In their paper arguing that ‘educational’ solutions to fake news are superior to ‘structural’ solutions, Michel Croce and Tommaso Piazza (2021) challenge my earlier (Rini 2017) claim that the spread of fake news results (partly) from an individually reasonable practice I call “epistemic partisanship”. Epistemic partisanship happens when you are more willing to trust a person’s testimony because they share your political values. Croce and Piazza (henceforth C&P) dispute my claim that partisanship in testimony reception is epistemically defensible. This leads them to conclude that fixing fake news requires correcting individual poor performance rather than making structural changes to the shared informational environment … [please read below the rest of the article].

Image credit: outtacontext via Flickr / Creative Commons

My reply has three parts. First, I clarify the relationship between epistemic virtue and partisanship; C&P have misread me on this point. Second, I try to draw out where and why we actually disagree. I think C&P are trying to apply ideal normativity—what would a perfectly rational agent do—to the problem of fake news, whereas I approach the issue through non-ideal epistemic theory. Third, I consider how this diagnostic disagreement informs our differing prescriptions and reaffirm the unique value of a structural solution to fake news.

The (Individual) Epistemic Virtue Consistency of Partisanship

I wrote my original paper five years ago for a special issue of the Kennedy Institute of Ethics Journal concerning the 2016 US presidential election. I deliberately avoided jargon and excessive formality, since I expected the topic would have an audience beyond academic philosophy. For that reason, I wrote in a way that might be called insufficiently defensive for an audience of analytic philosophers.

In particular, I did not explicitly distinguish two ideas—being consistent with virtue and constituting virtue—that make a big difference to my claims about partisanship in testimony reception. So let me be clear now: I think that partisanship is consistent with epistemic virtue. That is, a person whose epistemic practices generally tend toward promoting epistemic value may exhibit partisanship-in-testimony-reception among those practices. I do not think that partisanship is constitutive of epistemic virtue. That is, I do not think that exhibiting partisanship in testimony reception is among the things that makes a person count as epistemically virtuous, such that everyone ought to aim at exhibiting it.

C&P appear to attribute the constitutive view to me. They say that I think social media users who exhibit epistemic partisanship “can be praised as intellectually virtuous” (C&P 2021, 3) and also that on my view, “fake news consumption… may be underwritten by the exercise of an intellectual virtue that she calls ‘epistemic partisanship’.” (C&P 2021, 2) But C&P are wrong to attribute this constitutive view to me; as I said, on my view, epistemic partisanship is merely consistent with intellectual virtue.

To be fair to C&P, in my original paper I did slip a couple times (including in a regrettable section heading) into locutions like “the (individual) epistemic virtue of partisanship”, which does connote a constitutive view if taken too literally. But I meant this as elliptical for the ugly-long phrase “consistent with epistemic virtue” that I more typically used throughout the paper. If nothing else, I suppose this is a lesson in the perils of deploying variable phrasing for aesthetic elegance in philosophy writing.

Apart from scholarly niceties, does any of this matter? Well, yes. The difference between consistency with virtue and constitution of virtue is very important to how we address socially harmful practices like fake news consumption. To see this, consider a parallel to the problem of climate change.

I currently own a car that runs on fossil fuels, even though an electric car would contribute much less to harming future generations through climate change. I chose a fossil-fueled car because I live in an old apartment building where the parking garage does not have any electric hookups, and because electric car batteries aren’t yet good enough for a long intercity drive on a single charge. I do most of my day-to-day commuting on foot or public transit, but I own a car specifically to make spontaneous visits to distant family. So I will continue to buy carbon-burning cars until battery technology and charging infrastructure improve.

What should we say about my choice to own a fossil-fueled car? Is it morally virtuous? That depends on whether we are speaking of constituting virtue or being consistent with virtue. Clearly my choice is not constitutive of moral virtue. No one would say that it “is a virtue” to prefer a more polluting car. No one ought to praise me for this choice.

But it is harder to answer whether my choice is consistent with moral virtue. In principle, I could make a better choice: I could own an electric car if I moved to a newer, more expensive home and sacrificed most opportunities to see my family. I could even own no car at all. But those alternatives come with serious personal costs, which I regard as outweighing the fractional contribution my driving makes to global environmental harm. Is this decision consistent with my being a generally morally virtuous person? I’d like to hope so, and I am at least confident that this is open to argument, whereas regarding my choice as constitutive of virtue is not even minimally plausible.

