The new electoral system in California is a top-two majority runoff with the possibility of multiple candidates from one party. Please, do not call it a primary, because it isn’t. In a primary, a political party permits voters to select its candidate for the general election. However, under the new California system, the general election will now be just a runoff between the top two candidates, regardless of party. That is, at most, two parties will be represented on the general-election ballot, but it is possible for both candidates to be from the same party, or no party (if both of the top two in the first round were non-partisan).
We might call it two-round SNTV, for lack of a better term. The reference to SNTV–single non-transferable vote–calls attention to the fact that two or more candidates of the same party can be competing with each other, but co-partisans are unable to share votes with one another to ensure that they don’t divide the vote and cause none of them to advance. (As noted, the second round can also feature two candidates of one party, but then there is no risk of coordination failure, as the winner will be from that party, obviously.)
In the first use of this system this week, there are a few cases that could represent SNTV-style coordination failure. There will be several legislative races in which the November choice will come down to two candidates of the same party. Most of these are in districts with an entrenched incumbent who will happen to face a (token) intra-party challenger, so there is no coordination problem. There just is no opportunity for voters in November–who will be more numerous than they were in the first round–to register a partisan choice for one of the other parties. I will focus my attention, then, on a few cases in which the runoff contenders are from one party, and they did not combine for significantly more than half the first round votes. (This is not an exhaustive list.)
A particularly striking example occurred in US House District 8: The runoff will feature two Republicans, Paul Cook (15.5%) and Gregg Imus (15.0%). The third place candidate just missed qualifying for the runoff: Democrat Jackie Conway (14.7%). There were 11 other candidates, including a second Democrat who had 9.7%. While the combined votes of ten Republicans is over 70% and thus this was not a district a Democrat was likely to win, the Democratic Party nonetheless narrowly lost the right to even make their case to the general-election electorate.
In US House district 31, the top two candidates are both Republicans: Gary G. Miller (26.7%) and Bob Dutton (24.9%). There were four other candidates, all Democrats, and the top-scoring one, Pete Aguilar, had 22.8%, missing the runoff by just over 2 percentage points. While the two Republicans combined for a majority of the votes, they did so just barely, with 51.6%. It is not out of the question that a Democrat could have won this district–especially given the difference in turnout that we can expect, as well as the long gap between elections and the potential importance of candidate quality. But the Democrats will not get to make their case in this potentially winnable district.
In fact, this last example points to another potential pitfall of the system: even if some candidate wins a majority in the first round, there still must be a runoff. What will be really interesting is the first case in which the majority “winner” in round 1 loses round 2 due to the different turnout or other reasons. Something to watch for.
Naturally, if this is “part 2” there was also a part 1, complete with a pretty picture!
Fruits and Votes 
Isn’t this the same as the old Louisiana system that Edwin Edwards instituted? As I remember it, he was mad that he had to fight three elections (primary, runoff, general) and his Republican opponents (who were usually unopposed for the nomination) only had to fight one. Of course, Edwards’s brainchild nearly got David Duke elected.
LikeLike
One method that might maximise the pitfalls of both “highest from each party” and “top two overall” would give a second-round spot to each candidate who, on the first round, either:
(a) was one of the two highest, or
(b) polled at least (say) 15% of the votes in his or her own right as an individual, or
(c) was the highest-polling candidate of a party whose candidates collectively polled at least (say) 20% of the votes.
So you’d have a “party-list” aspect (see my earlier comment, on a different thread I can’t now locate, saying that US party-line primaries are to French runoffs what open-list PR is to SNTV) which would mean that the top Democrat would make it to the second ballot under criterion (c) (even if a crowded field meant he or she didn’t satisfy (a) or (b) also).
It would be slightly less party-based than list PR since, eg, Paul supporters whose votes for Paul on the first ballot help swell the “Republican” total and thus get Romney into the second ballot, under criterion (c), are still free not to vote for Romney on the second ballot.
A couple of issues to clarify:
i. You might still need a third ballot if no candidate on the second reaches a set threshold of votes (50%, 40%, the double-complement threshold, whatever).
ii. You would need some way to determine which candidates are “Democrats”, which are “Republicans”, etc – probably midway between “identifies with” self-declaration at one extreme and party-committee approval at the other. “Any registered Democrat” might work, say, or linking a party’s public funding and/or future ballot slot and position to its first-round votes (so the Laroucheans might baulk at entering the “Democrat primary” and thus help Howard Dean and Nancy Pelosi).
LikeLike
er, “minimise the pitfalls” or “maximise the advantages”.
(Strange indeed how much easier to spot typos in Times or Cambria than in Courier New…)
LikeLike
iii. Might also be a case for letting the plurality candidate on the first round win outright if he or she is supported by, say, 60% of the valid votes and 40% of all adult citizens. On the second ballot, this could drop to 50% of votes and 30% of the eligible voters. On the third ballot, a plurality should suffice, on the basis that the voters have had adequate opportunity already (if not as ample an opportunity as STV or exhaustive balloting would offer) both to express their sincere first preference AND to keep out the candidate they detest the most.
