I am writing a paper at the moment on some of my work with Steve Checkoway and Hovav Shacham on voting, which has involved a pretty broad literature search in social choice theory. I came across this quote about approval voting (AV) as an alternative to plurality voting (PV) in the paper Going from theory to practice: the mixed success of approval voting by Steven J. Brams and Peter C. Fishburn (Soc Choice Welfare 25:457–474 (2005)):
The confrontation between theory and practice offers some interesting lessons on “selling” new ideas. The rhetoric of AV supporters has been opposed not only by those supporting extant systems like plurality voting (PV)—including incumbents elected under PV—but also by those with competing ideas, particularly proponents of other voting systems like the Borda count and the Hare system of single transferable vote.
We conclude that academics probably are not the best sales people for two reasons: (1) they lack the skills and resources, including time, to market their ideas, even when they are practicable; and (2) they squabble among themselves. Because few if any ideas in the social sciences are certifiably “right” under all circumstances, squabbles may well be grounded in serious intellectual differences. Sometimes, however, they are not.
I don’t think it’s particular to the social sciences…
On another note, the IEEE adopted AV at some point but then abandoned it. According to a report on the (very partisan) range voting website, there are shady reasons.
An Ergodic Walk 
I am an author of some of the content on ScoreVoting.net (aka RangeVoting.org). I would like to know why you consider the site to be “partisan”.
The site was primarily authored by a Princeton math Ph.D. named Warren D. Smith, who has been studying election methods since at least the late 1990’s. His goal, and my goal, has been to study and report on election methods and other facets of group decision making, from an objective scientific point of view.
It is true that the site is strongly in favor of Score Voting, and Approval Voting. But that position is rooted in voluminous evidence that these systems are objectively simpler and more democratic than the other commonly proposed systems. So to call that “bias” is like going to a reputable site on biology and calling it “biased in favor of evolution”.
One other point is that Brams’s argument actually is somewhat particular to election theory. Election theory is phenomenally counter-intuitive, and much more complex than the vast majority of people realize. Most of the “debate” about which system is best is a result of logical fallacies and lack of knowledge about Bayesian Regret, and other related concepts.
For instance, Score Voting and Approval Voting are commonly criticized for having the potential not to elect a candidate who is the first preference of a majority of voters. This intuitively seems like a valid argument. However, it is mathematically proven that the “favorite candidate of the majority” may not be the “favorite candidate of the electorate”.
http://www.electology.org/majority-criterion
Similar arguments abound in the subject of social choice. In my experience, it is very difficult to get people to abandon their intuition in favor of science and axiomatic logic. This is why the election theory community has not unanimously backed Score Voting and/or Approval Voting.
I think the empirical evidence is not quite so voluminous. I say it’s partisan because the site is an advocate for a particular voting system (more or less). IRV has its partisans too, who will say how IRV is far superior to other systems based on their own metrics. Is there a mathematical definition of “more democratic?”
The analogy to evolution is pretty flawed I think. Firstly, evolution is a scientific theory and a consensus has developed that it is the best explanation for a number of observed phenomena. Voting systems are at best engineered objects, and flawed ones at that. Certainly there is not a widely accepted consensus on the superiority of range voting. Donald Saari is also a math PhD and noted researcher and he prefers Borda Count in a lot of instances.
For the record, I think plurality voting and IRV are pretty bad and that range voting is a good idea. But I’m certainly not going to claim that it has been “proved” to be better.
Yes, there IS a mathematical definition of “more democratic”! The definition is, “having a higher social utility” (or “lower Bayesian Regret”). This is axiomatically proven here:
http://ScoreVoting.net/UtilFoundns.html
That does not mean that we’ve “proven” Score Voting to be the best voting method. It means that we’ve proven that Bayesian Regret is “the one right metric” to assess the democratic-ness of a voting method. However, we use simulations to measure the Bayesian Regret of voting methods, and those simulations are not perfect. But they seem to be extremely reliable, because they found Score Voting to be the best “non-exotic” voting systems in all of 720 different electoral models. Score Voting was best with 100% strategic voters, and with 100% sincere voters, and best with all the incremental ratios in between those two extremes. It was best with different numbers of candidates, and with different numbers of voters, and even with different “ignorance factors”. It was best even with different utility generator models (e.g. Gaussian issue axis, random utilities, utilities based on real world ballot data, etc.).
