[EM] STV+AV

[Kristofer Munsterhjelm]

On 02/02/2012 07:24 AM, Bryan Mills wrote:

> Single-winner is required by 2 USC Sec. 2c:
>
>     [...] there shall be established by law a number of
>     districts equal to the number of Representatives to which such
>     State is so entitled, and Representatives shall be elected only
>     from districts so established, no district to elect more than one
>     Representative [...]
>
>
> I can't find a proper citation for requiring FPTP in the source where I
> saw it; that part may be mistaken.  So that might might admit the
> possibility of using an alternative single-winner method within
> districts, but it's not at all clear to me that that would help
> significantly given the susceptibility of single-winner districts to
> gerrymandering.

Could you get around this by making "show districts" that are linked to 
the candidates, but having the election happen on a larger scale?

Say you have a state with 5 districts, and you run a statewide 5-member 
STV election for the House. Say further that a {candidate, district} 
pair's score is the number of first preference votes that candidate got 
in that district.
After the 5-member STV election, you then assign winners to districts so 
as to maximize the sum of the districts' scores, subject to that a 
district can't be assigned to an out-of-district candidate.

Does that run afoul of "no district to elect more than one 
Representative"? If so, the question is how much interaction would be 
permitted.

For instance, you could imagine a party-based system where, if there 
were disproportionately many Democratic candidates, the power of each 
Democratic-candidate vote would be attenuated until proportionality is 
restored. That kind of scheme would have proportional representation, 
but no voter would vote for an out-of-district candidate. (It would also 
have the weird result that the FPP winner of a district might not 
actually win that district.)