LikeLike
What are the main differences between this system and the system used in French elections? The only one I can see right away is that the French allow third place finishers with a certain amount of support into the second round.
And the coordination problem mentioned actually happened in France, in a big way, in 2002.
If you are going to use single member majority (SMM), this is a big argument for doing it Australian style instead of French style, bigger than the cost of the second election that I’ve usually seen cited.
The one big advantage I’ve seen of the French practice of using run-offs, that it gives the chance for minor parties to convert their votes into something tangible by bargaining to throw their support to the major parties, of course isn’t present in the American party system.
LikeLike
@Tom
Let us try and avoid complications that do not make it transparent to an elector where there vote is going.
@Ed
Minor parties do regularly negotiate with the major parties for policy concessions in return for second preferences. The difference between what happens in Australia and what happens in France is that minor parties in Australia can only recommend to their electors. The decision remains with the electors where it should be, subject of course to issues around above-the-line voting.
LikeLike
@Alan, I think the idea that “if all the Republicans together poll over 20%, then the highest Republican gets a slot on the second-round ballot” is comprehensible to the average voter. Many US States already require a party’s primary turnout to reach a certain threshold (usually 1% or thereabouts) to give it a ballot line in the general.
Didn’t Earl Warren used to run in, and win, both the GOP and Dem primaries for California Governor?
The problem with a highest-each-party system is that it enables “cracking”. Warren has the support of 40% of the voters, against two rivals with only 30% each. Yet Warren gets trounced 20% to 30% (of the total turnout) by his rivals, in turn, in either given party’s primary.
LikeLike
@Tom
Why not 15% or 25%? Why not not 19% or 21%? With majority preferential voting you do not need the legislature to start fixing quotas because the quota is an inherent part of the system. Arbitrary quotas are different.
LikeLike
Alan, anything I propose that isn’t AV-STV is predicated on the assumption that AV-STV is off the table for a particular jurisdiction and some second-best option must be cobbled together. Illogical, I know, but from observing the debate over the UK AV referendum, there seem to be a lot of people out there who don’t like AV because “it’s too complicated/ it gives some voters more votes than others”, yet who are quite happy with using exhaustive ballot to elect the Tory Party leader or the Commons Speaker, or some variety of runoff to elect the Mayor of London or the President of France. Exhaustive bottom-up runoffs are not a live option for direct popular elections, but simple top-two also has problems, as shown two French presidential elections ago and California today. 15% is a fairly common threshold in US elections.
LikeLike
@Tom
I know. It is just weird seeing these cobbled-together systems based, as you say, on myth. The current Egyptian presidential election gives us another example of the excellencies of the second ballot as a way to give the electorate choices it does not want.
LikeLike
SNTV is a good analogy as top-two presents the same double-or-nothing gamble. Focus all your supporters’ votes on one single.nominee and you’ll probably assure her a place (assuming you have enough supporters). Try to divide your votes evenly among 2 or 3, and you might doubleor triple your harvest of places secured… or you might fall between two stoolsand end up winning zilch.
I wonder how/ if the dynamics of SNTV vary acc’g to factors other than the vote-counting rule. Eg,whether parties distribute how-to-vote leaflets, and whether Candidates ballot positions are rotated, orvaried in different precincts (as in New york andPuerto Rico). This would make the party vote-managers’ job easier.
LikeLike
[comment by Ed replanted. Thanks, Ed, for the interesting link and question!]
LikeLike
More on topic, one approach to holding a runoff could be as follows: the number of candidates that advance to the second round is equal to the total first round percentage obtained by all except for the top candidate, divided by the first round percentage obtained by the top candidate, rounded up.
In plainer English, that would mean that:
1. If the top candidate gets 50% +, then one candidate advances to the second round (eg the top candidate is elected).
2. If the top candidate gets between 33.3% and 50%, the top two candidates advance to the second round.
3. If the top candidate gets between 25% and 33.3%, then the top three candidates advance to the second round.
4. If the top candidate gets between 20% and 25%, then the top four candidates advance to the second round.
And so on.
There is only one second round and whoever wins the most votes in that round is elected, regardless of whether they have a majority, so this would not me a pure single member majority system. It would be more like a single member enhanced plurality, or single member majority-encouraged system.
Plus you could still have a situation where the votes of a substantial partisan bloc are scattered around several candidates, so that the candidates advancing to the second round all come from a rival bloc that doesn’t have substantially more support. The difference is that it would be harder to game this, and almost impossible if the bloc doing this doesn’t have majority support in the district in the first place. You could have a situation where a Bloc A has 45% support and Bloc B has 55% support, but candidate A1 gets 35%, candidate A2 gets 10%, and candidates B1, B2, B3, B4, B5, and B6 each get 9%. It would take such an extreme gap in the coordination of the A candidates vs the B candidates to get this result, because the top two candidates couldn’t sneak into the second round and monopolize it with voting percentages in the 20s.
