Continued from part 1. In my last post, I suggested that Trump’s recent meeting with Nobel Laureate Nadia Murad focuses our thinking on character-based voting. I take it to be uncontroversial that Trump’s behavior at the meeting was an expression of bad character, but the question is, how relevant is something like that to voting (e.g., voting for Trump in 2020)?

Jason Brennan has argued that character can only be relevant as a proxy for policy. He’s made the further claim in the comments here that “Policy > Character.” I haven’t so much disagreed with these claims as questioned their determinacy or clarity, and see the Trump/Murad meeting as an opportunity for getting some clarity. Doing so means going beyond what Brennan says on the topic (primarily in The Ethics of Voting, Against Democracy, and “Polluting the Polls“), because my point is that he doesn’t say enough to do justice to its complexity. I suggested in part 1 that there were three (really more than three) grades of involvement between character and voting. Here’s the first of the three I intend to discuss.

(1) First grade: character is instrumentally relevant to policy
Suppose we equate “governance” with “policy,” and understand policy in the narrow way familiar to us from ordinary speech. Is Trump’s behavior at the Murad meeting relevant to policy, so understood?

This is a less perspicuous question than a face value reading might suggest. It contains two unclarities, expressible in one question: Is Trump’s behavior at the Murad meeting relevant how to what policy? Explicating the whole of the “what” and the “how” would take far too long to untangle here. So for now, let’s gloss the question as asking: “Is Trump’s behavior at the Murad meeting somehow relevant to America’s policy toward the Near East, or even more narrowly, to Syria and Iraq?” There are two possible interpretations here, depending on how strong or weak an interpretation we give to “somehow.” Think of the weak interpretation as saying, “the meeting is relevant to policy somehow, in some indirect or indeterminate sense,” and the strong interpretation as saying, “the meeting is relevant to policy in some directly, clearly specifiable, determinate, and in the best cases, quantifiable sense.”******

Weak interpretation. If we take “somehow” on a weak interpretation, then Trump’s behavior will inevitably end up being (at least weakly) relevant to America’s policy toward the Near East or Iraq/Syria. Trump’s bad moral behavior in this case arguably reflects culpable ignorance, and the culpable ignorance plausibly reflects epistemic malfeasance indicative of epistemic vice: incuriosity, irresponsibility, bluster, and the like. These vices seem projectible well beyond this particular meeting. The meeting is just one occasion for their exemplification, and amnesiacs aside, we’ve seen plenty of others. So we’re not talking about isolated token behavior but evidence of a character flaw or personality disorder epistemically expressed.

Intuitively, you’d think that foreign policy toward the Near East (or really, any place) requires basic knowledge of the political geography and recent history of the place. It also requires (like all politics, really) a limited knowledge of human psychology, e.g., knowledge of the degree of tact (forget empathy) required to deal with a victim of wartime violence of the sort Murad has suffered.

A person possessing this knowledge knows, for instance, that you don’t ask the survivor of a massacre where her dead relatives are right now. A person morally bereft enough to have to ask such a question would lack the moral sensitivity to process the history of the Near East in the first place, at least in the way required to draw the right moral inferences from it. A person who both lacks basic knowledge of relevant history and (a fortiori) fails to make the right moral inferences from it, is in no position to be dictating policy toward the region. And that understates Trump’s ignorance. Here as elsewhere, he sounds like an undergraduate bluffing his way through Recent Iraqi History 101 after being put on the spot by a very gentle professor who knows that her student has not bought the textbook, much less read it. Of course, when it comes to the average undergraduate, this doesn’t matter that much. It matters when you’re the 21st century’s version of Nero.

More generally, a politician of epistemic virtue has the capacity to integrate what he knows about, say, geography, history, and psychology with the requirements of policy. Since it’s hard to know very much an exotically foreign country or culture in very much detail, perhaps the largest part of the relevant sort of epistemic virtue is knowing the limits of one’s knowledge, and grasping the pitfalls of policy-making under uncertainty.

An epistemically virtuous policy maker knows that even a person as well-educated and well-informed as John Stuart Mill or Robert McNamara can make serious mistakes when it comes to imperial policy. He also knows that it’s no surprise that the likes of Reginald Dyer or LBJ can be fooled by their expertise. So he knows enough to know how ignorant he is, and how much more he would need to know (per impossibile) to implement grand-scale imperial policies X, Y, and Z. And what he needs to know is not just a long list of facts, or even paragraphs of them, but a broader and more intensive sort of knowledge along with the disposition to improve his mastery of it.

