For the past six months or so, I’ve been working on a project on what I call “character-based voting” (CBV), construed as voting for a political candidate based on her traits of character, as contrasted with “policy-based voting” (PBV) which is voting for a political candidate based on the expected consequences of the candidate’s expected policies.

It’s a rough and in some contexts problematic distinction, but clear enough to work with. There’s a clear enough distinction to be drawn between voting for a candidate because you regard her as more honest than her rival, and voting for a candidate because you expect her to enact policies X1…Xn, which have expected consequences C1…Cn, which you regard as net favorable, but which you don’t expect her rival to enact. My modest claim is that CBV can in principle be justified, and has its place.

There is a literature on both CBV and PBV in political science, but not much of one in political philosophy. Political scientists have tended to argue that policy-based voting is the paradigmatically rational approach to voting whereas character-based voting is not. That older consensus is now giving way to a view that takes a more positive view of CBV (see the work of Loes Aaldering and Scott Clifford), but the literature is in its early stages and leaves a lot undiscussed. The only philosophical discussion of the topic that I know is Jason Brennan’s in The Ethics of Voting (and elsewhere). Brennan’s arguments against CBV strike me as both cursory and weak, but that’s a topic for another post.

No matter how positive a view one takes of CBV, however, it’s obvious that one has to distinguish the cases in which it has a hope of being justified from those in which it really is irrational. It’s one thing to vote for a political candidate because she’s more honest than her rival; it’s another to vote for her because she’s sexier than her rival. Both are traits of character, but only one is relevant to the task of holding political office, and voters have a real tendency to confuse the two.

Honesty is a politically relevant trait because the assessment of policy depends so heavily on it, especially in democracies that rely heavily on voter consensus created through open political discourse. Policy wonks often seem to forget that a political candidate in a democratic election isn’t just an impersonal lever for the promotion of policies; she’s a person who has to convince her would-be constituents (and colleagues) of the rightness of her proposed policies, or at the very least, to inform people of what those policy proposals are. A dishonest politician suffers from an obvious liability in this respect. You can’t convince people of anything unless you induce them to believe you. But if they regard as you as dishonest–and in particular, as extraordinarily dishonest–they will tend to disbelieve you. If so, they will tend not to be persuaded by what you say about policy, or suspect it (and suspect you) even when you’re telling the truth, or at least trying to. And if you are dishonest, they’ll largely be right.

A cynic might at this point be inclined to laugh, or at least chuckle. Honesty can’t be that important, he might say. Politics isn’t the Groundwork for the Metaphysics of Morals. Politicians lie all the time about everything. We all know this, and all just tolerate it, correcting for the inevitable signal-to-noise or truth-to-falsity ratio we encounter. Once we do that, we’re then free to focus on policy, which is what we should focus on. So CBV is beside the point, even when it comes to honesty.

The cynical view has a point—up to a point. But it seems to me that there’s a point at which it clearly fails, and that’s when one candidate is basically honest (by political standards), but the other tells blatant, Orwellian lies with a straight face, expecting to be believed–often actually being believed by large swatches of the electorate–but giving the real impression of being either an all-out con man or a sociopath, or both.

The question then arises whether a candidate’s character can be so fucked up that his putative policy positions cease to matter—either because you can’t really trust him to tell you what those policies are, or because his character is so defective that it makes more sense to elect the decent-but-policy-suboptimal candidate than the psychopathic-but-putatively-policy-optimal one. Believe it or not, stuff like this sometimes happens, even in America.

Let me try to simplify the preceding scenario for purposes of a thought experiment. Of course, when it comes to thought-experiments, “simplifying” means “abstracting from real-world complexity,” which really means…

making complex stipulations in order to simplify a situation for purposes of analysis, in the hopes that the reader will make the right inferences from the analysis, and also in the hopes that the reader will then re-introduce the relevant complexity when he applies the inferences back to the real world that the thought-experiment was intended to illuminate.

Which never happens. But anyway, take two candidates, Smith and Jones.

As far as traits are concerned: Smith is fundamentally honest—doesn’t lie, doesn’t stretch the truth, doesn’t aim to deceive, tries her best to be as candid as political life permits her to be, and is willing to pay the political price for her candor.

