Maskin on single-winner systems and the US electoral college
Prof. Eric Maskin came to Georgetown today to advocate for Condorcet systems, which he branded “true majority rule.” In the process, he gave nicely intuitive explanations of that system and the Borda count using the Florida 2000 and France 2002 examples.
Maskin’s main concern was how to solve the “spoiler problem” in plurality elections whereby a majority-opposed candidate wins because the majority splits its support. I’d take issue with two parts of the presentation: his treatments of instant runoff voting and the prospects for reforming the American electoral college.
He was not openly hostile to IRV, which I appreciate. He readily noted that system would, like Condorcet, be preferable to plurality because it lets voters register more than one preference using rankings. One might say Maskin’s preferences are: his favorite system > any reform > status quo. The same cannot be said of other Condorcet (and approval, and range) advocates whose preferences appear to be: favorite system > status quo > any reform.
But he was somewhat dismissive, lumping instant runoff with two-round runoff systems (TRS). In the 2002 French presidential election, Jean-Marie Le Pen entered the runoff because of first-round spoiler problems on the left, and Lionel Jospin, who may have been the majority-preferred candidate, ultimately lost to Chirac. This example supposedly highlights the deficiency of “runoff” systems, including instant runoff.
This is something of a straw man. Maskin’s main concern - he returned to the point when pressed by an audience member - is that runoff systems are eliminative, and the order of candidates’ elimination can produce ‘wrong winner’ outcomes. While that deficiency is possible under IRV, it is far less likely than under the French TRS. In the former, candidates are eliminated one at a time. In the latter, all but the top two are. In fact, IRV likely would have ‘worked’ in France 2002.
To adjudicate the relative merits of Borda and Condorcet (and, implicitly, IRV), Maskin applied five criteria from voting theory: consensus, anonymity, neutrality, independence of irrelevant alternatives (IIA) and decisiveness. Here we enter the realm of value judgements. Borda and IRV do not satisfy IIA, but Condorcet may not satisfy decisiveness. That is, it may fail to elect a winner when no one candidate is preferred by a majority over every other candidate in pairwise contests (called a “Condorcet cycle”). For Maskin, this deficiency is less offensive than the failure of IIA.
Moreover, if we assume with Maskin that most voters have ideologically driven preferences, a vote distribution resulting in the above is unlikely. (Say, Bush > Nader > Gore > Buchanan.) And when it occurs, we should not get too upset because such preferences are ideologically inconsistent. The voter has been illogical.
As a thought experiment, I’ll buy the claim that ideology, however understood, governs preferences. Too much criticism of reform efforts has been predicated on hypothetical preference orderings that seem schizophrenic.
Maskin also made some comments about the electoral college. When I asked him for thoughts on the National Popular Vote plan, he said it was a “cute idea” and that my question was “not on point.” But one of his arguments for using state-by-state Condorcet in presidential elections was that moving to direct election is variously “not possible” and “not likely to happen.” As such, the question is “on point” because he raised the issue.
Replacing the electoral college with a popular vote, as Maskin noted, would not address his core concern about plurality rules. His objective must have been rhetorical, therefore - to seize on reform energy to promote a reform unrelated to the desire for a direct election.
In all fairness, Maskin paid heed to the electoral college’s malapportionment and sidelining of voters in non-swing states, noting there may well be reasons to reform the institution.
But I vigorously dispute the claims that reform is “not possible” and “not likely.” One, states have constitutional authority to choose their electors in any way they want. Two, there was no single, sincere justification for the institution when the convention adjourned in 1787. (Hence states reserve the right to do what they want.) Three, National Popular Vote may have been a “cute idea” two years ago, but two states have since signed on, and it’s expected a third soon will. If that happens, we will be “46 electors closer to more democratic presidential elections.”
AllAboutVoting on 17 Jan 2008 at 7:37 pm #
>The same cannot be said of other Condorcet (and
>approval, and range) advocates whose preferences
>appear to be:
> favorite system > status quo > any reform.
For what it’s worth, my view is:
Approval > Range (w/ good instructions) > most Condorcet > | status quo > IRV > truncated IRV
I would support any of the reforms above the ‘|’.
In some ways I prefer Range to Approval but I think that Approval is a better reform to push for first.
Where:
* range w/ good instructions = instructions clearly state that votes that are not max or min scores can be thought of as partial abstentions.
AllAboutVoting on 17 Jan 2008 at 7:44 pm #
>When I asked him for thoughts on the National
>Popular Vote plan, he said it was a “cute ideaâ€
>and that my question was “not on point.
I like NPV but have some reservations about the NPV compact plan. As someone interested in alternate single winner election methods my biggest concern is that I have not heard a clear explanation of where the data for what the ‘national popular vote’ was is collected.
It seems to me that if a state that is not in the compact uses an alternate election method they will not be able to provide NPV data for use by the compact states. How is this coped with by the plan? It seems cruel to punish states for trying other methods by disregarding their popular vote.
It seems complex and distorting to intuit what their popular vote would have been from ballots that collected votes for use in a different method.