Some Microfoundations for Pragmatist Democracy

by Henry Farrell on February 13, 2013

One of the arguments that Knight and Johnson make is that standard ‘epistemic’ accounts of democracy do not provide a good foundation for understanding what democracy actually does. Such accounts argue that democratic institutions can do a good job at capturing and aggregating the knowledge of citizens, so that the collectivity can make better decisions than any individual. For example, Condorcet shows that if everyone is slightly more likely to be right than wrong, and if they make their judgments independently, then the more people who vote on a question, the more likely that they will collectively reach the right decision.
Knight and Johnson want to provide an account of democracy which also captures the benefits of collective knowledge, but which is much messier and more rumbustious, and better able to deal with stark disagreements among people with genuinely clashing interests and beliefs. Here, for example, they don’t think that people can or should make their judgments independently – they should and will argue with each other, and try to influence each other before voting. The value of democracy then, comes from its capacity for spirited argument – for exposing individuals, with their quarreling, diverse views, to each other, so that they can better evaluate politics, even if they never, ever converge on a common understanding.

This all seems good to me – indeed it converges with arguments that Cosma Shalizi and I have been making (this is unsurprising, since my share of this collaboration has been deeply influenced by Jack and Jim’s work in the past). What would be nice though, would be an _alternative_ set of microfoundations, which might provide a better understanding of when these arguments will work and when they don’t. To put it another way – it is plausible that Knight and Johnson’s argumentative account of democracy will work better (or perhaps: only work) under certain conditions, just as epistemic arguments for democracy only work under certain conditions. However, while bits and pieces of an argument about these conditions are visible in parts of the book, they aren’t drawn together in any very systematic way. So what I propose to do in this post is to set out a very general set of these conditions, drawing heavily on Cosma’s and my own arguments (while absolving Cosma for blame for any stupidies), as well as Knight and Johnson’s own work. This might or might not reflect Knight and Johnson’s own notion of what their microfoundations might be, but at the least, it should help them to figure out where they disagree …

So what might these microfoundations look like? Most obviously – people can’t murder each other over politics. Slightly less obviously, they cannot be idiots. Less obviously again, there needs to be some bounds on the space of politics – no political system can reasonably handle all problems (there are bandwidth constraints). Finally, even if people disagree vigorously over how to solve problems, they need to agree what the problems _are._

First – as Jack and Jim say in the opening pages, the conflicts of democracy have to be _constrained._ That is, people are going to disagree with each other, but they cannot disagree with each other so much that they are prepared to resort to undemocratic means (i.e. take up arms) in order to pursue their disagreements. Knight and Johnson’s account of how this might arise is more a sketch than a complete picture, although as they note, their argument at least highlights the need for such an account, where other normative accounts might gloss over the problem. One wants sufficient conflict to properly represent the genuine divergence of interests and beliefs in a given society, but this conflict should be entirely channeled through argument.

In some circumstances, this may be a heroic aspiration, but a number of societies seem to have managed (perhaps temporarily, perhaps provisionally) to have figured this out. This condition seems obvious, but it does suggest limits to the benefits of argumentative democracy. Specifically, we can expect rather more benefits from vigorous political disagreement in societies where people are prepared to settle their differences with votes than with knives. There is likely a sound pragmatic case for dampeners on disagreement in societies where democracy and the rule of law are very fragile.

Second, the Knight Johnson account of democracy shares some features with the Condorcet approach, in that it plausibly requires that people not be badly stupid or misinformed if they are to put collective knowledge to good work. Condorcet’s rule suggests that people individually need to be slightly more likely to be right than wrong for aggregation to produce good results. Cosma, relying on machine learning theory, would frame it a little differently – people only need to be _weak learners_ to contribute usefully to knowledge aggregation, but they do need to be weak learners. Different people may have different perspectives, each of which captures a different aspect of some underlying reality- but they have to have some imperfect glimpse of that truth to contribute usefully to democratic learning. If one is so badly misinformed, or confused, or intellectually warped that one’s perspective systematically detracts and distracts from the search for some underlying insight, then one’s participation in argument hurts rather than helps the democratic search for better understanding.