By contrast, Jones is not just a liar, but a fucking liar: he lies about everything, consequential and inconsequential; half the time, you get the sense that he’s lying for fun, and half the time, you get the sense that he’s lying to himself. But you also wonder whether his whole lying routine is intended to deceive you into complacency: if he lies all the time, there’s a danger that people will start to take his dishonesty for granted, get jaded by the sheer quantity of lies, equate his extraordinary dishonesty with everyone else’s garden-variety dishonesty, and finally get conned into something really big. Assume that for now, nothing of that sort has actually happened, but that as far as Jones is concerned, it could.

Let’s hold the preceding descriptions constant, but imagine five different cases which vary in their policy implications.

1. Smith, the honest candidate, is, all things considered, policy-superior to Jones.
2. Smith and Jones are policy-equivalent. Neither is perfect, but all-things-considered, Smith is no better than Jones, and Jones is no better than Smith.
3. Jones, the dishonest candidate, is all-things-considered slightly preferable to Smith on policy, making Smith the policy-suboptimal candidate. But Smith’s suboptimality doesn’t put her too far behind Jones. In other words, as far as policy is concerned, Jones is good, but Smith is acceptable.
4. Jones, the dishonest candidate, is all-things-considered substantially preferable to Smith on policy. Dishonesty aside, his policy preferences are spot-on, at least if he can be trusted to enact them. Smith, by contrast, has some seriously problematic policy preferences—not disastrous, but discernibly, seriously worse than Jones. But at least she’s honest.
5. Jones, the dishonest candidate, has spot-on policy preferences (as [4] above). But Smith’s preferences, though honestly expressed, will very likely lead to all-out disaster.

With that in hand, let me make three (or so) simplifying assumptions.

For one thing, let’s set aside the possibility of single-issue or litmus test voting; when I say that a candidate is “all-things-considered policy superior” to another, I mean that the one candidate is better than the other on the basis of a global consideration of all of the policies he or she can be expected to adopt, not on the basis of a single policy that clinches the case while ignoring all the others. In saying this, I’m really borrowing an assumption from the literature on PBV to the effect that there is such a thing as a “global consideration,” and that voters know how to engage in one.

As it happens, I’m skeptical that anyone really does have a clear account of “all-things-considered” policy assessments, or that voters engage in them. (I don’t even think act-consequentialists have a workable account of either thing, and of course, a commitment to act-consequentialism creates problems of its own). Unlike, say, Michael Huemer, I don’t reject the legitimacy of single issue voting, either. It just simplifies things here to make some of the same assumptions as defenders of PBV, so I’ve done so. But part of the problem I have with them, not further discussed here, is that I reject both their one-eyed insistence on PBV as contrasted with CBV, as well as their conception of each half of the CBV/PBV contrast, which then affects the nature of the (supposed) contrast itself. But as that’s too much to deal with in one place, I mention it and move on.

Second, I’m assuming that we can make trait-ascriptions to people like honesty, and that we have an account of a politically relevant conception of character traits (not just a generalized account to the effect that “honesty” is politically relevant, but an account of the specifically political version of honesty that is relevant). This is another big assumption, one that defenders of PBV might well reject on grounds that trait-ascriptions all tend to involve bias or error. Again, that’s too large an issue to adjudicate here, so I mostly set it aside. (I don’t know where I stand on Roderick Long’s overall argument in this paper, but I agree with enough of his criticisms of “the trait-ascription error literature” to want to cite it here.)

Third, I’m assuming (very implausibly) that we can predict the likely consequences of the policies a candidate is likely to enact. Actually, the very idea of “a policy that a candidate is likely to enact” strikes me as semi-nonsensical, but I’m going to ignore that fact here. Obviously, political candidates do not enact policies on their own, and aren’t, even by their voting records, obviously responsible for all of the consequences of all of the policies they do support. Nor, by the way, do social scientists even pretend to track, or claim the ability to track, all the relevant expected policy consequences of a given candidate’s expected policies in all of the cases where tracking them would be politically important (I take that to be obvious, but if you don’t think it is, see this paper by Hendrik van den Berg).

But don’t worry yourself about the preceding assumption, because for purposes of this post, I’m pretending to be an analytic philosopher, which permits me to over-simplify obvious facts by legislating them away by means of under-analyzed Latin phrases like “ex hypothesi” and “ceteris paribus.” It would over-complicate things to deal with issues of causal attenuation or unpredictability, so I’m not going there, at least “for present purposes.”

OK, I’m done “simplifying” things. By the way, if you think I make big assumptions, you should read the other guys in this literature.

I take it that (1), (2), and (5) are easy, or relatively easy, cases—but also relatively uninformative for that very reason.

