In November 2012, San Francisco voters will vote on a measure introduced by supervisor Mark Farrell, which would replace Instant Runoff Voting (aka Ranked Choice Voting) with traditional vote-for-one ("Plurality" or "First-past-the-post") voting, for city- and countywide offices like mayor and district attorney. Supervisor elections would retain the current Instant Runoff Voting system.
Under Farrell's proposal, general elections would occur in September. And if no candidate received at least a 65% supermajority of the vote, then a November run-off election would be held between the two top vote-getters.
This measure would give the conservative wing of the board every advantage:
1) A low-turnout primary.
2) A virtually guaranteed runoff that gives the Chamber of Commerce another bite at the apple.
3) Voters whose sincere favorite candidate was not a perceived frontrunner would tend to cast an insincere tactical vote for the "lesser evil" among the frontrunners (e.g. a Nader supporter who votes for Gore so as not to get Bush) — which means all they'd have to do to prevent a principled populist from winning is to convince her supporters that she "couldn't win".
A countermeasure should try to address both elements as simply as possible. It should ensure that:
1) It has the highest possible chance of defeating the Farrell measure. (That is, it doesn't have to be perfect — it just has to be better than Plurality-with-runoff voting, and it has to have a good chance of success.)
2) Voters are never punished (get a worse result) for supporting their sincere favorite candidate (e.g. get Bush instead of Gore because you voted for Nader).
3) A popular political platform never loses just because too many candidates run on that platform and "split" the vote. This is also called the "spoiler" problem.
We therefore propose a modified form of Plurality Voting that we call "Count All the Votes", with elections held in November, covering both citywide elections and supervisor elections.
As the graphic at top demonstrates, the ballot looks exactly the same as an ordinary ballot. The only difference is that you may vote for one or more candidates. Or put another way, instead of treating over votes as spoiled ballots, we count them — thus this system makes it virtually impossible to spoil one's ballot. And just as with Plurality Voting, the candidate with the most votes wins.
This system has been studied and advocated by political scientists for decades, under the name "Approval Voting". This is due to the similarity to a political approval poll, asking which candidates the pollee approves. Here's a 1983 article about the use of Approval Voting to gauge support for the Democratic nominees for President. Here's a 1989 article in favor of the system, by the current provost and senior vice president of Carnegie Mellon University. (While Approval Voting is the prevailing name for the system we're proposing, we also like the term Count All the Votes, because it emphasizes that this is just ordinary simple Plurality Voting, with the modification that over votes are counted rather than discarded.)
To address our three previous bullet points:
1) It has the highest possible chance of defeating the Farrell measure. Polling shows that Instant Runoff Voting in any form is likely going to lose against the Farrell measure. But the Count All the Votes measure creates a better compromise — virtually all proponents of IRV have good reason to support it over Plurality with runoffs, and it makes some concessions to the anti-IRV crowd, which we discuss below.
2) Voters are never punished (get a worse result) for supporting their sincere favorite candidate (e.g. get Bush instead of Gore because you voted for Nader). Not even Instant Runoff Voting can say this!
3) A popular political platform never loses just because too many candidates run on that platform and "split" the vote. For instance, if several progressives are running, and you fear that this will lead to the election of a moderate or conservative due to the progressive vote being spread too thin, you can vote for as many of the progressives as you want, to maximize the chance that a progressive wins. Or vice versa if you're conservative or moderate.
This page describes some of the benefits of the Count All the Votes system, and responds to some common questions and criticisms. Here is a page detailing the reasons we believe this system is much better and more democratic than Instant Runoff Voting. Here's a Youtube clip of Clay Shentrup, of The Center for Election Science, speaking in favor of Approval Voting (and the closely related Score Voting) at City Hall.
The Count All the Votes system also makes a concession to the folks who feel that Instant Runoff Voting has been too complex and/or confusing. They get their ordinary single-column ballot back — in all races. And if they want to keep it simple and just vote for one candidate, they can do that. To them it will feel like they're using ordinary Plurality Voting again.
But folks who want to express support for multiple candidates (including long-shot candidates whom they passionately support) may do so. They will never feel like a vote for their sincere favorite is "wasted". They can still cast a strategic vote for the "lesser evil" among the frontrunners, if they choose to.
Further, there will be no multiple rounds of elimination, but instead a simple sum. The results will be easy to understand. And the simplicity of this system makes for better election transparency and integrity. It can be handled on ordinary out-of-the-box voting machines without any upgrades or modifications required to process a ranked ballot. And unlike IRV, it can be subtotaled in precincts, so it's not necesary to transport the ballots to City Hall to centrally count them.
We urge the Board of Supervisors to introduce a measure to compete with the Farrell measure on the November 2012 ballot, to switch to Approval Voting for all citywide and countywide offices, with elections held in November, and without a runoff.