California is voting today. As a supporter of democratic institutions to the maximum extent compatible with broadly republican government — in other words, as someone who is predisposed to like the institutions of ballot initiatives and petitions, and who thinks that recalls do have a place in a well-run representative democracy, I take away two and a half lessons from the California fiasco: The two main lessons are that the bar for a recall needs to be somewhat higher, but not too high, and that if ever there was on object lesson on the virtues of single transferable vote, this is it. (The half lesson is that I need to rethink the virtues and vices of postal voting.)
The first lesson is obvious and has been noted often: it's dangerous to set the bar for recall (and perhaps for petitions) too low. In California, the state currently requires a minimum of 897,158 signatures of registered California voters to recall a Governor, that being 12% of the of the votes cast for Governor in the most recent election. I think this experience shows that this number is too low. Only 36.1 percent of the 21.5 million eligible voters in California voted in the November 2002 election. If the requirement for a recall were 12% of registered voters, that would almost have tripled the number of signatures needed. California's total population is about 35 million so ten percent of that would require 3.5 million registered voters to sign, or about four times as much as currently required.
The second lesson also seems obvious to me, but strangely got little traction. Many people have pointed out the danger that comes (if the incumbent is recalled) of allowing the replacement to be selected by a mere plurality. It's entirely conceivable, for example that 49% of the voters might choose to retain Gov. Davis, but that his replacement might have the support of only, say, 25% of the electorate. That is not a result that serves the state well.
The traditional US approach to this problem—when we deal with it at all—is to require a run-off election between the two top vote getters when no one has a majority. That would be better than what California is doing now, but I'd like to suggest that it is nowhere near the best system.
The state of California was badly served by the conduct and structure of the replacement campaign. The special election brought forth
125 135 candidates. I thought that was great. Yes, some of them were jokes—the bar to getting on the ballot was set too low—but others were real people who ought not to have been dismissed quite so quickly as possible candidates. It would be wonderful if we lived in a world in which you didn't need to be a professional politician and have millions of dollars in order to get the media to pay attention to you when running for office, a world in which you would at least get one chance to make an impact on the quality of your person and your ideas.
The solution to both the problem of the candidate with a tiny plurality and the problem of over-rapid media-driven winnowing of candidates is to have the ballot be by single transferable vote
The Single Transferable Vote is a system of proportional representation that allows voters to vote for individual candidates (as opposed to party lists) in order of preference. Voters rank the candidates in order of preference; first preference votes are the first to be looked at, and the votes are then transferred if necessary from candidates who have either been comfortably elected or who have done so badly that they are eliminated from the election.
STV is obviously much fairer than first-past-the-post, plurality-take-all for elections with multiple candidates and no runoffs. It produces a winner who has more real support, more visibly, and thus more legitimacy—which makes it easier for the victor to govern effectively.
I'd argue that STV is also usually superior to first-past-the-post elections with runoffs.
For example, suppose there are three candidates. A and B each have their passionate supporters representing 40% of the electorate. C has 20% support but is the second choice of a substantial majority of A and B's supporters. In an STV system, C wins. In a first-past the post system with runoff he doesn't even make the final cut. [This is was a really lousy example; see the comments. A better example is this one: A is a candidate with a strong following among a minority of the electorate. He gets 30% of the first choice votes, which is more than any other single candidate, but everyone else hates him. B, C, D split the rest of the vote. As each gets eliminated their votes are redistributed to everyone except A; eventually one of them gets over 50%.] It's true that one can craft other hypothetical preference distributions in which an STV system is less likely to produce a compromise candidate. STV might in some cases reduce the pressure on candidates and parties to moderate their views to appeal the marginal swing voter, but I think European experience shows that this tends not to be the case in single-member districts. (There is something to this criticism when the elections are for candidates in multi-member districts, or for national electoral lists. But we basically don't have elections like that in the US, so for us this is a red herring.)
Many people will spin the results in California as democracy gone wild, or 'oh those nutty Californians'. For me, though, it's mostly a state let down by its electoral architecture.