Instant Runoff Voting: A Means to Destroying the Two Party System?
America’s electoral system is poorly designed and in desperate need of reform. It encourages toxic, negative partisanship, where people are apt to vote out of fear of their least favorite candidate getting elected. Additionally, the status quo entrenches a two-party duopoly that, given low primary turnout rates, is lacking in democratic legitimacy.
Luckily, people are starting to pay more attention to electoral reform. Unfortunately, the reform option that has gotten the most traction is Instant Runoff Voting, a subset of electoral systems known as Ranked Choice Voting. The logic behind it is seductively simple. You rank the candidates on your ballot, if your preferred candidate is eliminated as the person with the least amount of votes, your vote gets transferred to your next favorite candidate, and so on until someone wins a majority of the votes. While it would represent a marginal improvement over the current way of doing things, IRV is a seriously flawed electoral system. Resultantly, we should look elsewhere to one of the many other choices for electoral reform.
The primary reason why IRV is flawed is that it violates what’s known in social choice theory as the monotonicity criterion. Essentially, under IRV it is sometimes possible to make a candidate lose by ranking them higher up on your ballot or make someone win by ranking them lower.
A real-world example of this occurred in the 2009 Burlington mayoral election, which was contested between three parties: the left-wing Vermont Progressives, the centrist Democrats, and the conservative Republicans. In the second round of this election after a minor candidate was eliminated, the Republican candidate got 37% of the active vote, the Progressive got 34%, and the Democrat got 29%. The Democrat was then eliminated, and because 63% of Democratic voters switched to the Progressive candidate, the Progressive won the election with 51% of the third round vote. However, if just a third of Republican voters (~13% of all voters) in the second round had voted for the Progressive instead, the Republican candidate would’ve been eliminated, but the Progressive would have ultimately lost the election to the Democrat. (I calculated from data here that 84.6% of GOP voters preferred the Democratic candidate to the Progressive.) So, Republican voters actually helped their least favorite candidate, the Progressive, win by ranking him last instead of first. To illustrate this more clearly I’ve created a couple of Excel tables.
Crucially, simulations suggest that in competitive three-way instant runoff races, there is somewhere between a 15% and 51% chance of the monotonicity criterion being violated. As elections become more competitive, the likelihood of monotonicity failure increases. Since competitiveness is measured by the level of support for the third-place candidate, monotonicity failure doesn’t really affect elections with only two competitive candidates.
If we did implement IRV, we would face two horns of a dilemma. Either it would be successful at tearing down the two party system, or it wouldn’t.
If America did become a multiparty democracy, then, with IRV, its elections would effectively turn into random winner generators, which is obviously undesirable. It simply makes no sense to have an electoral system based on ranking people in order of preference if you will often accidentally help candidates by ranking them lower on your ballot.
Meanwhile, if America remained a two-party duopoly, then, well, it would remain a two-party duopoly. American voters would be no less inclined to voting out of fear of the other candidate. So, both of the issues mentioned in the first paragraph would remain serious problems for American democracy.
Furthermore, we shouldn’t underestimate the possibility of America retaining its two party system under IRV. Australia is one of a small number of countries that use IRV for electing its legislators, and they have used it for over a hundred years. In the 2019 election for the House of Representatives, 96% of the seats were won by one of the two major parties there. In fact, the Great Depression was the only time period where the two-party seat share fell under 90% of the total seats. It has never fallen below 80%.
Now, if the second horn of the dilemma comes to fruition, this would be better than the status quo. Fringe candidates would no longer play spoiler to major party candidates; Al Gore probably would’ve beaten George Bush in Florida.
Due to their issues in 2009, the city of Burlington voted to get rid of IRV in 2010. America would do best not to waste its time further with this lousy system, and start exploring the many better options out there for breaking up its two-party duopoly.