As Melissa suggests in her post, this suggests that a pragmatist account might be read to imply limitation of the franchise if it were shown that other kinds of rule produced better results . One possible response might be to accept this, and to say that a pragmatist account is incomplete because it doesn’t reflect basic norms of equality and the rights of citizens to engage in democratic politics, even if their contribution is wrong headed or obtuse. Another might be to acknowledge this as a theoretical problem, but to argue that in practice, it will be usually impossible to distinguish in advance between useful and not-useful views in any systematic way, given the complexity of major social problems, and the strong likelihood that diversity of viewpoint trumps individual sophistication (Scott Page’s work is very useful here). This would accept the _ex post_ participation of some non-useful viewpoints on the grounds that one can’t be sure _in advance_ that they won’t be useful. Obviously, you want to avoid these voices derailing discussion, but this may be tricky. The most unusual views may sometimes (often?) simultaneously be the most likely to be viewed hostilely by others as irrelevant, and to be the ones that are most valuable to democratic argument, precisely because they provide vantage points (and hence understandings of the underlying problem) that differ sharply from those held by the majority population.

Another interesting question concerns neurodiversity – neurodiverse people will often view the world in non-orthodox (and potentially very valuable) ways, that could have particular advantages for democratic debate. It is at least possible that some forms of diversity are more cognitively valuable than others. Certainly, there is only a partial fit between cognitive diversity and the more conventional kinds of diversity that institutions such as universities conventionally try to promote.

Third – and I can’t do more than to sketch out the problem here – the political issue-space needs somehow to be _bounded._ That is, there is a potentially very large number of possible problems that democracy might engage with, but an actually relatively small set that it is capable of handling, given institutional limits, the patience of people involved (the [Oscar Wilde]( problem) and so on. The bandwidth of the political system is going to be limited. The problem is, of course, that decisions over which issues get on the agenda of politics, and which do not, are inherently political decisions. If one takes seriously the arguments of a variety of political scientists, stretching from policy scholarship (Frank Baumgartner and his colleagues) through historical institutionalism (Pierson and Hacker) to various agenda setter models in rational choice, the choice of a small set of issues to be debated from the vast space of possible issues is likely to be just as politically tricky as argument within the issue space. The Knight-Johnson framework points towards this problem but doesn’t, as I see it, resolve it.

One possible external source of insight here might be Nancy Rosenblum’s [arguments]( about partisanship. Adapting her claims a bit loosely, Rosenblum makes a good case for how partisanship (when it works properly) can organize issue space usefully. Parties seek to respond to their beliefs about what the electorate wants or might want by bringing forward new issues and problems, and seeking to organize politics around these problems. They do so in a competitve arena, where other parties may seek shamelessly to steal their solutions, to redefine the problems in ways more congenial to them and so on. And they respond (when partisanship works properly) to voters, gradually abandoning issues which are vote losers. This has useful implications. First, it suggests how the issue space of politics may be bounded, so that it is not completely chaotic. Second, it suggests how it may be bounded in ways that are at least roughly responsive to the broader wants and needs of society.

Of course, this is neater in theory than in practice. Actual parties in the US are less electoral coalitions responding to voters, than policy coalitions working in a context where voters impose an outside bound on what they can get away with (the new party organization literature is good on this). In this world, the policy space will be bounded in ways that sometimes are intended specifically to frustrate the desires of majorities. However, this also provides a specific place where pragmatists might want to get to grips with institutional reform that could have salutary benefits for democracy. Making parties more responsive will make democracy better able to identify salient problems and, perhaps, to solve them.

Finally, there needs to be some consensus on _what problems are_ for argumentative democracy to work well in problem solving. If, to take a contemporary example, a large segment of the voting population think that human-caused global warming is not a problem, and are sufficiently well-entrenched, then the problem solving capacities of democracy aren’t going to be worth squat. Either the issue is never going to get political debate, or it will be possible for people who refuse to believe that this is a problem to block any action taken on the basis of debate. I don’t see how this can be solved within the limits of the democratic system itself. What it suggests to me is that the pragmatist agenda, if it is to be pursued, needs to look to institutions of knowledge formation outside of democracy proper, and how they can best intersect with democratic institutions. Philip Kitcher’s work is one obvious reference point, which pragmatist political theorists might want to grapple with, either to agree or to disagree with. But the more general point is that (a) passionate argument will only work well to reveal solutions to problems when a sufficient number of the arguers are well acquainted with reality, and (b) that this knowledge of reality is unlikely to be generated by democratic institutions themselves. The intersection between this condition and the second ‘people cannot be complete idiots’ condition above is too obvious to be worth belabouring at length.

