Sunday, July 11, 2010

Saving the Republic: A Primer

You've no doubt noticed that the views represented in Congress these days seem more extreme than in the past, with moderates disappearing. As a result, it has become harder and harder for parties to work together, and therefore harder and harder to get anything done. I don't think even libertarians could rationally consider this a good thing.

A large part of the problem can be attributed to party primaries, as I observed in my post "Bring Back the Smoke-Filled Room." When party nominees are chosen by primary, they represent the mainstream of opinions in their own parties, which may be far from the mainstream of opinion in the country as a whole. Even though candidates start scuttling toward center as soon as they're past the primary, they still have to act in Congress with one eye on the next primary, lest they  find at re-election time that their disloyal moderation is still resented by the party faithful. (See under McCain, John.)

It is perhaps a sign of how desperate things have become that people have started seriously discussing alternatives to the party primary. It turns out getting rid of party primaries is fairly easy, as the Constitution gives states a lot of freedom to decide how to run their elections-- no amendment required. There are a couple of prominently proposed replacements. 

The Jungle Primary. The simplest alternative is to hold non-partisan primaries, with the top two candidates going on to the general election. This alternative is sometimes referred to as the "jungle primary" (I have no idea why) or the "Louisiana primary." The idea is that now candidates will be able to appeal across the full spectrum of ideologies, rather than having to cater to a narrow slice of it. This alternative was recently urged in an Op-Ed in The New York Times by Phil Kiesling that does a good job of summarizing the "pro" side of the argument. More remarkably, California just passed a ballot measure establishing a non-partisan primary, referred to as "top two," for all elections other than presidential. This, of course, has generated lots of comment pro and con.

Some of the pundits warn of unanticipated consequences-- particularly, that the added expense of appealing to a broad electorate (without a party to pick up the tab) will further empower interest groups and the rich, and force out third parties. But the biggest unknown is: will it work? Will more moderates get elected?

At, Nate Silver has done a simulation of the two different electoral systems, and finds that moderate candidates do much better with the jungle primary. His post not only gives a good summary of the logic, but has very pretty bar graphs showing the difference in outcomes (Bactrian versus dromedary, or to be pedestrian, bimodal versus unimodal). I am not wholly convinced by this approach, however. Silver's simulation generates candidate ideologies randomly, but real candidates are likely to adjust their ideologies depending on which other candidates are running-- for example, running as an extremist when many moderates are running. In technical terms, it's less a statistical problem than a game-theoretic one.

Did someone say game theory? Eric Maskin, a Nobel laureate in economics, points out in a reply to Kiesling that a top-two system is essentially what made Jean-Marie Le Pen, an extremist's extremist, a candidate in the final election in France in 2002, pushing aside a candidate who would easily have beaten him in a head-to-head matchup. Similarly, the Louisiana ancestry of this voting system is not reassuring; this is the system that led to David Duke, former Grand Wizard of the Knights of the Ku Klux Klan, getting more than 40% of the vote for the U.S. Senate. True, neither Le Pen nor Duke was elected, but when one of the two final candidates is a wacko, voters could not be said to have a real choice.

There's not much empirical research on the effects of non-partisan primaries. What there is suggests a real but modest moderating effect. We'll have more evidence soon in California.

Instant Runoff Voting. In Instant Runoff Voting (IRV) voters rank-order some or all of the candidates first, second, third, etc. Then the last-place candidate is eliminated and those voters are reassigned to their second choices. This process continues until someone has more than fifty percent of the votes. So there's no need for a runoff; in effect, the primary and the general election are the same.

Like the top-two (jungle) primary, IRV keeps a minority from being gatekeepers to a general election. Unlike the top-two primary, it keeps an extremist from profiting from a split vote among moderates. To my knowledge, no one has mentioned another benefit, which is that it would give much more information on what voters actually intended when they voted for a particular candidate, because we can see whom else those voters liked.

IRV has gotten support from some surprising places recently. Thomas Friedman endorsed it in an op-ed in The New York Times, and even more remarkably, it was used in the last Academy Awards, where it is credited with allowing "The Hurt Locker" to beat "Avatar." It's in use in Australia (in case anyone in America cares about that), and, under the name "alternative voting," it recently became an issue in coalition negotiations in Britain, where the centrist Liberal Democrats have gotten tired of being everyone's second choice. (The Conservatives agreed to hold a referendum on it next year, which should give it more visibility in the US.)

IRV does seem to me to have significant advantages over the jungle primary. It might be a little tricky to make it easy to use with optical scanning, but the main obstacle is probably getting people used to the idea. A good place for some of that "states as laboratories of democracy" stuff that lovers of federalism are always talking about.

What's Wrong With Partisanship? There is a counter-argument that, rather than being a problem, partisanship is a good thing, because it gives voters a clear choice and accountability; voters should know what they are getting when they vote for a party. The argument is stated by Matthew Yglesias here. Moreover, some argue, other countries, such as those in Europe, seem to get along fine without bipartisanship. (See, for example, comment 1 here, and my reply at comment 3.) Even in those parliaments where coalitions are the norm, nobody worries about reaching across the aisle to parties outside the coalition.

The problem with this argument is that we're not a parliamentary system. In our system, extreme partisanship does not produce accountability. It produces paralysis, which is pretty much the opposite of accountability. But we theoretically could, and perhaps should, tinker with the system to make partisanship much more workable. The obvious place to begin would be to eliminate or drastically curtail the filibuster. The Founders listed quite specific situations where supermajorities are required in the Senate, and never intended to make it a body where nothing can happen without three-fifths support. Other Senate rules are even more inimical to accountability, such as the one that allows a single Senator to keep a bill or appointment from coming to the floor.

The argument that paralysis has more to do with structural rules than with partisanship has even more force in California. The surest way to overcome dysfunction there would be to abolish the two-thirds supermajority requirement for tax increases-- itself, ironically, a product of the referendum process as part of Proposition 13.  It's going to take an awful lot of moderation to overcome that rule.

But at the Federal level, there are Constitutional limits on how far we can go toward workable partisanship. As long as there's the possibility of a House and Senate, or a Congress and Presidency, controlled by different parties, the parties will have to work together to get anything done. So it looks like we may need to take seriously the idea of reforming the primary system.

1 comment:

  1. Top Two Open Primary: We already have the evidence. It's in Washington state. Maskin was right.

    California is full of partisan candidates because California is full of partisan voters.

    IRV isn't much better.

    Have you looked at approval voting or score voting?