The issue is not so much that knowledge dictates policy–I’m guessing it dictates more policy omissions than policy–as that epistemic virtue functions as (a weak but determinate) side-constraint on tempting forms of epistemic hubris. Since the side constraint is always relevant, so are the behaviors and the states of character that make those behaviors counterfactually stable. So a candidate lacking epistemic character lacks a basic credential for office, even if it’s hard to spell out precisely how possession of the relevant trait generates policy outcomes (or avoids bad ones by omission). Anyway, on this view, a person who regularly flouts the side constraint, and/or does it in an egregious way, disqualifies himself as a candidate for an office where possession of the trait is an asset.

That’s one interpretation, anyway.

Strong interpretation. If we take “somehow” on a strong interpretation, however, Trump’s behavior at the Murad meeting will at some point turn out to be irrelevant to policy, depending on how strong an interpretation is involved. To take “somehow” on the strong interpretation is to conceive of epistemic virtue as strongly (as well as clearly, determinately, perhaps measurably) related to policy-making, as though possession of epistemic virtue made for good policy, and its absence made for bad policy.* But (it might be argued) the stronger the interpretation, the less plausible the underlying claim, at least as applied to this particular example. In other words, the closer we come to regarding epistemic virtue as sufficient for making good policy (not just relevant in some relatively vague way), the less plausible the claim.

Put it this way: It could with some plausibility be argued that a chief executive doesn’t need knowledge of the kind that makes for a successful meeting with Nadia Murad, at least if the policy objective concerns how to handle Near East/Iraqi/Syrian policy.** The whole point of being a chief executive is to surround yourself with knowledgeable people, to delegate knowledge-intensive  decisions to them, and to restrict yourself to making the information-poor choice-point decisions that executives are supposed to be so good at making. If knowledge is involved, it’s knowledge of how to pick the knowledgeable.

Yes, the Trump/Murad meeting may look bad and reflect ignorance, even culpable ignorance, but so what? Ignorance is policy irrelevant (says this latter argument). A president could flub the whole meeting, reveal himself a complete ignoramus, and still appoint the right generals, etc., have them kill all the right people, and make all the right deals with all the right powers, etc. to get the right things done. (Obviously, this approach presupposes some controversial things about what counts as success in foreign policy, a complication I can’t address here but that needs addressing. How we conceive the character-policy relation depends on the policy outcomes we have in mind.)

Indeed, someone might argue that the whole problem with the Mills and McNamaras of the world was that they knew too much and over-thought things. Better to go with an epistemically vicious ignoramus like Ronald Reagan, George W. Bush, or Donald Trump than a Jimmy Carter-like egghead. Donald Trump may sound like an ignorant dick, and be one, but politics doesn’t require more of the President of the United States. So epistemic virtue is neither here nor there. If you have it, well, OK. If you lack it, fine. A good executive is good because he has the executive skills to do without epistemic virtue. He has how-to skills, not virtues. ***

So there’s another interpretation.

How do we decide between these two interpretations? Obviously, what we do is consult The Empirical Literature on the Subject. Now, I can’t say that I’ve looked specifically at the empirical literature on this particular issue–“Does culpable ignorance of the kind that Trump displayed at the Murad meeting bear on Near East policy?” But having read a fair bit of the empirical literature on character-based voting and related topics, my working hypothesis is that no single study and no combination of them is likely to give a clear-cut answer to that question, or to any relevant question in the vicinity. At least not in the foreseeable future. The lesson here is: we consult The Empirical Literature if there is one.

If any reader can dig up the studies that point clearly in one direction or the other, I’d be much obliged. (And I’m not being a sarcastic dick here. I really would be obliged.) But the absence of studies confirming one or the other interpretation doesn’t cut it as an argument against character-based voting. That’s just an argument from ignorance. What I’m looking for are studies that positively confirm or disconfirm character based voting’s status as a proxy for policy. I’m skeptical that there are any such studies out there, simply because this is not the kind of topic political scientists are apt to pursue, at least in a direct way. The studies that I’ve read are proxies for a study of the topic of whether character is a proxy for good policy making. They don’t directly conceptualize or answer the relevant question. Maybe they can’t. Maybe they never will.****

The corollary is that if there’s no Empirical Literature to consult, we don’t consult it, and a fortiori don’t bank on it, either.

Now what? I guess one thing we could do is to opt for evidentialist agnosticism: whereof the data do not speak, thereof one must remain silent; and whereof we remain silent, we cannot vote. In other words, since we don’t know which option to take, we take none. The implicit inference here is that since we don’t have social scientific evidence for a connection between character and policy, we act as though there isn’t one.

