[EM] Re: Improved Generalised Bucklin

[Eric Gorr]

(oops...needed to make an obvious correction and more observations)

At 3:05 AM +0930 8/23/03, Chris Benham wrote:
>In reference to this example:
>31:  B>A>E>C>D
>23:  C>B>A>E>D
>25:  D>A>C>E>B
>11:  D>C>B>A>E
>10:  E>A>C>B>D
>100 voters, the Smith set is ABC.
>On Thursday,August 21, 2003 Eric Gorr wrote:
>"This example contains a simple cycle between ABC. It is clear that 
>DE are not preferred over ABC by any majority and shouldn't have the 
>possibility of winning.
>So, the question becomes, who should be victorious in the cycle. I 
>cannot come up with any good reason to believe that should be anyone 
>other then A."
>Eric, for this observation to be especially relevant,I assume that you CAN
>"come up with a good reason to believe" that A should win. What 
>reason might that be?

I assume you have run your example through a RP implementation and 
taken a look at the detailed output.

If not, you can head to http://www.ericgorr.net/condorcet/.
In the output you will see:

   1 0 65 30
   0 2 66 32
   2 1 69 38

0 = A
1 = B
2 = C

Now, the defeat that could cause A to lose is the weakest defeat here.

More people would rather have C then B.
More people would rather have A then C.

Since the fewest number of people would rather have B then A, it gets ignored.

Simple cycle. Simple & clear resolution.
I can think of no reason why it should be resolved any other way.

One of the primary flaws in IRV is that it places far to much 
emphasis on the top choices of the voters, while assuming the middle 
and lower choices are less important.

Now, your method comes along and solves part of that flaw by assuming 
the emphasis should be placed on the top and the bottom preferences, 
but that the middle choices are less important.

It is interesting to note that this is where A has been placed in 
your example...in the middle....no one has placed A as their top 
choice and no one has placed A as their bottom choice.

IRV selected C because of it's emphasis on the top choices of the 
voters, but it is also interesting to note that more people (66%) 
would rather have A then C.

Candidates BDE all appear at the bottom and top of the preferences 
and it would appear that B is generally as liked as it is disliked.

A benefit of Condorcet is that it makes no assumptions about the 
importance of any of the preferences expressed by a voter.