Obviously, these conditions are not stated as precisely as they might be (this is a blogpost, not a philosophical treatise). They are not exhaustive, and are certainly open to challenge. Even so, I think that the Knight-Johnson approach needs to identify scope conditions, and these are as good as any. First, and much less importantly, clear statement of such conditions will make it much easier to organize debate between this approach and others (such as epistemic democracy). Second, and far more importantly, figuring out these conditions will make it easier to figure out precisely when this account is going to be _useful_ in guiding thought and reform, and when it will be less useful, or even perhaps fail.



Mike Huben 02.13.13 at 5:36 pm

It strikes me that “weak learners” as a microfoundation is too simplified to represent what really happens in democracy. Learning does not happen out of the blue, or even simply through discussion. It is disseminated through news sources, propaganda and other social sources. Learning also has bandwidth issues.


mpowell 02.13.13 at 5:46 pm

I don’t really understand your point about people needing to agree on what problems there are. It’s actually pretty frequent that the major political disagreement is over whether something is actually a problem (the deficit, segregation, etc). I don’t even know how you can really explain the difference between disagreement over the solution to a problem and disagreement about whether a problem exists. It doesn’t take much rhetorical contortions to frame most political disagreements one way or the other. And what does it matter, anyhow? The system doesn’t have to get the ‘right’ result, does it? It just has to do better than the alternatives. Another way of putting this is that people will have disagreements about which metrics to use to evaluate policy outcomes. This seems to me to be a better way of expressing what you may be trying to get at here, but surely you can’t rule this case out of your microfoundations, can you? Because it’s pretty hard to find societies where this isn’t an issue and plenty of reasonable successful democracies where it is.

I do like your point on ‘weak learning’ though. I think this is a much better way of putting it than to require that people be slightly more likely to be right. Because that frequently won’t be true initially. It’s also hard not to immediately think of the Fox News viewer demographic when you bring up this point… and think uh oh!


Glen Tomkins 02.13.13 at 6:18 pm

This approach starts off reasonably enough. It seems perfectly sensible to posit govt as a mechanism to reach sound decisions about public policy, and decide which forms of govt are best, based on their efficacy at reaching such sound decisions.

But see how quickly you run off the rails if you start from that seemingly reasonable position. Soon enough you’re up against the need to deny the franchise to people with a proven track record of reaching unsound decisions about public policy. But do any of that franchise denial at all, and soon thereafter we’ll be murdering each other for sure over who gets excluded, and we’re back to a Hobbesian state of nature in no time.

I don’t think the problem here is a bad choice of theoretical framework. The problem is the whole project of a theoretical framework that seeks to be prescriptive, when description is all we’re capable of. Precisely because this stab at a prescriptive framework is so sensible, and made with such careful attention to its inherent limitations and problems, the fact that it pitches us right back to Hobbes should tell us there’s something wrong with the whole approach, not the particular theory.

Descriptively, I would posit that democracy is what we get after we try forms of govt that put competent people in charge, and that turns out not to work in the long run, so we settle for the prospect of needed govt change happening only after a process of trial and error. We don’t expect the governors to get it right all of the time. In fact, their competence is likely to decline over time as they try to apply what worked before to a changing reality. So we have a system whose only stability is that the governors can be thrown out should they prove empirically t0 be bums. But that turns out to be better, or at least less bad, then a system that tries to do better than trial and error by making good or competent people the permanent governors.

Of course this is frustrating to reasonable people. It seems that we should be able to do better than trial and error, that we should be able to come up with reasonable public policy using forethought, rather than having to wait for the existing policies to fail to the point that even the middling voter sees the need to throw the bums out and change course by electing a new set of merely potential bums to run things until such time as they actualize their inherent bum nature.

Reasoned argument doesn’t play any immediate or direct role in this process, because the existing policies are pretty impervious to rational refutation right up until reality refutes them, and then the other side gets its turn to mess up. But this view of things is absolutely not a call to intellectual nihilism. The current regime will eventually fail in ways so dramatic that even the midlling voter, who has his or her own life’s business to attend to right up until govt failure makes them pay attention to public policy, will wake up sufficiently to throw them out of office. At that point, the other side needs to be ready with an alternative, hammered out in those long years of shouting reasoned arguments to the trees and rocks of the wilderness, for all the immediate good it does.