Another option is to take a willful Jamesian leap into the epistemic darkness, gambling that ignorance of the Trumpian variety must somehow (whether weakly or strongly) be policy relevant in ways social science can’t easily discern but that seem plausible anyway. In other words, on this view, voting on the basis of epistemic character is one relevant factor among the many others that go into the deliberative pot, on the assumption that it’s just pre-scientifically or sub-scientifically plausible that there’s got to be some connection (however indirect or weak) between epistemic virtue and good policy. Even if the connection is insufficient on its own to produce bad policy in every case, it’s normatively relevant once it crosses a threshold–relevant enough at least to disqualify candidates who exemplify mind-blowing epistemic vice a la Trump (or Kamala Harris).

It doesn’t take a lot of specialized knowledge, or even all that much general knowledge, for a voter to disqualify a candidate on the basis of vice, at least if the vice is broadly policy relevant, and either extremely egregious or indefinitely iterated (or both). Think of this as the “Complete Asshole Threshold” (CAT): cross it, and you’re done, electorally speaking.

The Jamesian option is not as quixotic or silly as you might think. It’s all well and good to trumpet the virtues of social science, but the hard truth is that there is no scientific evidence for social-scientific evidentialism about politics. There’s no super-duper study anywhere that says: “For any factual claim C relevant to any normatively significant political deliberation D, there exists a study or set of studies that conclusively confirm or disconfirm C with respect to D, so you’d better consult it.”  Data (meaning, for the grammar nazis out there, data-intensive information) exists because someone has the incentive to collect it. If the incentives don’t favor the collection of normatively significant data, that data doesn’t come into existence, and neither do studies based on it. There’s no reason to think that political scientists invariably conduct studies that track precisely those issues that philosophers regard as having the greatest normative importance. They certainly don’t when it comes to character based voting.

It defies instrumental rationality to insist that unless you have confirmed scientific findings in hand, you’re obliged to bring life crashing to a halt until you do. That strategy would induce paralysis, which we avoid in non-political life, if possible. In some parts of life where lack all the evidence we might want, we adopt versions of the precautionary principle; in other parts, we engage in Popperian conjecture and refutation.

If you doubt this, try to find scientific studies to justify looking both ways before you cross the street, or driving more slowly to reduce the probability of having a collision, or showing that heavy pot smoking will adversely affect an intellectually demanding career. You won’t succeed in finding the studies. But you’d be an idiot to stop looking both ways, to drive faster, or to toke up every night. Lack of scientific data is lamentable, but it doesn’t veto the epistemic right to make educated inferences on non-scientific evidence. It certainly doesn’t on what things in life to avoid. And part of voting is avoiding in the form of exclusion: if one candidate crosses the CAT, whatever you do, you don’t vote for him. If they all do, you don’t vote for any of them. Insofar as they do, you avoid voting for the ones that do. And so on.

So what’s the upshot here? The upshot is that reflection on the Murad meeting suggests that epistemic virtue could be a route to character-based voting, even on a relatively narrow, instrumental understanding of the relation between character and policy, and on a relatively conventional understanding of what “policy” means. I don’t mean to imply that that’s the only possible relation the two things bear, or that it’s the most important one, or whatever. I don’t even mean to imply that epistemic virtue is a sure-shot recipe for defending character-based voting (hence the “could”). I just mean: if you want to consider the plausibility of character-based voting, disqualification on the basis of extreme, iterated epistemic vice is where some of the relevant action is. If you don’t address it, you haven’t done justice to the topic.

Barring the appearance of good social-scientific evidence strongly suggesting a positive or negative relation between character and policy, a great deal turns on armchair plausibility judgments about the possession of various epistemic traits and the outcomes they produce in the wider world. That said, I wouldn’t wait too long for those social-scientific studies to come along, either. Patience may be a virtue, but that would be taking a good thing too far.

In the next installment, I’ll consider the possibility that character is relevant to a conception of governance that is broader than, and not reducible to, “policy” in the conventional sense just discussed. One implication is that if we’re to get clear on the topic, the relevant contrasts here all have to sharpened: “character,” “governance,” “policy,” etc. But one thing at a time.

---

*I put the point in this hedged way to include two possibilities: that epistemic virtue is sufficient for making good policy, and that epistemic virtue is strongly conducive to good policy formation in the sense of strongly probabilizing good deliberation about policy outcomes without by itself being sufficient for the outcome. Unless we front-load the concept of “epistemic virtue” in politically loaded ways, the sufficiency claim is going to be implausible. But we could front-load it: arguably, Aristotle does that with many of the intellectual virtues discussed in Nicomachean Ethics VI.

**I keep wavering between “Near East,” “Iraqi and Syrian,” and “Iraqi” policy because it’s an interesting but distractingly difficult question which description is the right framework to adopt for optimal policy making. Is the ideal policy maker the one who can see Iraq policy in the context of Near East policy, or the one who compartmentalizes Iraq policy by abstracting from the broader issues? Is there such a thing as someone genuinely capable of both?