Case (1) is easy because it involves over-determination: since it makes sense to vote for Smith on policy grounds by itself, that fact tends to conceal the causal role of the character-based considerations that might simultaneously be operating. So the choice is as easy, but the role of CBV unclear.

Case (2) is easy because it’s a tie-breaker: it suggests that character-based considerations become relevant when all or most of the others are exhausted. That’s not nothing, but it’s pretty faint praise. It says that we vote for the honest candidate because honesty is at least a weak consideration in favor of the candidate, even if it’s the last consideration in a lexical ordering of legitimate considerations. That proves that some CBV is justified, but only under rare and narrow circumstances.

I suppose that ties do arise in any election where the policy issues are relatively settled so that the election turns in some way on which candidate will operate in good faith. Primaries are sometimes like this. So are lots of local elections, e.g., for judge or sheriff, where the law is basically settled, and the officeholder is just adjudicating or enforcing it in a routine way. Of course, lots of elections aren’t like this, and it’s probably rare to find an election where the candidates are precisely policy-equivalent ex ante for any given voter.

Case (5) is easy (or seems easy) because it allows (or seems to allow) us to set character-based considerations aside as obviously irrelevant so as to avert an imminent disaster. But its ease is of no help to a defense of CBV, since the ease of the decision arises, precisely, from a case in which CBV seems obviously silly.

On the other hand, case (5) can in a certain way be construed as offering backhanded support to CBV. For some voters, case (5) might give reason not to vote for anyone, or to vote “None of the Above” despite the disaster that might arise, something I ignored in the other two cases above as unlikely or irrelevant (unless you oppose voting altogether). At a minimum, abstention or a NOTA vote in case (5) implies that immoral traits of character can disqualify an otherwise qualified candidate once the candidate crosses a certain threshold of immorality.

Of course, it’s a further question whether it really makes sense to disqualify an immoral candidate at the price of allowing the sky to fall. But if you don’t think that the sky’s falling is your problem–you are, after all, merely allowing it to fall by failing to make a very, very minimal contribution to its not falling–then there’s nothing wrong with this line of reasoning. To be clear: the preceding reasoning is most relevant in the case of the otherwise-eager voter driven to abstinence by the repulsiveness of case (5), not the voter who sits out case (5) because he’d sit most of the other cases as well.

That leaves us with the really contentious cases, (3) and (4).

Case (3) is in some sense the crucial one, the one that most clearly distinguishes the monistic defenders of PBV from anyone sympathetic in any context to CBV. Here is the CBV-friendly interpretation: There’s a point at which a candidate’s immorality is such that it either disqualifies or tends to disqualify the candidate for office in ways that can’t be compensated-for or offset by his policy-superiority to rivals who are at least in the ballpark of policy sanity. In a case where the rival candidate is substantially more honest but only marginally policy-inferior, I would say that honesty is the kind of electoral asset that can tip the scales in favor of the honest-but-slightly-policy-suboptimal candidate as against the dishonest-but-policy-optimal candidate.

The question is how to cash this out; for now, I just want to say that the claim is plausible enough to be worth cashing out, and that its plausibility poses a problem for any rigidly one-dimensional defense of PBV that flatly excludes CBV. The clearest case for CBV is the one where the difference between the two candidates in, say, honesty is dramatic but the policy differences are relatively minor, even if they redound to the (marginal) favor of the dishonest candidate. It seems perfectly rational to vote for the honest-but-policy-slightly-inferior candidate to the fucking liar whose policies, if enacted, would be marginally better than the honest candidate’s (i.e., would be slightly superior in relative terms to policies that, absolutely considered, are not-bad but not as good).

Of course, though honesty is an awfully convenient example to use, it’s also slightly problematic and potentially misleading. In relying on honesty in our analysis, there’s a danger of confounding moral and epistemic considerations: a dishonest candidate is immoral but also a source of uncertainty; the first is a moral consideration, the second an epistemic one.  In saying that dishonesty disqualifies a candidate in an election campaign, it’s unclear whether that means that dishonesty disqualifies the candidate qua moral consideration or qua epistemic one or both.

It sounds like it shouldn’t make a difference, but it does. The moral issue is: Jones is dishonest, but the consequences of his proposed policies are better, and arguably more choiceworthy than Smith’s. So should we opt for dishonest consequences or consequence-suboptimal honesty? The epistemic issue is: as the honest candidate, Smith’s policy platform can rationally be believed; Jones’s cannot. So should we go with the credible candidate, or gamble on the non-credible one? In other words, if we disqualify Jones “on grounds of dishonesty” without specifying which issue we mean, it’s unclear whether we’re disqualifying Jones because…

• It’s a moral fact that: Smith’s sheer possession of honesty is more choiceworthy than the best consequences of the best policies that Jones enacts (taking their enactment for granted), or because
• Smith is simply more believable than Jones, so that we can’t rationally expect Jones to enact the optimal policies he claims to promote.