The detractors of Score Voting do not even attempt to refute these calculations, or offer alternative ones. They instead focus on alternative metrics of quality, which can be proven to be the wrong metrics (indeed, most of them are actually self-contradictory).
You talk of there not being a “scientific consensus” that Score Voting is best. But science is not democratic. If 90% of the world’s scientists say creationism is a better theory than evolution, that does not make it true. Creationism is not falsifiable, whereas evolution is. Creationism does not predict hierarchical taxonomy of life, whereas natural selection does. The same goes for vestigial features (e.g. the broken ascorbic acid gene in apes and old world monkeys). It’s not merely subjective. There are arguments which can be objectively shown as fallacious, no matter how many people are in one camp or another. Consensus is just a pragmatic tool which is generally more efficient than gaining one’s own expert understanding of a subject. We can objectively show that the vast majority of criticisms of Score Voting are flawed, regardless of any consensus.
This is precisely the situation with the arguments of Don Saari. They are simply riddled with logical and mathematical fallacies. Moreover, he does not cite any Bayesian Regret figures to back up his claims. Our calculations actually find that Borda voting is remarkably susceptible to strategic voting. Saari doesn’t refute that; he merely ignores it. We have pointed out his errors to him in detail, and he has simply not responded.
The problem with the metrics advocated by IRV advocates is that they also are virtually all refutable via reduction ad absurdum. That is, they literally contradict themselves or can be demonstrated to be faulty in highly objective ways.
For instance, IRV advocates will criticize Score Voting and Approval Voting for having the ability to elect a candidate who is the favorite of ZERO voters, unlike IRV. As bad as that might sound, it can be mathematically proven that the “favorite candidate of the electorate” can indeed be a candidate who is no one’s favorite. (Nevermind the sheer improbability of this in the first place.) Furthermore, IRV can elect a candidate who is the favorite of only two voters — an insubstantial difference that reveals how absolutely ludicrous that criticism is.
IRV advocates will also regularly tout the Later-no-harm criterion, which they claim is good because it ensures that a vote for your second place candidate can not help that candidate to defeat your favorite. Again, that intuitively seems like a good property. But the problem is that if you sincerely prefer e.g. Green over Democrat over Republican, you don’t want to rank Green in first place to begin with. Your strategic best bet is to rank Democrat in first place, to ensure that the Green won’t be a spoiler. This means that if there are enough strategic voters, then the mere appearance of being unelectable is enough to become a self-fulfilling prophecy. Whereas with Approval Voting, a strategic vote for the Democrat would in no way dis-incentivize a voter’s sincere vote for Green. Then if enough voters truly support the Green, he can win regardless of any assumption of inviability. Here is a very amateur Youtube video I created to try to describe this with a visual aid.
Science means avoiding dogma, and always being open to a change of stance on the basis of new information. However, we have spent many years looking into this subject with extreme rigor. And we have unfortunately found that the “controversy” is nothing but a host of widespread misunderstandings and logical fallacies. This is not just a claim. I can prove it on a point-by-point basis.
Thanks for the long comment. However, I have to disagree. What you’ve done is to say “if I model social utility in a particular way then the natural voting system is range voting.” This is a particular modeling assumption, as “natural” or “unnatural” as demanding, as in Arrow’s Theorem, that preferences be non-cyclic. If my preferences are truly cyclic, then it seems range voting (or Borda, or pretty much anything in use now) would be a poor choice.
Hence your claim that regret is the “one true metric” is rhetorically equivalent to claiming that a particular book is the “one true holy book.” It’s all about modeling. And so whatever you can prove mathematically rests on these modeling assumptions. So I stand by my earlier claim that your site is rather partisan, because your concept of “truth” is hardly dispassionate.