LikeLike
@Ed: Applied to the French 2002 presidential election mentioned by other commentators
Chirac: 19.88% in the first round
Other candidates: 80.12%
80.12%/19.88%=4
Thus Chirac, Le Pen, Jospin and Bayrou would have gone into the second round. I can’t imagine it would have changed the final result, though Jospin could have slipped through on a sub-30% plurality if enough leftist voters had taken fright and Bayrou had mounted a credible challenge to Chirac.
LikeLike
In part 1 of this discussion, Ed summarized the intentions of the proponents fairly well. Business interests are trying to weaken both major parties (without actually replacing them with a multi-party system), in order to put their money behind candidates who they don’t think will get nominated in partisan primaries. (The code word for such candidates is “moderates”, which really means pro big business.) They believe, with some justification, that they have lost control over both major party organizations.
The net effect of top two — if it works on its own terms — is to make money more effective outside of the major party organizations. The evidence from Louisiana, the State of Washington, and now California is that it doesn’t work all that well on its own terms.
If Top Two continues to have weak or non-existent benefits for its backers, it will be interesting to see what they come up with next. A new centrist political party? I wouldn’t rule that out. Americans Elect provided a rationale for its nominating procedure that is identical to the rationale for Top Two. According to both, the voters, not the parties, should nominate [sic] candidates to run in elections. Like the folks who funded the campaign to pass Top Two in California, the folks who funded Americans Elect are business and financial leaders who don’t feel they can gain enough influence in the major parties.
LikeLike
I totally love Ed’s idea on a runoff system! Totally awesome!
LikeLike
Can I suggest a series of quotas dividing by successively larger numbers so that where V is the total vote the descending quotas would be V/2, V/3, V/4 etc etc
I think thats what you’ve done by using fixed numbers, but using the calculations would make the rationale of the system clear.
LikeLike
Alan @10: “he current Egyptian presidential election gives us another example of the excellencies of the second ballot as a way to give the electorate choices it does not want”
– Ahdaf Soueif, “‘Down with the next Egyptian president’: Whoever is declared the next president of Egypt will not be the person most Egyptians want.” guardian.co.uk (17 June 2012) http://tinyurl.com/c4dxh7p.
LikeLike
“Or how about California’s 51st U.S. House District in San Diego? A strongly Democrat district, the lead candidate, a Democratic state senator, spent nearly $50,000 in support of a penniless Republican opponent to prevent his strongest rival, a fellow Democrat, from making the November election. The ruse worked, and now the Democrat will soundly trounce his Republican opponent in the runoff.”
–Steven Hill, writing in the Sacramento Bee
LikeLike
It would be interesting instead to try the traditional* French runoff system: No forced elimination at all after the first round, election by plurality in the second.
However, runoffs in California should not include any new candidates (except possibly write-ins and emergency replacements). Those wishing to advance should also explicitly confirm this. Or perhaps the party for which the primary candidates stood ought to perform this task? That should eliminate coordination failure, but would on the other hand formally make the primary a non-binding test vote for the party (instead of a test vote for the individuals).
Regardless, as commonly seen in historical cases, one would expect the Duverger effect to be minimised at the first vote and maximised at the second. It would be a constitutional wide-open primary system that allowed small parties to participate fully, and that prevented coordination failure unless there was a breakdown in party discipline.
* It was in use during most of the Third Republic (also, in Norway 1906-18 and at presidential elections in Weimar Germany).
LikeLike
In the Netherlands, a similar system to the new Californian one was used from 1848 to 1918. It was manipulated in a rather different way: it was possible to stand as candidate in more than one constituency, and quite a few leading politicians took advantage of that rule. After the general election, they would have to choose a constituency to represent, while in the other one(s) by-elections would follow. Unsurprisingly, the double candidates always chose to represent the constituency where their party would have faced a tougher by-election battle, thus leaving the other one(s), which were safer, to another candidate of the same party or faction.
LikeLike
JD @21, that type of manipulation isn’t specific to, or inherent in, SMD runoffs. Multiple candidacies are allowed in India and Bangladesh (FPTP) and Malta and formerly Ireland (STV). I believe The Netherlands still allows multi-district candidates with PR-List – a candidate can stand on different versions of the same party’s list in different districts, or a party can run the same version of its list in different districts. If elected in two or more districts, I think the candidate is credited to the one where s/he polled the most personal votes, and deleted from the list in the others. In the other cases I mentioned, the candidate gets to choose which seat to keep.
I stand by my earlier proposal of “the top two, plus any other candidate over 20%, plus the top candidate of any party over 20%.” Perhaps the runoff ballot should be by Approval so a majority can be certain of keeping out candidates they detest (while still ensuring that a candidate with majority first-preference support would be elected on the first ballot).
LikeLike
Tom Round: You’re right, that manipulation actually has little to do with the two-round system. Thanks for that – I didn’t know about India and Bangladesh. Do they have many by-elections soon after general ones?
In the Netherlands, there are actually no districts in the true sense of the word… There are only ‘districts’ where parties can present different lists – this influences the distribution of seats to candidates (but not lists) in a very complicated way. However, the recent tendency is for each party to present one list, something many large parties have already been doing for quite a while.
LikeLike