These same points apply to my discussion of partisanship and epistemic virtue. Obviously partisanship is not constitutive of epistemic virtue; C&P are simply wrong when they suggest that I regard it as “praiseworthy”. But I do think that a generally epistemically virtuous person can frequently engage in epistemic partisanship, given the informational environment we live in, just as a generally morally virtuous person might not yet replace their carbon-burning car, given the energy infrastructure we currently have.

And that leads to my substantive disagreement with C&P. They frame our disagreement as whether partisanship in testimony reception is “epistemically criticisable conduct” (C&P 2021, 2). They think that epistemic partisanship is criticisable and that I think otherwise. Partly this is because they misinterpret me as speaking constitutively of epistemic virtue (obviously if something is constitutive of virtue then it ought not be criticized). But they are right that I think people often do not deserve to be individually criticized for falling for fake news due to epistemic partisanship, just not for the reason they suggest. To understand this disagreement better, we need to think about criticism and non-ideal normative theory.

Testimonial Dynamics and Non-Ideal Epistemology

Does a person deserve criticism for displaying partisanship in receptivity to testimony? That is, if a social media user is more willing to accept a testimonial assertion about a factual political matter when it comes from a co-partisan than from some other testimonial source, is that person liable to epistemic criticism? Croce and Piazza think so: they say sharing political commitments provides “no reason” to infer greater likelihood of the truth of a co-partisan’s testimony (C&P 2021, 4).

Interestingly, C&P don’t provide any argument for this claim, which is really the crux of the debate. They seem to think that I, along with the benighted practitioner of epistemic partisanship, have simply confused the testimonial significance of normative claims (e.g. it’s wrong not to pay your taxes) with the testimonial significance of descriptive claims (e.g. this candidate didn’t pay her taxes). Once we carefully distinguish the two, it’s meant to be just plain obvious that the latter shouldn’t be the target of epistemic partisanship.

What’s odd about this style of argument is that my original paper does explicitly distinguish normative and descriptive claims in exactly this way, and I explicitly argue that even descriptive claims can be the target of reasonable epistemic partisanship (Rini 2017, E-50-53). I can’t recount that entire argument here, but the gist of it is this:

(1) We have a default entitlement to trust the testimony of others (otherwise we’d know very little about the world).

(2) This default reliance on testimony is rationally conditioned on sensitivity to situation-specific indicators that certain testifiers are incompetent or dishonest.

(3) Social media testimony is one such situation, given its preponderance of something I call ‘bent testimony’: the unreliability of usual norms for holding people accountable for spreading false information.

(4) However, co-partisanship, as an indicator of shared values, can make it the case that a person is reasonable in selectively ignoring the problem of bent testimony while evaluating social media testimony from some co-partisans.

That’s the argument. Of course, maybe there’s a problem that I haven’t noticed. But what’s odd is that C&P don’t even acknowledge this argument in their critique of my position. They don’t mention the concept of ‘bent testimony’, which is the dialectical hinge for my view. Instead they attribute to me a “general maxim” to the effect that one ought to proportion the evidential weight of a piece of testimony to one’s normative proximity to the testifier (C&P 2021, 4). But I don’t say anything about such a maxim, and in fact I don’t accept it. So why do C&P attribute this confected maxim to me, rather than discuss the argument I actually made?

The answer, I think, is that C&P want to do ideal normative theory, and I don’t. In epistemology, ideal theory aims to identify the criteria for changes in belief states that would be employed by a perfectly rational agent. Wherever actual human practice deviates from this ideal, it is thereby liable to epistemic criticism.

I think epistemic ideal theory has a certain limited range of value. It’s useful when we want to abstractly analyze the meaning of concepts like justification and evidence. Presumably that’s why it gained so much influence in 20^th century analytic philosophy, when philosophical method focused primarily on the semantic properties of concepts. But when it comes to real world phenomena, and especially to making prescriptions to guide actual humans through fraught social circumstances, I think that this approach rises only marginally above uselessness.

Real humans operate under an enormous range of doxastic constraints. Some of these constraints are inescapable, like human cognitive limitations or the amount of time a temporally finite being can spend reasoning. Others are contingent constraints of technology and social background that can be changed and improved, though not by any one agent alone.

