Saari reply

[Dave Ketchum]

On Sat, 22 Jun 2002 20:54:04 -0700 (PDT) Alex Small wrote:

>>>No, sorry, there does not need to be a near-tie in order to have
>>>cycles.
>>>
>>
>>MUST be approaching a tie, even if you do not like "near" as an
>>adjective.
>>  We know that while A>B and B>C are significant enough for both to
>>  head
>>for the winner's circle, C>A also MUST have significant backers or we
>>could not be cyclic.
>>
> 
> First a simplified example of cycle without near-tie, then discussion:
> 
> 33% A>B>C
> 33% B>C>A
> 34$ C>A>B
> 
> Obviously these exact numbers will happen about as often as elections
> where the crucial electoral votes come from a state whose governor is
> related to one of the candidate--rare, unlikely, but ever discount it
> because then it will happen.


When I look I see:
     67 A>B vs 33 B>A
     66 B>C vs 34 C>B
     67 C>A vs 33 A>C
My eyes see equal strength, a perfect tie, between A and C.

Agreed that unwanted events SOMEHOW gain in likelihood if we do not 
prepare for them - a reason for arguing against IRV.


> Now to the real world:
> 
> If we stick to a world where most people view politics via a left-right
> spectrum then you are right, cycles only occur in close elections.  But we
> just saw in 2000 that close elections can happen, and when they do it's
> best if everybody has already agreed in advance on what the rules are. 


2000 was a close election, but I see no cycles for it was basically a 
2-candidate election (Nader disturbed some, but was not strong enough to 
make a cycle).


> Otherwise, some people might become very angry...
> But, in a world where third parties fit into categories other than "Even
> more liberal than the Democrats" or "Even more conservative than the
> Republicans" (a world I'd like to inhabit some day) then situations which
> are roughly similar to my simplified example can happen.
> Imagine three candidates:  A liberal (in the usual, contemporary American
> politics sense of the word), a conservative (same disclaimer), and a
> candidate who's somewhere in the middle on domestic policy but has a very
> different vision of foreign policy (a field where left-right isn't as
> well-defined).
> The liberals mostly place the third party candidate in second place
> because they're mostly interested in domestic policy, and they agree more
> with the third-party candidate.
> The third-party people, being free-traders, tend to place the conservative
> second because of the ostensibly market-oriented policies of
> conservatives.
> The conservatives decide they prefer the status quo, and place the liberal
> in second place (better the devil you know than the devil you don't).
> Three factions, each with plausible and respectable reasons for their
> preferences, and together they form a cycle.  Not likely to happen as long
> as we keep plurality, because all new ideas are filtered through a
> two-party system.  But when extra parties become competitive it is
> plausible that a third coalition could form that doesn't fit on a neat
> spectrum.


Plurality voting encourages having two parties, each positioning 
themselves on opposite sides of the center, but as close as possible for 
strategy, and discouraging voting for third parties by complaints about 
spoilers.  Condorcet is one method that does away with the spoiler problem 
and thus gives third parties the right to the votes their platforms deserve.


> If you disagree with my characterizations of the left and right, or of
> foreign policy, well, that isn't really the issue.  The issue is whether
> or not cycles require near-ties, and I'm arguing that with the injection
> of a very new perspective cycles can occur without near ties.


Let me try a different vote:
  49% A>B>C
  49% B>C>A
   2%

When I look at this I see:
     51 A>B vs 49 B>A
     98 B>C vs  2 C>B
     51 C>A vs 49 A>C

Or:
  50% A>B>C
  50% B>C>A

When I look at this I see:
     50 A>B vs 50 B>A
    100 B>C vs  0 C>B
     50 C>A vs 50 A>C

This crew clearly likes B, and gives A and C a tie as losers - not near a 
tie, but hard to get farther away and still have a cycle.
-- 
  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.

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em