Voting and the Wisdom of Crowds:
Several commenters on my post on the rationality of voting brought up James Surowiecki's interesting book, The Wisdom of Crowds. It so happens that I wrote a review of the Wisdom of Crowds last year, where I explained why his argument that large numbers of individually ignorant decision-makers can make seeemingly well-informed collective choices does not work well in the context of voting. For those interested, the review is available here.
And both markets and democracies have properties not mentioned by Suriowecki that enable them to match the performance of groups of experts remarkably well. For example, both disproportionately reward first-past-the-post winners. This property greatly increases the influence of the well-informed: assuming that the ill-informed are easily persuaded (by irrelevant appeals of one kind or another) to make essentially random choices, the well-informed, who are amenable to rational persuasion, become exceptionally valuable to competing products or candidates, because their choices are dependable and predictable. Similarly, the bandwagon effects that undermine the independence needed for Suriowecki's "wisdom of crowds" effect disproportionately affect the uninformed, who are therefore more likely to line up behind the informed than vice versa.
Actually, most elections depend on winning over swing voters, who tend to be the least-informed of all voters. Moreover, the choices of the ill-informed are not "essentially random." They fall into predictable patterns of error caused, in part, by their lack of information. For more details, see my paper here, among other writings of mine.
Lastly, it's worth noting that most markets are not "first past the post" systems and neither are elections in proportional representation system.
I think you mean that most close elections depend on winning over swing voters. (Most elections aren't close enough for swing voters really to matter.) Either way, your observation merely demonstrates that the market is efficient. Candidates craft their positions so as to appeal to a maximal number of thoughtful voters and their faithful bandwagons. Usually, one candidate does so substantially better than the others, and wins handily. Occasionally, two candidates do roughly equally well, and find themselves scrambling for the remaining highly random voters. Hence the "swing voter" phenomenon.
Moreover, the choices of the ill-informed are not "essentially random." They fall into predictable patterns of error caused, in part, by their lack of information.
I wish I had time to find and read your references 93 and 94, which apparently back up this claim. What "predictable patterns of error" are observed?
Lastly, it's worth noting that most markets are not "first past the post" systems and neither are elections in proportional representation system.
In many markets market leaders benefit disproportionately, creating at least a mild first-past-the-post effect. And proportional representation systems usually result in broad coalitions competing for de facto first-past-the-post electoral victory.
Actually, very few national elections are won by more than a 60-40 margin, and independent voters are at least a third of the electorate. So swing voters decide most, if not all, elections. Whether or not the market is efficient depends on what you mean by efficiency. It is efficient in the sense that politicians usually act rationally to maximize their chances of winning, but not efficient in the sense of necessarily creating optimal outcomes (or even close to it).
Candidates craft their positions so as to appeal to a maximal number of thoughtful voters and their faithful bandwagons. Usually, one candidate does so substantially better than the others, and wins handily. Occasionally, two candidates do roughly equally well, and find themselves scrambling for the remaining highly random voters. Hence the "swing voter" phenomenon.
The remaining voters are not "random" at all, but influenced by their ignorance. MOreover, "thoughtful voters," in the sense of voters who have even a basic level of political knowledge about candidates, issues, etc., are a relatively small proportoin of the electorate, probably no more than 20 or 25 percent. For lots of data on this, see the paper cited in my previous comment.
First of all, given that "independent voters" include many libertarians, far-left and far-right voters, single-issue voters, and de facto Democrats or Republicans who just happen to register as independents, I'd be very leery of equating "independent voters" and "swing voters" who are the "least-informed of all voters". I'd estimate that true swing voters of the kind you criticize make up maybe twenty percent of the electorate at most.
Regardless, though, a 60-40 margin of victory is more than enough to be consistent with swing voters having been irrelevant, when you consider that swing voters are at least somewhat random in their choices. (I know, I know--you assert that ignorant voters are also biased in their ignorance. I didn't see much detail on this in your paper, apart from those references 93 and 94 that I'm still hoping you'll summarize for me. But biased randomness can still contain plenty of randomness.)
Let's assume that in a given election, swing voters (I'll split the difference with you and call them a quarter of the electorate) split 50-50, while committed voters split 60-40. That would give a 57.5-42.5 margin in the final result--a blowout, by national election standards. Now, it's true that enough bias among swing voters would change the election result--but it would have to tilt 80-20 in favor of the underdog to do so. It wouldn't take much randomness on the part of swing voters to make that result highly unlikely in a large election. The winning politician in that scenario could therefore ignore swing voters completely, concentrating instead on keeping his or her bloc of committed voters loyal.