It’s worth noting that the imperatives of policy making diverge from those of academic careers. Policy makers often integrate across whole academic disciplines and sub-disciplines, claiming policy relevant expertise in whole regions (“Dr. Smith is an expert in Near East policy”). Academics tend to specialize more narrowly, disclaiming knowledge even slightly outside of their official areas of specialization (“Dr. Jones works on intersectionality in Palestinian women living in communities adjacent to the Israeli separation wall”). But these tendencies are driven as much by practical contingencies as by genuinely epistemic considerations, so it’s unclear how to figure out the policy implications of either approach. I can’t resolve this issue here, so I just unapologetically hedge.

***It’s ironic that the more lax interpretation of epistemic virtue yields the one most favorable to Brennan’s thesis. The more we devalue good epistemic character, the better the case against any conception of character-based voting that makes epistemic virtue a part of the character that’s the object of voting. It’s ironic because Brennan wants to disqualify voters on epistemic grounds: they should lose the right to vote because they’re epistemically incompetent. But if epistemic virtue is conducive to policy competence in leaders, then voters who discern that a candidate possesses the relevant sort of virtue should be entitled to vote on that basis (even, arguably, if they’re unable to explain the exact connection between virtue and policy).  (Mutatis mutandis for vice.) They lose that entitlement the less that epistemic virtue matters.

Another complication: “leadership” could be its own character trait, one not eliminated by the interpretation in the text. It’s often treated that way in the political science literature. The problem with “leadership” is that it’s not clear whether leadership is a trait or just a loose bundle of skills. In the latter case, voting for people with good leadership skills is not an instance of character-based voting.

That said, I find Brennan’s discussion of this issue in The Ethics of Voting inadequate. “Still, even if voters are good judges of such political skills and vote accordingly, such skill could mean bad policies will be enacted” (Ethics of Voting, p. 85). I don’t dispute that, but it does nothing to rule out the possibility that possession of the relevant skills could lead to good policies. It also does nothing to rule out a view that holds that possession of the relevant skills is deliberatively relevant to voting without being  sufficient for determining the voter’s vote. Brennan claims that voting on skills is “not obviously”  a reliable way of generating good policy outcomes. It may not be, but it may still be a normatively relevant consideration when voting for a candidate.

Brennan’s discussion of character-based voting attacks two targets: that (a) it’s justified to treat character as a normatively relevant consideration of any significant weight in voting, and (b) that it’s justified to make character the sole basis of one’s vote for a candidate. Whether intended or not, claim (a) ends up collateral damage of Brennan’s attack on (b). But (a) is a lot more plausible than (b). The conflation may arise from an ambiguity in something’s being X-based. In one sense, “X-based” means “based exclusively and asymmetrically on X”; in another, “X-based” means “having some normative basis in X, but of unspecified weight.”

****I’m baffled by Brennan’s citation of the empirical literature on this subject, particularly in The Ethics of Voting. He cites two sources to make the strong claim that “[t]o a significant degree, voting for character is voting for the wrong reasons” (p. 84n.22): Bernard Manin’s Principles of Representative Government, and Russell Hardin’s How Do You Know?.  As far as I can see, neither source is relevant to Brennan’s claim, much less bears it out. And neither author offers social scientific evidence to clinch the case against character-based voting, or even claims to. I’m traveling, and don’t happen to have any of the three texts here with me, but can address this when I get back.

*****PoT blogger Hendrik Van Den Berg has written very cogently on the limitations of conventional social science methodology. Though his work is mostly on economics, it has implications for political science, as well.  For starters, take a look at this paper, as well as this one. Van Den Berg aside, many of the criticisms made here remain relevant as well, and here, too. (Though Hendrik has never actually blogged for us, or even accepted the invitation I sent him to blog for us, he once said he would blog for us; hence I only slightly flout the data by calling him a PoT blogger.)

******Added later, August 8: On re-reading this, it occurs to me that I put these claims less clearly than I should have. I mean the strong and weak interpretations to be criterial, not merely descriptive. In other words, what I mean is: to count as satisfying the weak interpretation, the meeting need merely be relevant to policy somehow, in some indirect or indeterminate sense. Whereas to count as satisfying the strong one, it has to be relevant to policy in some directly, clearly specifiable, determinate, and in the best cases, quantifiable sense. The weakest versions of the weak interpretation will be too weak to make the meeting relevant to policy, but the strongest versions of the strong interpretation may well be too strong to be plausible, unless “front loaded.”

Thanks to Michael Young and Michael Brown for some helpful input on this post.