perspicuity mentioned the Australian multi-party system and suggested some kind of vote-for-and-against system for the US, in which candidates had to receive more yes than no votes in order to be able to win. This got me thinking, and I ended up wondering whether something like the following would work:
- Your ballot lists all declared candidates for each position, plus two write-in candidate boxes per position, plus a line for NONE OF THE ABOVE.
- For each of the above, you award a number of points from 1 to 5, where 5 points is STRONGLY LIKE, 3 is DON'T CARE, and 1 is STRONGLY DISLIKE. Any line left blank, including NONE OF THE ABOVE, scores an automatic 3 (neutral/don't care).
- When tabulating votes, 3 is subtracted from the points awarded, to yield an actual adjusted vote score in a range of 2 (STRONGLY LIKE) through -2 (STRONGLY DISLIKE). All votes cast for that candidate, for and against, are then summed to arrive at a total.
- Every candidate (except write-in candidates not otherwise appearing on the ballot) who does not receive more net adjusted votes than NONE OF THE ABOVE is eliminated, and is disqualified for life from running again for that office. If no candidate running for a given office in a given district receives a higher net adjusted vote total than NONE OF THE ABOVE, then all candidates for that district are disqualified, and the district must be resolved separately in a by-election with all new candidates. A candidate disqualified by a NONE OF THE ABOVE result is not a valid write-in candidate for future elections.
- If candidates remain who scored a higher net adjusted vote total than NONE OF THE ABOVE, the candidate having the highest adjusted vote total in any given district wins that district. If two candidates have the same total score within a margin of error of 1%, the tie is resolved by determining which of them received the lowest total of NEGATIVE adjusted votes, i.e. which of the tied candidates is least disliked. If no candidate has an adjusted vote total greater than zero (neutral), then since NONE OF THE ABOVE receives a default neutral vote, it is likely that NONE OF THE ABOVE will win.
(The 1% margin of error is intended to reduce the number of squabbles and re-re-re-recounts over handfuls of votes. The exclusion of write-ins from NONE OF THE ABOVE elimination removes the possibility of a well-intentioned write-in vote for a candidate who receives few votes from having the unexpected side-effect of disqualify that candidate from any future election well, is completely moot actually, since as cipherpunk points out in the comments, it won't pass Constitutional muster.)
There is one obvious weakness here. If any single party has overwhelming (>50%) support in a given district, its voters can permanently eliminate all other candidates in the election by voting a 5 for their candidate, a 4 for NONE OF THE ABOVE, and a 1 for every other candidate. However, assuming that this gambit is uniformly applied by voters of all parties, in any district in which no party has greater than a plurality the result will be to eliminate all candidates from all parties running in that district; thus it is self-defeating except in districts completely dominated by a single party, in which case that party was going to win the seat anyway. And that entire paragraph is now moot as well. In any case, I've since realized that it's not a "safe" tactic anyway unless the majority party has considerably more than 50% of the vote ... I haven't done the math, but can safely say that if you have a sufficient majority of voter support to make it a safe tactic, you're almost certainly in no danger of ever losing the seat anyway.
This still doesn't yet address the issue of proportional representation, i.e, if party X receives a significant statewide vote, but does not have sufficient votes clustered in any single district to actually win a district, its voters go unrepresented. But here I propose a means to fix that, by taking a different path at step 5:
- [Alternate] After candidates and districts eliminated by NONE OF THE ABOVE are removed from consideration, all remaining votes are added up and the net adjusted vote totals for each party are used to determine the number of districts won by each party.
- Once the number of seats won by each party is determined, winners of individual districts are selected, starting with the party that won the LEAST number of seats, and awarding each party its allotted number of seats in order of those seats in which it received the greatest proportion of the vote, EXCLUDING seats in which all candidates were eliminated by NONE OF THE ABOVE (which must be resolved separately in a subsequent by-election). After all seats due to the party winning the least number of seats have been awarded, seat selection passes on to the party winning the next greater number of seats.
Step 6 here is designed to ensure that minority parties receive their fair share of seats in the legislature, and at the same time ensure that their representatives in the legislature come from the districts in which they have the strongest support, even if they do not have sufficient support in any single district to win it in a first-past-the-post winner-takes-all election.
This will of course still result in some seats electing representatives who are not the favorites of the majority of voters in their district. That is almost inevitable in any multi-party system, though. What this system should ensure is that, even if your vote does not elect a candidate of your party in your district, at the state-wide level no vote for a minority party is wasted unless that party has so little support it cannot earn enough votes to win even a single seat, and also that minority parties earning enough state-wide votes to (proportionately) win at least one seat will represent the electoral districts in which they have the strongest support. It does this without requiring multiple recounts, as transferable-vote systems almost invariably do.
So. Thoughts? Did I miss anything obvious, besides the one obvious weakness discussed above?