Nick 02.13.13 at 6:30 pm

Since two out of the first three comments made this mistake, it might be worth pointing out that “weak learner” is a statistical term for (translated to this context) someone who is right a little more than half the time on a specific subset of questions, and really has nothing to do with “learning” in the sense of knowledge accumulation.


shah8 02.13.13 at 6:44 pm

/me tosses an anthropology textbook at Henry as a friendly gesture.

Central underlying question seems to be “how do we motivate people to be responsive agents inside a democratic system?” There is a lot of “assume a can-opener” element here. Not least of which seems to be an absence of consideration that people as a whole only really count voting as an act of privilege (for which there *must* be outsiders that they know and interact with) and not really as a general inalienable right. People might say otherwise, but they generally act as I have said.

Then there are the problems of social leverage and deliberate impairment. Violence is not just snipers but also about blockades or incestuous amplification of malevolent concepts.

When it comes down to it, complex societies deliver rights to the most people, and not necessarily democratic societies. As such, it was only ever complex industries and war that truly powered democratic (and hey, Stalinistic, too) forms of governance. Humanity is a fire ecology. When there aren’t enough scared, anxious, elites out there, politics resolves down to extractatory and very reactionary autocracy.


Wonks Anonymous 02.13.13 at 7:43 pm

Your Oscar Wilde link is broken by a superfluous underscore at the end.


Trader Joe 02.13.13 at 7:47 pm

Interesting thought questions.

It seems like 5 people debating where to go to dinner is one sort of democracy – everyone votes, everyone voices an opinion, everyone is impacted.

500 people debating who should be town mayor is another – most people vote, most people are impacted and many people voice an opinion.

At some level of participation however it turns into 5 million people voting on American Idol – which is to say much less like a participative democracy and much more like a beauty contest and one we can’t be sure isn’t rigged.

It seems like part of the question being asked is how large can we make a democracy before it loses its democratic identity and is hyjacked by the systems designed to fulfill the democracy.


Tony Lynch 02.13.13 at 11:39 pm

@3: “So we have a system whose only stability is that the governors can be thrown out should they prove empirically t0 be bums. But that turns out to be better, or at least less bad, then a system that tries to do better than trial and error by making good or competent people the permanent governors.”

“They prove empirically to be bums” – to whom? On what (shared) criteria?

Besides if we “throw out the bums” who says we don’t simply want DIFFERENT bums? – one who’ll skew things our way? (this interest determining what we put into the “empirical” claim.)

And how do you know – even make a persuasive case – that this “system” – IS better than the alternatives? Any actual system always claims it is better than putative alternatives (which is why Rawl’s Difference Principle is only operationalisable as a conservative device). The real has a certain solidity and power that tends for obvious reasons to trump the merely imaginary.

Finally, what is the “democratic system” that is at least better than any other system? Does this system qua system include the USA, Venezuela, Zimbabwe, Iran &c. as – at this level of analysis – an undifferentiated monolith? If not, then how is “the system” usefully and non-trivially defined? What internal differences between regimes may it include? &c.


Omega Centauri 02.13.13 at 11:48 pm

I’m highly skeptical of the truthfinding capability of democracy. Were there any major scientific paradigm changes that would have happened if the issue had been decided by a majority vote of the respection field of science. New and revolutionary ideas have to be vigorously fought for, and an overwhelming body of evidence accumulated before a plurality switches sides.

Of course we have the issue of having people/organisations with loud megaphones, which ruins the prior assumption of independence. Even worse some of these organisations are not willing to make a good faith effort to only advance what they think is the truth. Thus the system becomes vulnerable to manipulation. So you need some sort of mechanism to enforce/encourage good epistemology. How do you do that, without running this risk of being overcontrolling?


Omega Centauri 02.13.13 at 11:49 pm

respection should have been respective.


John Quiggin 02.14.13 at 12:43 am

One problem with this kind of microfoundation is that it works most naturally if you assume decisions are made individually, either in a public assembly or in a referendum.

The closest feasible approximation is a PR system where the legislature can be seen as a representative sample of the population, delegated to work full-time on decisionmaking.

But even here, the reality is that parliamentary parties are different from the people they represent, and that’s even clearer in systems with more majoritarian characteristics.