In the first case, we’re comparing a trait-possession (Smith’s honesty) with a set of policy consequences (the actual consequences of the policies for which Jones is responsible). In the second, we’re comparing two different sets of probabilities with respect to the policy consequences expected to arise on each candidate’s being elected (i.e., policies that are more probable because you can trust the person promising them, versus policies that are less probable because you can’t).

These are different comparisons involving different considerations, and it’s a conceptual mistake to conflate them, even if it’s very hard to disentangle the different factors in real-world deliberation and action. As I defined it at the outset, CBV includes both moral and epistemic traits of character (and others as well). Since honesty has both epistemic and ethical dimensions, the distinction between immorality and unpredictability may not matter all that much in most practical contexts, or at least not matter enough to affect the point I’m making, viz., that traits matter (should matter) to voting.

Exactly how or why traits matter is a separate issue. I suspect that the epistemic reason for trusting an honest-but-policy-suboptimal candidate is easier to grasp than the ethical one. But that doesn’t mean that the ethical one should be dismissed. Better, I think, to imagine an an inclusively disjunctive account that allows traits to matter in both epistemic and ethical ways. Maybe honesty trumps good policy in case (3) both because when purchased dishonestly, good policy is an ill-gotten good, and/or because you can’t trust a dishonest person to do anything, much less enact the good policies he promises. I concede that that formulation allows me to cover the bases without really resolving the hardest issues in play. That said, I still think the claim I’m making about case (3) has a certain plausibility about it. (The issue here is similar to the one that comes up when we think about the exclusionary rule in criminal procedure: do we exclude procedure-violating evidence because it’s epistemically unreliable, or simply because, being wrongly acquired, it would be immoral to include it? Same with torture: morally wrong or just epistemically unreliable?)

Case (4) strikes me as less clear than (3), but still seems to point in a determinate direction: abstention. It makes no sense to vote for an honest duffer. But it seems immoral to give political legitimacy to a psychopath who gets the trains to run on time. I don’t know how to decide between the honest duffer and the psychopath, so I’m inclined to decide (4) by recommending non-decision, i.e., by recommending the refusal to vote, or write-in vote for a third option. A case like (4) ought to dampen the enthusiasm of even the most zealous advocate of voting–and throw cold water on any advocate of compulsory voting.

I should lay my cards on the table and tell you that the preceding exercise was inspired less by a purely theoretical interest in CBV (though I have one), than by a response to a specific incident which outraged me, and which I subsequently decided to theorize. I don’t apologize for that, or regard it as a theoretical skeleton in my closet. I’m inclined to think that the issue of CBV could use more attention to and analysis of particular cases. Anyway, here’s the incident in video form.

Jay Webber is the Republican candidate for Congress in New Jersey’s 11^th congressional district; Mikie Sherrill is the Democrat. A few months ago, Sherrill appeared and spoke at a demonstration critical of the Trump Administration on immigration policy; her sympathies were clearly with the demonstrators. Some of the demonstrators there wanted to abolish ICE, a view that Sherrill herself doesn’t happen to hold. Though she’s explicitly disavowed the view, Webber wants, for reasons of sheer expediency, to pin it on her. So he’s insisted, over and over ad nauseam, that she once held the view and is now backtracking on having held it only because of the attention he’s brought to it, and through the pressure he and people like him are exerting on her. (Here’s a similar but more ambiguous case that deserves a discussion of its own.)

Webber is lying: there’s no other way of describing the speech acts in question. Pressed to justify his claims, he simply repeats the lie—again and again and again, each time with a renewed conviction that seems as much designed to convince his listeners as to convince himself. How a man can manage to bullshit himself and others in so egregiously obvious a fashion, I don’t know. But here he is, doing it on tape—insisting on it as though all-out fucking nonsense were the most obvious thing in the world. (To be clear: Webber is not the model for Jones in the thought-experiments above, but Sherrill is the model for Smith. In technical terms, I regard Webber as a mere liar, but not a fucking liar.)

