[EM] Request for help: complex election

[Adam Tarr]

Nikhil Nair wrote:

>The modifications I was thinking of was simply dividing the number of
>votes by the turnout (e.g. 12 out of 15 from the Council would be 75%),
>before multiplying by 0.6 or 0.4.  Then we're comparing apples with
>apples, and variation in turnout wont' matter.
>
>My final twist was the veto (at least 20% from both), so we get less than
>4 if there aren't 4 good candidates.
>
>So, have I missed a fatal flaw somewhere?

Only that there's no proportionality in this system.  So, if 75% of the 
voters want one set of four candidates, and 25% of the voters want another 
set of four candidates, then you will get a 4-0 split rather than the 
appropriate 3-1 split.  Even worse, 51% of the voters could get 4 
candidates, leaving the other 49% of the voters with no representation.

If you insist on not using a good proportional method like STV-PR or PAV 
(Proportional Approval Voting), then a simple alternative is "cumulative 
voting".  It works exactly the same as regular voting, except you may cast 
any or all of your four votes for any candidate.  So you could vote for one 
candidate twice and another candidate twice, or four different candidates 
once, or one cadidate four times, or one candidate three times and another 
once... you get the idea.

This allows minority factions to still get their fair share of the 
electorate, by concentrating their votes on a smaller set of 
candidates.  So, the whole method would go:

1)  Every voter may cast up to four votes, and may vote for the same 
candidate multiple times.

2)  Multiply all votes from the general population by .4/(voters), and 
multiply all votes from the council by .6/(council members)

3)  Total the votes, and eliminate all candidates whose vote total does not 
exceed .2.

4)  Elect the four remaining candidates with the highest vote total.

Hope that helps,
Adam