John Quiggin 02.14.13 at 12:46 am

A more modest case for representative democracy is

1. Elections are the best way to achieve peaceful alternation of executive power between rival claimants

2. Among all electoral systems, one-person, one-vote is salient and stable.


Glen Tomkins 02.14.13 at 5:35 am


Lord, why would anyone want to defend democracy? That would be as beside the point as commending or disapproving of entropy. Entropy just is. Things tend to decay to the lowest energy level they can reach.

Democracy is a lot like entropy. It’s the heat death, end-state of forms of govt. It’s the form of govt that doesn’t allow anyone, not the one (monarchy) and not the few (aristocracy), to govern. Instead it gives us long periods of the inattention of the many to public affairs — during which the set of ideas and personnel left over from whatever side was on top when the dust settled from the last crisis is allowed to putter on along rails — punctuated by periods of crisis when the established set of ideas and personnel fail and things in the real world go to pot enough to rouse the many to a momentary fit of attention to public affairs, and then to their throwing the bums out.

The downside to this non-system of a system is obvious. We careen from crisis to crisis, without much forethought in between, because no one’s at the wheel in between the crises that force the many to actually vote in real change, we’re on autopilot. The upside is that it is therefore easy to make a change should some real-world disaster prove that we’ve been on the wrong course. We can just vote the bums out of office, with no need to guillotine them, and all the havoc that can create.

Of course there’s no guarantee that the many will not then choose worse bums, or will not do some throwing out of non-bums when it isn’t really warranted. No rational person would undertake what Swift has The Hack notionally write, “A Modest Defense of the Proceedings of the Rabble in All Ages.” But I think that would be an easier book to write than the defense of the proceedings of all the monarchies and aristocracies of history. These people definitely pay more attention to public affairs, they may even average out as more competent than the common herd (I doubt that), but leave them in charge, leave them in a state where their continuance in power is not completely in the power of the rabble, and they stop being empirical.

If we were capable of gathering an elite that wouldn’t have to rely on trial and error to run things, then we could do better than democracy. We would find these more competent people and put them in charge so that they could set public policy deductively from first principles. But as long as we don’t live in that world, as long as we can’t mint Philospher Kings whose “science” of governing is real and not mumbo jumbo, we will continue to need an empirical approach, we will need democracy’s ability to throw the bums out when they fail. We will continue to need to choose not to choose any form of govt, as Socrates ends up recommending in the Republic.


SamChevre 02.14.13 at 1:51 pm

These accoutns of democracy seem to me to be looking at a rather purer form of democracy than is the practice in current democratic societies.

Current OECD democracy has a critical feature that needs attention in an account of democratic decision-making. That is, it has a stable, professional Civil Service with very substantial power. (In countries where courts have substantial discretion to ignore democratic decisions, like the US, I’m including the courts in that Civil Service.)

This Civil Service looks to me like an attempt to deal with the knowledge problems and boundary cases listed above. Basically, the Civil Service serves as a limitation on the responsiveness of policy to “everyone democracy”, but is itself in many ways a limited-franchise democracy.

I will note that I prefer the older American system, where it requires consensus to decide that government should deal with a problem, and democracy only comes in afterward to decide HOW the government should deal with it. This, too, can be seen as a response to the limits above.


Jameson Quinn 02.14.13 at 2:08 pm

I see two things missing from this analysis.

First, there’s the whole issue of game theory and strategy. Plurality (FPTP) systems are nearly inevitably two-party systems (Duverger’s law) and thus they can get stuck in pernicious equilibria where majorities of voters agree on several needed changes, yet a lack of perfect unity combined with the first-mover problem prevents any of these changes from finding effective voice. I’d say that’s pretty much the situation in both the US and UK at the moment (with gerrymandering and the undemocratic filibuster providing an assist in the US case). This fall I start a PhD in which I hope to prove that other voting systems (such as approval voting or majority judgment; IRV is in this sense a dead end) are not subject to such pernicious equilibria.

This analysis has the advantage of actually offering useful advice. If you’re interested in following up on this advice: (not affiliated and I don’t endorse everything they say but they have the right basic idea), #endPlurality, #approvalVoting.