I shouldn’t need to make this explicit, but I guess I have to: there’s a big difference between calling for the abolition of ICE and appearing at a demonstration where someone else calls for the abolition of ICE. There’s even a big difference between calling for the abolition of ICE and being a sympathetic participant in a demonstration where some people are calling for the abolition of ICE. A person unable or unwilling to make such distinctions—more precisely, a person positively intent on subverting or erasing such distinctions—has in my view disqualified himself from office. One also lacks a certain moral standing to make such charges if one accepts an endorsement from the likes of Donald Trump.

I take a lie of the sort Webber told to indicate dishonesty as a general trait of character in the person telling it. Yes, it’s “only” one lie, but it’s a bald-faced lie intended as character-assassination, and repeated with a persistence that indicates a propensity to lie about others’ character (a propensity borne out by his performance in this debate with Sherrill). In this respect, Webber’s dishonesty resembles that of Duncan Hunter in this now-notorious ad accusing Ammar Campa-Najjar of terrorist associations, something that strikes me as even more obviously disqualifying for political office than Webber’s claim. (Hunter is probably closer to a fucking liar than Webber.)

I happen to think that Sherrill is policy-superior to Webber, so I regard myself in this instance as being in case (1) above. But I’d vote for Sherrill over Webber in cases (2) and (3), as well. Case (5) is too distant from the actual case to be conceivable, at least by me, but case (4) is just barely conceivable: I could at least imagine someone’s convincing me that Sherrill’s policy preferences were a lot worse than I realized (I disagree with some of them as it stands), and she could in principle adopt some new policy position between now and the election that was problematic enough to put us in case (4). If case (4), arose, I’d abstain. But unless I somehow found myself in case (5), I couldn’t bring myself to vote for someone like Webber.

Maybe all this is just my priggishness, naivete, moralism, squeamishness, policy-illiteracy, humanistic soft-headedness, Democratic partisanship, and/or dogmatism talking, but I don’t think so. My defense of CBV reflects what I take to be the fact that character-based considerations play a subtle, powerful, justifiable, but generally under-theorized role in electoral behavior.

In my experience, the most prodigious crappers-on-character-in-politics are social scientists who insist that an emphasis on “morals” (as they like to put it) interferes with the objectivity required of a truly wertfrei policy wonk (itself the model of The Perfect Voter). We get this problematic idea from another Weber, Max Weber, who argues for it, among other places, in his Vocation Lectures. This isn’t the place to start a fight with a dead German political scientist, much less to throw the Weimar Republic,  the Third Reich, World War II, the Holocaust, and the Cold War in his face. I’ll just say that it seems to me that in practice the Webers of the world tend to facilitate its Webbers, which seems a bad thing. Oh what tangled webs they weave, when first they practice to deceive.

CBV is not a panacea for all of our political ills, or always the right thing to do, but it isn’t the paradigm of irrational or wrongful voting that so many political scientists have made of it. Call me crazy, but I choose my doctor, dentist, mechanic, accountant, real estate broker, and lawyer(s) as much for their traits of character as for the policy outcomes I expect from working with them (recall that competence is a trait). Mutatis mutandis, I’ve even chosen a couple of wives that way.

Something similar might be said of the instructors I’ve hired as a department chair, the job candidates I’ve preferred in job searches, and the bosses/supervisors I’ve tried to work for (or avoid) both in academia and outside of it. I “voted” for them as much on character-based as on policy-based grounds in ways that seemed to run the two things together. When it comes to specifically political elections, character-based considerations are obviously relevant to the election of, say, a sheriff; sometimes, they tell you when it’s time to get rid of one. Actually, examples like these are really the tip of a little-examined iceberg at the intersection of business ethics, political science, and political philosophy. Maybe a collision would be a good thing.

To choose someone over her competitors for a contested position is often to choose her for her “professionalism,” a trait that denotes the integration of technical competence with epistemic virtue and moral probity. If it’s pointless to divorce character from expected outcomes in so many professional contexts, as it seems to be, perhaps it’s pointless to do so in politics. Or so I’m inclined to think. But if you disagree, feel free to explain why. But you’ll have to hurry. The Hunterdon County Clerk, a very competent and conscientious lady named Mary Melfi, tells me that my mail-in ballot is on its way to my mailbox. In fact, she says it’s supposed to get there sometime today. And I believe her. Go figure.

Thanks to James Boettcher, Roderick Long, Ben Rossi, and Michael Young for pressing the objections that led to this post. No, I haven’t (yet) done justice to all of your objections. But I’m not done yet!