[EM] STV-CLE

[James Green-Armytage]

Dan Bishop <daniel-j-bishop at neo.tamu.edu> writes:
>Good point.
>Well, what if instead of eliminating the overall Condorcet loser, we 
>eliminated the Condorcet loser *of the excess votes*?
>That is, once A and B had their 25 votes, the elimination would be based
>on:
>16: C>E>D
>14: D>E>C
>20: E>C>D
>and D, being the Condorcet loser of this, would get eliminated, causing 
>E to meet quota.  This would produce the same result as STV and CPO-STV.

Dan,
	I think that the basic criticism still holds. Try this modified version
of my example.

3 candidates, 100 voters, NB quota is 25.
21 ABXCDE
16 BAXCDE
14 XABCDE
15 CEDABX 
14 DECABX
20 ECDABX

	The main difference is that I've added another candidate to the AB
faction, so that nobody has a quota to begin with. E is the Condorcet
loser, so I assume that the CLE-STV's first step would be to eliminate E.
However, I believe E is selected by both STV and CPO-STV, and I think this
result {A, B, E} is most consistent with the goal of proportional
representation. It's very late at night here, so I apologize if I've made
any typos.
	I remember discussing this method back in 2003; I found one entry in the
archives for August... there may be more. My question for you is: why not
CPO-STV?

my best,
James