Second, democracy is not merely a system for making decisions and fomenting debate. It’s also a system of legitimating a government; in itself, an argument for putting down the guns. This is yet another reason why “stupid people don’t get to vote” is a pointless idea; it robs democracy of its legitimacy. The issue of legitimacy is also a basis for the people to demand better democracy (such as the improved voting systems I discussed above).

Finally: about the stupid people (we know who they are). I’d say that despite all the worries about epistemic closure and demagoguery and other stupidity-enhancing processes, a strong majority are at least “weak learners” if reality hits us upside the head hard enough. That blow can be in the form of either (electoral) defeat or (economic/environmental/civil rights) disaster; obviously, the former is preferable, as it does not extend to those who didn’t earn it. Plurality voting and its pernicious equilibria discussed above postpone defeat, and make disaster more likely.

(note: this comment was initially and mistakenly posted in the wrong thread. Sorry.)


Mao Cheng Ji 02.14.13 at 2:41 pm

“It’s also a system of legitimating a government; in itself, an argument for putting down the guns.”

Is it, really? What is the argument, exactly; what about the tyranny of the majority? I’d say the divine nature of the king (or of The Constitution, or the Vanguard Party) would be a much stronger argument for putting down the guns, than the result or an election or a referendum.


Jameson Quinn 02.14.13 at 4:37 pm

The majority can claim to be divine by Condorcet’s theorem. (Or just because: “majority rules, minority drools”). That’s a better claim in my book than the king/constitution/vanguard. And while tyranny of the majority happens sometimes, it seems less common/absolute than when you have a king/vanguard, because pity from a small fraction of the majority scuttles it.


Mao Cheng Ji 02.14.13 at 7:19 pm

Suppose there are 3 of us, and we are ordering food from a restaurant.We take a vote, I choose Italian, you and the third guy Chinese.

So, I’m eating Chinese today. The utilitarian calculation doesn’t work here, because it’s possible that you and the third guy only slightly prefer Chinese, and I really-really hate it. My intense displeasure is stronger that a slight increase of the pleasure in the two of you. Nevertheless, you win.

But you’re right, there is something there indeed, something that makes me put down the gun. It’s the fact that the two of you, a majority, will probably be able to kick the shit out of me. And it seems to me (at the moment. I may not be thinking straight, got a touch of flu) that this is the only ‘legitimacy’ that you can claim. Which is, actually, exactly what you do (“majority rules, minority drools”), right? But then, if I manage to get a machine gun, the ‘legitimacy’ is mine, right?


Omega Centauri 02.14.13 at 7:50 pm

“majority rules, minority drools”.
“pity from a small fraction of the majority scuttles it”
Except when it doesn’t. Which is often the case if one group (a majority narrow or not) considers an issue to be of overriding importance. That seems to be the case in significant parts of the world today (thinking Arab spring countries here), were it is religious fundamentalists versus secularists. In this case it looks like whomeverwhenever the fundies gets power, they push through their agenda regardless of the feelings of the “minority”.


John Quiggin 02.15.13 at 1:21 am

” stuck in pernicious equilibria where majorities of voters agree on several needed changes, yet a lack of perfect unity combined with the first-mover problem prevents any of these changes from finding effective voice – IRV is in this sense a dead end”

How so? For concreteness, suppose the issue is environmental preservation. Green candidates can run, directing preferences to whichever major party candidate is most favorable on environmental issues. If they get a substantial vote, and their recommendation on preferences has some effect, the major parties have a strong incentive to shift position. This happens routinely in Australia, on a wide variety of issues.


Jameson Quinn 02.15.13 at 3:09 am

Why do I say IRV is a dead end? Play with (or watch the video at of someone else playing with it if you can’t figure it out.)

The basic problem that shows is nonmonotonicity, in the form of “center squeeze”. Taking your issue of environmentalism, say the Green candidate got 30, the mainstream left party got 25, and the mainstream right (after all fringe right parties are eliminated) got 45. So the left is eliminated, and 10% of that vote goes right, and the right party wins. But if the Green were not in the race, the left would win easily with up to 55%. Insofar as people realize this can happen (perhaps from experience) they will strategically vote for the mainstream left, and you’re back to a strategic (pernicious) equilibrium of two-party dominance. Which, by the way, definitely holds in Australia (well, the dominance; though strategy is probably not the sole reason).

I’m currently running an experiment on Mechanical Turk which is validating this sort of problem for IRV, and shows that Majority Judgment is (in my scenario) immune and Approval Voting (relatively) resistant. If you want to know when I have a preprint available, email me (firstname dot lastname at gmail).


