Eric Maskin and Partha Dasgupta are smart guys, and it’s hard to believe they are totally ignorant of what happens in the Southern Hemisphere. So how can they justify writing a piece promoting a system of “rank-order voting” as superior to the existing American (plurality) and French (top-two runoff) systems, without mentioning that Australia has had this system (in a range of variants) for many decades[1].
A minor side point is that, in addition to having the world’s most complicated voting systems, Australia also has compulsory voting[2]. Typically more than 95 per cent of votes are formal, that is, list all candidates in order of preference, with no missing numbers or repetitions. In Dennis Mueller’s generally excellent book on Public Choice, he discusses the single transferable vote and suggests that, while attractive in theory, it’s too complicated to work in practice. Either Australians are a lot smarter than everybody else, or public choice theorists aren’t as smart as they think they are.
fn1. To be precise, Maskin and Dasgupta advocate the Borda weighted vote, whereas Australia has the single transferable vote (called preferential voting in Australia), but nothing in their argument distingushes the two.
fn2. More precisely, compulsory registration and attendance at the polling station – there’s nothing to stop you casting a blank ballot.
John Quiggin 
“In Dennis Mueller’s generally excellent book on Public Choice, he discusses the single transferable vote and suggests that, while attractive in theory, it’s too complicated to work in practice. Either Australians are a lot smarter than everybody else, or public choice theorists aren’t as smart as they think they are.”
Think it through. You are a cultural product of this system yourself (I’m not). You are likely to have a bias in favour of construing whatever it does as “success”. To me, it looks as though he is perfectly correct: it does not work in practice. This is mainly because it is more focussed on reckoning a result better than no result (in many circumstances I would rather have a deadlock), and on not “wasting” a vote – the idea that there are no issues of principle, deal breakers, and that all compromise is good. Added to that, what you ask for, and what you see, are most definitely what you do not get (most of the time). Clearly, this makes it a failure for me, but a success for anyone who actually wants that.
I have already expressed my preference for term limited cumulative voting, combined with an upper house chosen on a very different basis (without deadlock breaking), to get round Arrow’s theorem problems.
I’m using a very weak criterion of “works in practice” here, namely that the system should elicit formal votes from most electors, which I assume was what Mueller meant when he said it was too difficult.
I’m pretty sure that John’s interpretation of “works” is closer to what Dennis Mueller has in mind (having been at a few seminars with Dennis while he was at the ANU last year). There are of course a number of models of preferential voting, and they produce divergent results – eg in a contest for multiple positions should you first distribute preferences from a top ranked candidate who has a large surplus of votes (well above the number required to win) or distribute from the bottom first? IMO, a preferential system is highly desirable for a single member electorate, but is more open to manipulation and sneaky practices by the contesting parties when electing (as for example in the case of the Australian Senate) a larger number of candidates. That said, we’re not going to change it short of a constitutional revolution! (path dependency and all that)
Careful, John. Footnote 2 could land you in gaol.
AFAIK, it’s legal to mention the possibility without advocating it. It’s also legal, again AFAIK, to advocate repeal of the law used to silence Albert Langer, and I hereby do so.