[EM] Fluffy the dog: rejection of Condorcet

[Dave Ketchum]

On Tue, 4 Sep 2001 23:21:57 -0700 Craig Carey wrote:
> 
> At 01.09.04 16:40 -0400 Tuesday, Dave Ketchum wrote:
>  >On Tue, 4 Sep 2001 08:35:54 -0700 Craig Carey wrote, in part:
> ...
>  >This caught my eye so, thanks to eGroups/Yahoo, I could and did go back
>  >to this item in Nov. 00 archives:
>  >
> 
> I was thinking of my repair to that Fluffy example.
> 
>  >I do NOT see this as an indictment of Condorcet.  These voters ALL SAID
> 
> Condorcet finds the wrong number of winners sometimes.
> Condorcet is able to be seen as already rejected.

I do not understand these statements.
     All of these methods seem to be in the business of finding a single
winner, with the exception of an occasional tie.
     If Condorcet really fails, I am all ears.
> 
> ...
>  >     Aside from the basement thought, I do not like anything that
>  >invites strategic voting.  For those who understand the strategy, it
> 
> That could be similar to expecting monotonicity to be held.
> 
> ...
>  >As to hundreds of candidates - how can any voter be expected to pick
>  >intelligently among such a crowd, regardless of method of counting votes?
> 
> Post up the equations and I will have a look at them. Unless there is
> some problem with that (nonexistence?).

In one sense, a good formula should continue to satisfy for any number
of candidates.

In a second sense, if the formula fails after 100 candidates, and there
would never be more than 50, the formula is acceptable.  We do have,
assuming the district is a whole state, need to consolidate results from
multiple precincts.  Results tables quickly grow beyond manageability
for IRV; remain tolerable with Condorcet.
> 
>  > davek at clarityconnect.com    http://www.clarityconnect.com/webpages3/davek
>  >   Dave Ketchum     108 Halstead Ave, Owego, NY  13827-1708    607-687-5026
> ----
> 
> This original "Fluffy" example has such problems with its numbers that new
> numbers could be used and the original discarded. Here is an improved
> version:
> 
>        AB  48     : 1 winner        (no. 1)
>        B    3
>        CB  49
> 
>      Condorcet: B wins   :   A:B = 48:52, B:C = 51:49, C:A = 49:48
>      FPTP:      C wins
> 
> Candidate B wins, and: (1) the outcome is too different to First Past the
> Post to some, and (2) allowing that would lead into problems.
> 
> Condorcet can be rejected for picking the wrong number of winners even
> without this example.

I like B winning, for every voter considers B to be acceptable, while
more than half declare each of A and C less acceptable.

Why am I supposed to care about First Past the Post, considering the above?

Why wrongly?  Certainly it should not surprise if altering votes makes
results change.
> 
> If Mr Layton writes, perhaps an 'improved' Fluffy the Dog example could
> be named.
> 
> At 2000.11.13 12:05 +1100 Monday, LAYTON Craig wrote:
>  >There is basically one cogent criticism of all Condorcet systems, of which
>  >you're all aware; that it puts too much importance on middle preferences,
>  >especially when not all candidates or parties are known to voters.  Example:
>  >
>  >In a two candidate race:
>  >A 51
>  >C 49
>  >
>  >C wins
>   A wins
>  >
>  >A and C are diametrically opposed, there has been a long and dirty smear
>  >campaign, so that C voters hate A and A voters hate C.  My dog, fluffy (B),
>  >joins the race at the last minute;
>  >
>  >ABC 49
>  >CBA 49
>  >BAC 01
>  >BCA 01
>              A:B = 49:51,  B:C = 51:49,  C:A = 50:50
>  >
>  >Fluffy is the Condorcet winner!.

And I am happy, for ALL the voters agree that B is acceptable, more than
half like B better than A, and more than half like B better than C.

That Fluffy may have had a misleading resume is not our problem.
> ...
-- 
 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