John Quiggin 02.15.13 at 7:10 am

The scenario you describe almost never happens. The equilibrium is one in which the major parties nearly always take first and second place, meaning that it’s optimal for third party supporters to vote sincerely. So, the Greens (or for that matter, gun lobby supporters) can signal the strength of their support, and bargain over their preference allocations, as happens at every election. In the rare cases when a non-monotonic outcome appears likely, it’s more common for the weaker of the mainstream parties to withdraw, or run dead, allowing the third party to win a seat at the expense of their main opponent. Everyone in Australia knows about this, which leads me to think that you aren’t Australian and that your claims are based on ignorance or, at least, very limited information.

In any case, you would be better advised to look at the ample data actually available, rather than simulations, which can easily be rigged to favor some particular choice.


Z 02.15.13 at 9:04 am

but a number of societies seem to have managed (perhaps temporarily, perhaps provisionally) to have figured this out.

I agree with shah8 that at least part of the historically accurate answer is going to come from anthropology. More precisely, I would venture that the rise of alphabetization triggers a recasting of some anthropological values prevalent in a given population as abstract political (or religious) ideals and that these ideals are (frequently) the common ground on which the civic debate can take place. From that point of view, the rise of political islam (for instance) is the historical norm rather than an aberration.

See for instance Inde : la démocratie par la caste from C.Jaffrelot for a case study and L’Invention de l’Europe from E.Todd for a (tentative, but quite impressive and intellectually very challenging) formalization at the european scale.


Mao Cheng Ji 02.15.13 at 9:05 am

So, the problem with IRV, if I understand correctly, is that everybody’s second choice is going to be eliminated. And the creator of that website believes that everybody’s second choice is the ideal candidate. But maybe it isn’t, maybe it should be eliminated. If so few voters nominated it as their first choice, something is probably wrong with it. Centrism is not necessarily better. As they say, there’s nothing in the middle of the road but yellow stripes and dead armadillos.


SusanC 02.15.13 at 10:07 am

Finally, there needs to be some consensus on what problems are for argumentative democracy to work well in problem solving.

In the hypothetical example where most of the voters don’t think global warming is a problem, then there is a consensus on what the problems are: and the consensus is that global warming isn’t a problem.

And if they turn out to be right, then this is the system working.

The potential problem is that the consensus opinion might turn out to be wrong. You’re asking voters to predict the future, and they might be mistaken.

I think there’s a fundamental difference between polling people on their current preferences — in which there is no objective fact of the matter — and polling them on a matter of fact (about which they can potentially all be mistaken), or their future preferences (about which they can also be mistaken, because people’s preferences can change over time in ways they did not predict, possibly because the physical world changed in ways they did not predict).


Jameson Quinn 02.15.13 at 11:53 am

JQ: I think we actually agree substantially. My parenthetical comment about “dead end” was intended to suggest a path that allows a certain limited amount of progress, but that ultimately leaves two-party domination in place. I do think IRV is a better system than Plurality. It’s clear that the Australian system allows minor parties more influence, but it does put an artificial limit on their growth, such that third-party wins in the Australian House of Representatives (and similar state/territorial bodies which use IRV) are about as rare as in the US. You argue that people are aware of the potential problem of nonmonotonicity and effectively prevent its occurrence; that’s exactly what I was saying, and indeed the very phrase “strategic equilibrium” implies it. And by the way: of course I’m not Australian (though my mother and her family are), or I would have said “preferential vote” instead of “IRV”.

As to the importance of real-world data rather than simulations, I also agree there. That’s why I’m doing my current research on Mechanical Turk, which uses human subjects. However, I am using induced preferences (ie, payments) to investigate strategy, as real election returns do not directly show honest preferences for strategic voters, and also because there is little data for some of the systems I’m studying, such as Approval Voting and especially Majority Judgment.



Jameson Quinn 02.15.13 at 12:09 pm

Mao@24: You’re right that sometimes the compromise candidate deserves elimination. But sometimes they don’t. Voters are pretty good at distinguishing these situations, and systems like Approval Voting or Majority Judgment allow them to vote honestly to attain either outcome. But with IRV, there are cases where in order to prevent the elimination of a desired compromise, voters have to strategically misstate their true preferences. This incentive for dishonesty, in turn, can cause unhealthy equilibria which tend to stymie political change.

