[EM] Fwd: Is Condorcet The Turkey?

[Dave Ketchum]

I read three later posts debating how you count votes in Condorcet. 
Rather than getting in the middle of that:

In Condorcet each candidate ranked by the voter gets counted as better 
than EVERY other candidate, except those the voter lists as preferred over 
this candidate.  Thus, with candidates A, B, C, and D and 9 voters voting 
A>B, we add 9 to the matrix for EACH of:  A>B, A>C, A>D, B>C, and B>D. 
These voters did not rank C vs D so nothing gets added to the matrix for 
this pair.

On Sat, 7 Jun 2003 07:53:35 EDT Dgamble997 at aol.com wrote:
>  
> 
> 
> ------------------------------------------------------------------------
> 
> Subject:
> Is Condorcet The Turkey?
> From:
> Dgamble997 at aol.com
> Date:
> Sat, 7 Jun 2003 07:33:35 EDT
> To:
> election-methods-list at egroups.com
> 
> 
> Hello List
> 
> Recently there have been a large number of postings regarding something 
> called the turkey problem- indifferent/poor candidates winning as the 
> least worst choice in a condorcet ballet.
> 
> Due to the nature of the Condorcet method- which considers lower 
> preferences before the fate of higher preferences is decided- condorcet 
> would appear to me to be a turkey electoral system.

First note:  Condorcet considers ALL preferences in pairs simultaneously, 
though often in two steps:
      If one candidate is preferred over each and every other candidate, 
that is the winner.
      Otherwise we have a cycle such as A>B, B>C, and C>A, the other 
candidates are outside the cycle, with each of the cycle candidates being 
preferred over each of them, and the law BETTER provide a formula for 
choosing among these near ties within the cycle.
> 
> The fundamental difference as regards the results between condorcet and 
> irving is that  condorcet allows the election of candidates of 
> candidates with little or no first or higher preference support. It is 
> these candidates who are likely to be the turkeys.
> 
> Take the following example:
> 
> Candidate A takes distinct policy positions on a range of issues, 
> because of this he gains a great many enthusiastic supporters but also a 
> great many vehement opponents.
> 
> Candidate B takes opposing positions to candidate A, because of this she 
> also gains many supporters and many opponents.
> 
> Candidate C does nothing, says nothing and offends nobody.
> 
> The votes are cast as follows:
> 
> AC  498
> BC  497
> CA     3
> CB     2
> 
> Under condorcet candidate C as the least objectionable wins.
> 
> It occurs to me that the best way to win a single seat condorcet 
> election is to stand on a platform of bland, vague and generalised 
> policies that nobody could disagree with or be offended by.

Only works in this special case, where A and B have each offended about 
half the voters and half have said A is better than B while half have said 
B is better than A.  I expect each half would prefer this compromise over 
a total loss to their worst enemy.
> 
> Politics is fundamentally about disagreement, about different people 
> supporting positions and policies that are different to each other and 
> often in opposition to each other. Compromise and consensus are good 
> things but not at any cost.Condorcet appears as a ' compromise at any 
> cost' electoral system. It will elect candidates whose only merit is 
> that they offend no-one.

IRV can also produce ugly results.
> 
> Irving also elects compromise candidates, however under irving a 
> candidate must get a reasonable proportion of first/higher preference 
> support to reach next stage of the election and win. A candidate cannot 
> win by being everybody's last preference.

Go back and look carefully, and you should notice that C won by being 
NOBODY's last choice.  To win in Condorcet you must convince a lot of 
voters that you do not deserve being last preference.

My favorite IRV example has 40 A>B, 60 B>A, and IRV awarding A the win 
because 31 voting C>B>A makes if fail to notice the basic preference for 
B.  Here the voters are agreed that B is preferred over A, but a majority 
of the B voters like C even better than A or B.

Note:  Here C got "a reasonable proportion of first/higher preference 
support" and reached the nest stage by clobbering B before B's total 
strength was seen by IRV.
> 
> Compromise candidates who attract a reasonable level of higher 
> preference support are probably good people to elect. They have said or 
> done something to make people positively support them. Compromise 
> candidates who attract little or no higher preference support, who have 
> said and done nothing and whose sole merit is that they offend and upset 
> nobody are probably not good people to elect.
> 
> Irving elects the first type of compromise candidate but not the second, 
> condorcet elects both. This is why irving is better than condorcet.

Most of the time IRV and Condorcet produce identical results.  We only 
debate around the edges, and I claim your Condorcet example is invalid 
because it involved people voting FOR C as better than A and better than B.

I claim approval for Condorcet for noticing how most voters ranked A vs B, 
even though some of these ranked neither A nor B as best.
> 
> David Gamble
-- 
  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.