Jameson Quinn, Nov. 2015
The indefatigable Jameson Quinn with a 3rd proposal of a one concrete tentative system suggestion for Canada. I am putting it here (/CanadaSA8.html) just to serve as a reference point for further discussion.
I have a new version of my latest proposal. There are four basic principles:
Here's a more detailed explanation:
1. Each riding's ballot lists the candidates, and allows voters to "prefer", "approve", or "disapprove" of each. A voter gets their local ballot by default but may choose another ballot. Each candidate has two tallies: their score and their transferrable votes. A candidate gets two points of score for each "prefer", and one point of score for each "approve". Each ballot gives one transferrable vote, divided evenly between "preferred" candidates.
2. a. The candidate with the highest score in each riding wins.
b. The transferrable votes for each party are totaled. For each seat the party has already gotten, one Droop quota of transferrable votes is subtracted proportionally from all candidates in that party. (If this results in a negative total for any party, that party is eliminated and the size of the Droop quota is adjusted for following steps according to the remaining seats and votes).
3. For each full Droop quota of transferrable votes a party still has, they get 1 seat. Seats are assigned to the candidates from that party who did not win locally, in order from highest to lowest score. (If any party does not have enough candidates to fill all their seats, they transfer their extra votes as in step 4.)
4. a. Whichever party has the fewest transferrable votes remaining is eliminated. Their votes are transferred to other surviving parties according to the preferences of the candidates who hold them.
b. Repeat steps 3 and 4a until all seats are filled.
There are three additional optional modified rules you can use:
m2a: Tally score from within-riding ballots and that from out-of-riding ballots separately; and also count the total ballots cast from each riding (that is, by voters living within that riding). If any candidates' within-riding score is more than the number of ballots cast from that riding, then one such candidate must win in 2a (even if another candidate has a higher score when counting out-of-riding ballots). Since scores are added 2/1/0, this has the effect of guaranteeing that any candidate preferred by a local majority or unanimously approved locally cannot be unseated by votes from elsewhere.
m2b: Subtract transferrable votes from elected candidates, down to zero, before subtracting the remainder proportionally from all candidates not elected in a party.
m3: Keep the same separate tallies as in m2a. No candidate whose total score is less than half of the ballots cast in their riding may win a seat. This works like Germany's 5% thresholds to prevent a small party from getting seats with only a small minority in each riding. In particular, a party would need a "stronghold" where they could get at least 25% preference and/or 50% approval. Note that small organized parties could choose to concentrate their votes in one riding and still get a seat; but halfhearted "protest votes" spread out over a country would not end up electing a poorly-vetted candidate.
This is an extremely streamlined system. It's easy to understand and easy to use. It ensures party proportionality, with enough incentives for unity to prevent useless party-splitting, but still gives a fair shake to credible independent candidates and well-organized small parties. It ensures that no candidate is "voter-proof." It gives some degree of voter control over ideology at an intraparty level. The only jargon in the definition is "Droop quota", which can easily be defined in one simple sentence or equation. In particular, it is much, much simpler to understand from a voter's point of view (as well as being better at ensuring candidates are not voter-proof) than anything involving "juice", "evaporation", and/or time-limited "free-for-all" vote transfers.