Archive | January, 2014

Why we should adopt multi-round preferential voting

18 Jan

It’s very commonly suggested that the American political process is unfairly captured by the two major parties and that we should rectify this by adopting instant-runoff voting, a.k.a. preferential voting. In such voting schemes, instead of simply choosing a single most preferred candidate, a voter may rank all candidates in order of preference, with the idea that this makes third-party candidates more viable. For example, if I most prefer the Libertarian party candidate but also prefer the Republican candidate over the Democratic candidate, then I can express these preferences without concern that my vote won’t advance the Republican over the Democrat.

The important questions about an instant-runoff voting system are:

  1. Would it actually make third parties viable?
  2. Would that be a good thing?

My answer on both counts is, ‘probably not’:

Assume for a moment that we can accurately distill voter preferences into a single dimension of ideology, such that a given electorate has a collective preference for a candidate on a particular spot along the left-right, liberal-conservative axis. What we tend to get in our elections, such as our US House and Senate races, is a dynamic wherein the Democratic and Republican candidates occupy the middle of both sides while 3rd party candidates flank them on the ends. Though many members of the electorate identify as centrists, in truth they mostly sit ideologically clustered into two major peaks not much further towards the center than the Democratic and Republican candidates.

So assuming that the ideological positions of the two major parties, the third parties, and the electorate remain the same after the introduction of preferential voting, what then would be the point? Democratic and Republican candidates already best represent the electorate’s ideological preferences, so third party candidates would never win without shifting to the same positions, and if third parties can only win by becoming just like the two major parties, why do we need third parties? Sure, unpopularity isn’t the only reason that third parties can’t compete in our current system, but it’s likely the primary reason: given their current policies and ideology, the positions of the Green Party, the Libertarian Party, the Constitutional Party, et. al. simply aren’t popular, and preferential voting won’t change that.

Which is not to say that American democracy doesn’t have a systemic gap between what the electorate wants and what it actually gets. I just think that proponents of third parties have the wrong diagnosis and hence have the wrong prescription.

I actually agree that we want a preferential voting system, but one modified in a few crucial ways. The preferential voting system we need follows a simple strategy, the same strategy which people commonly use to make optimal choices in other contexts. This strategy starts by, from the set of all possibilities, identifying a subset of the most favored choices, then in successive rounds, whittling that subset into smaller subsets until left with a single choice. So for example, in a 3-round system, the voters might collectively select the top 30 candidates in round 1, then, from that 30, select the top 6 in round 2, and then, from that 6, select the final winner in round 3. The last round would use Instant Runoff Voting, and all previous rounds would use Single Transferable Vote, a preferential voting system that selects multiple winners.

Now, in a sense, our current system is already a multi-round system because we hold primaries to select party nominees. The scheme I have in mind, however, gives no official role for political parties. If a party wants to select just one member to appear on the first-round ballot, that’s the party’s business; its members would all still be free to appear on the ballot under their party affiliation. So even if, say, the Democratic Party selects one nominee, any other candidate is free to appear on the first-round ballot as a Democrat. If a party wishes it to be known which single candidate it actually endorses, that’s the party’s responsibility. Any candidate could still appear on the first-round ballot claiming to affiliate with any party.

So consider, again, how our current elections play out:

The electorate in the US is clearly ideologically clustered around two peaks, such that only the Democrat and Republican nominees have real chances of winning in the vast majority of our elections. The third parties simply can’t compete because they occupy the wrong ideological territory. Because most voters have a partisan Democratic or Republican affiliation, they end up voting for their party nominee, even if they don’t like the nominee very much.

A non-partisan, multi-round preferential voting system, however, should work quite differently. Assuming again a 3-round system, the first round would whittle down the space of all possibilities down to a more manageable number, say 25 or 30, and then the second round would whittle the field down further to say, 5 or 6. Because each round selects the most preferred candidates of the entire electorate rather than just one party, the final ballot will likely then consist of several very competitive candidates, even in districts usually dominated by one party. In a heavily Republican congressional district, for example, a final ballot of 5 candidates might include 4 Republicans and 1 Democrat. Thanks to instant-runoff voting, the Republican voters would rank their preferences without concern of throwing the election to the Democrat.

What, though, would be the point? In all likelihood, this electoral process would produce Republicans and Democrats who occupy more or less the same positions on the ideological spectrum as they do today. If Republican districts still elect Republicans, Democratic districts still elect Democrats, and competitive districts still swing back and forth, what then would be the advantage?

Well, consider three problem cases in our current voting system:

  • First, in theory, an incumbent with an unpopular position should get voted out, but partisan voters are very reluctant to contest their own incumbents, mainly because primary challenges significantly increase the odds of handing the seat to the opposition.
  • Second, underfunded candidates with low name recognition are greatly disadvantaged. Even when such candidates best matches voter preferences, they tend to get little attention from funders and voters because they lack legitimacy, a sense that they could actually win.
  • Lastly, consider a candidate who deviates from their party on one or two major issues. Even if such a candidate most closely mirrors the electorate, they generally get filtered out by the party primary, where candidates must appeal first to the party base.

In all three of these cases, our current voting system often filters out candidates who best match voter preferences. Multi-round preferential voting shouldn’t suffer the same problems:

  • With preferential voting, voters need not worry about throwing their votes away and so can express their true preferences. Thus, incumbents who deviate from the electoral will on one issue would be more likely to get replaced by a similar candidate without that deviation.
  • Lesser-known candidates would have better odds of winning because making it through the first round of voting should itself lend them some credibility.
  • Because candidates needn’t pass through a party primary, they would be able to mix-and-match ideological positions on different issues.

So, that’s the gist of why I favor multi-round preferential voting. Of course, a new voting process isn’t a panacea for fixing our politics. Our deeper problems stem from the dozens of veto points in our legislative process that severely mute the consequences of our elections. I have some ideas on how to reform our legislative process, but I’ll save them for a later post.