IRV is well-suited for Australia, a parliamentary system where people are used to it. In the US, with a strong tradition of “independent” voters, a system which doesn’t force “forming a government”, and voters who don’t and probably wouldn’t understand IRV (if the SF Bay Area experiences are anything to go by), it would have greater problems, though it would still be an improvement over the current broken system.


EqualToJake 02.16.13 at 5:35 pm

Can’t the problem with the second microfoundation (improving democratic problem solving by franchise exclusion) conflicting with “justice” be resolved simply by prioritising the first microfoundation (constraining conflict)? If the excluded people will no longer accept conflict limits then they have to be included, even if their problem solving limitations reduce the overall effectiveness of democratic problem solving?
And in any case why is franchise exclusion necessarily a bad thing? All democracies exclude children and some exclude felons and the mentally incapable. And historically there are many examples of democracies which where very successfull at problem solving while excluding large segments of their population. What matters is whether the excluded people will accept it.


Lt Collin Street 02.17.13 at 11:04 am

It always struck me that the problems with condorcet become obvious when you remember that elections aren’t one-off events but are repeated: condorcet elects a middle-of-the-road compromise candidate, won’t elect anyone off in the fringes. That’s its big selling point. Each and every time, without exception. For a once-off that’s OK, I guess, but a process that election-after-election returns a chamber of a hundred and fifty, or four-forty-odd, or six-fifty-odd, more-or-less identical middle-of-the-road compromise candidates might I think be called with some justice less than a perfect mirror of the electorate.


Bill Tozier 02.17.13 at 3:46 pm

I arrive late, and need to catch up in a big way (not least by grabbing a copy of the book). But a few immediate reactions are provoked:

– It’s interesting and noteworthy whenever anybody discussing a work in the Pragmatist tradition, critically or even in a comment, invokes the notion of “truth”. ’nuff said.

– One of the reasons I’m late to the discussion is that I’ve been up to my elbows in Pickering’s Mangle of Practice, and I have to say there are many overlaps and congruences between Pickering’s study of how “science” gets done, and what I infer as the Knight–Johnson approach. But there’s something I haven’t seen mentioned, and if it’s present or implied in Knight–Johnson then it’s surely a microfoundation of their treatment of democracy as much as it is Pickering’s discussion of science:

Isn’t the democratic “planning” system the one in which a diversity of people can try things out? In other words, isn’t the creation and enforcement of policy and law the mechanism through which idealists of all types—whether in the majority or minority—are exposed to the practical consequences of the laws that are passed?

This isn’t a question about “truth-finding”, but rather one of the dynamics by which democracies undermine ideologues: Isn’t it the case that in a (functioning) democracy, the occasional instantiation of ideologically-informed policies makes their consequences more apparent? It seems to me that there’s feedback there which is not only itself necessary, but should probably suggest more than just “good communication”: The inefficiencies of the democratic notion of “truth” mean that even recently popular models, policies and stances are subject to reexamination.

I guess what I’m trying to ask is: Isn’t it the case that democratic societies are more consistently made aware that they’re collectively exploring what might work in any given context, and are structurally better able to adapt when that information changes opinion?

As long as the participants know what’s happened recently, that is….


Jameson Quinn 02.18.13 at 7:37 pm

LtStreet@29: It appears your comment was prompted by my raising the issue of voting systems. Excellent; it appears the derailing of this thread was partly successful.

But nobody above you suggested Condorcet was an ideal system. The systems I suggested — approval and Majority Judgment — both allow an electorate to vote so as to elect a centrist, or not, in both cases honestly. So I think your criticism is valid, though its seriousness in practice is unknown.

(Well, mostly; if people really got tired of guaranteed centrist wins, they could strategically bury the centrist; even if they still preferred the centrist to the other extreme, if that preference were weak enough such burial could be a rational gamble, deliberately creating a Condorcet cycle which anyone might win rather than continuing a centrist dominance. But while that would cure the problem you note of centrist dominance, the problems of dishonest strategic voting it created could well be worse.)

ps. Also I wanted to apologize to the other JQ. I didn’t realize you were Australian and I didn’t mean to, as we Americans say, teach my grandmother to suck eggs. Sorry.

Comments on this entry are closed.