its possible consequences. In this particular example, what would
> worry him (I think) is that A and B voters will both try to get a bury
> each other, and will inadvertently elect C as a result, which I agree
> would be a public disaster.
> 
> James
>  

Perhaps too many things are getting labeled simply "strategy", defined as 
an undesirable activity.  Perhaps it can be divided into "normal strategy" 
and "devious strategy":

Normal strategy:
      In Plurality this is about all I see.  Looking above, A and B 
backers need not think beyond voting their preference.  C backers get to 
choose among what I see as sincere votes (or imitations that outsiders 
cannot distinguish) (agreed that this makes Plurality an undesirable method):
           Vote C to demonstrate backing and express neutrality as to A vs 
B or
           Vote A or B to help their preference between these win.

      In Ranked Balloting the above shows A and B competing in the same 
area, thus backing each other as second choice.  C backers can both 
express C as first choice and their preference among A and B as second choice.

Devious strategy (trying to accomplish your desires by voting something else):
      In Plurality I see nothing.  A and B backers cannot do better than 
their own candidates.  C backers are simply expressing their strongest desire.

      In Ranked Balloting:
           I read above of "burying" - ranking a competitor at the bottom 
of the pack, BUT:
                If the competitor is that undesirable, this is normal voting.
                If this causes a less desirable competitor to win, this 
was self destructive deviousness, visiting deserved pain on those so dumb.

      In IRV there are many examples of undesirable results, tending to 
depend on vote patterns hard to duplicate for devious purposes in real 
elections.

      In Condorcet another example shows B backers switching from B>A to 
B>C and causing B to win.  Agreed this is devious, but it depends on exact 
vote patterns (beyond what can be planned in normal real elections).  A 
voters can extend this by bullet voting (saying they do not care between B 
and C).  Agreed this is devious, but the A voters are accomplishing 
exactly what they claimed to want (if they are unable to win, they choose 
to leave the decision to B and C voters - who are saying C is acceptable).

      In Approval borderline candidates force a difficult decision:
           Rank them as if being as desirable as those you like best - 
perhaps causing them to win over those you like better.
           Rank them as if being as UNdesirable as those you like least - 
perhaps causing them to lose to one you like less.

An example I used last month, designed to show difference between IRV and 
Condorcet:
>      40 A
>      30 C>B
>      30 B
> This is incomplete - one last ballot to count.  Last voter votes:
>      A - odds are against this, and leaves us a tie problem.
>      B - IRV and Condorcet agree that B wins:
>           IRV - C becomes loser.
>           Condorcet - B>A and B>C.
>      C>B - disagreement:
>           IRV - B loses; C loses; A wins; GREAT unhappiness among B backers.
>           Condorcet - a cycle, B>A and A>C and C>B, but B>A is stronger 
> than C>B, so B backers are pleased.

Room for IRV deviousness here - getting any voter to convert to C>B lets A win.

-- 
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.