In principle, we can adapt ideal theory to address complex reality by imaginatively grafting a series of constraints onto epistemic angels, iteratively building a prescriptive matrix for hypothetically limited agents which approximate real humans. That is certainly the approach taken by many epistemologists in recent decades. I think C&P have interpreted me as trying (obscurely, confusedly) to do this as well, which is why their paper offers me the helping hand of a top-down ideal normative maxim that I neither expressed nor accept.

But that’s just not my project here. As I said in the original paper:

[G]iven the realities of human psychology and politics, certain forms of epistemic partisanship are individually reasonable in the world as we actually confront it. This would not be the case in an ideal world, but that is not where we live. In effect, I am defending a form of non-ideal political–epistemic theory. Accordingly, I will argue that our normative focus should be on identifying realistic structural changes, rather than specifying idealized individual practice (Rini 2017, E-50).

Non-ideal epistemic theory begins by describing our actual social epistemic practice, then identifies undesirable features of that practice, and finally sketches realistic changes in the contingent epistemic environment that might lead to better practice. There’s no assumption that we can identify the top-down principles that would guide an epistemic angel, only that we can recognize local, relative practical improvements. (This approach, of course, is modelled on non-ideal theory in political philosophy, most especially the work of Charles Mills (2005). For related methodological points about the practice of science and ethics, see Kitcher (1993) and Moody-Adams (1999).)

If there’s any area of epistemology that most demands a non-ideal analysis, it has to be testimony. Testimony reception is an intrinsically social process; it is a category error to treat the evaluation of testimony as on par with any non-social evidentiary source (see Moran 2005 on this point). In testimony reception, a range of epistemic and social factors converge to produce fraught trade-offs. Accepting another’s testimony amounts to entrusting them with one’s epistemic well-being (see Baeir 1986); refusing to accept testimony is often rightly perceived as an insult (Hazlett 2017). Shared testimonial practices can also go awry in a variety of notoriously unjust ways (Dotson 2011, Davis 2016, Tanesini 2016).

All of this is why I framed my non-ideal analysis of fake news by first diagnosing the specific testimonial dynamics of social media. Epistemic partisanship is reasonable in the context of bent testimony, which is itself a localized deformation of default social testimonial practices. Perhaps none of these dynamics are even legible from the point of view of the epistemic angels.

Which gets us back to why I disagree with C&P about whether epistemic partisanship is criticizeable. If our standard for epistemic criticism is deviation from the practice of angels, then of course epistemic partisanship will come out worthy of criticism. But if we are trying to understand the real-world options of people with limited time to survey a hostile epistemic environment, I think we do better to acknowledge they are usually reasonable to let partisanship inform their testimonial practices. Which means that if we want to fix fake news, we should be looking to change the epistemic environment.

Social Epistemic Infrastructure

Sometimes I worry that philosophers have a professionally distorted view of most people’s epistemic lives. We are paid to spend large parts of our days thinking about thinking, and most of us enjoy doing that. From our point of view, it seems natural to suppose that people who fall for deceptive media just aren’t putting in enough work. After all, we have the time and resources to invest trying out alternative news sources or developing higher-order tests for reliability. We can easily miss that a person who spends most of every day working the register or caring for kids just doesn’t have that many leisure hours left over to do a meta-analysis of available media. Of course, some people will choose to spend their free time this way, and they deserve praise for their civic diligence. But most will not. If our epistemic expectations demand it from them, they will correctly roll their eyes.

This is why I am skeptical of Croce and Piazza’s view that the problem of fake news is best addressed through educational means, such as “the strategy of having social media users widen their sources of information, at least by consuming news from traditional media such as newspapers and magazines” (C&P 2021, 6). Now, to be clear, I think that is good advice. It’s certainly better to consume a diverse media diet than a narrow one. But I don’t think that this suggestion will accomplish very much on its own.

We can see the problem once we recognize that C&P demand more than just consuming traditional media alongside social media. It turns out people must consume only the right traditional media. In particular, C&P declare that “irresponsible journalism” such as Fox News does not “belong within the same broad category of mainstream media as more balanced news outlets” (C&P 2021, 7). But Fox News has been the most watched cable news channel in the United States for years (Joyella 2021). If Fox News does not count as ‘mainstream’, then C&P’s seemingly low-cost recommendation to consume mainstream news turns out to be a much more taxing demand. To avoid epistemic criticism, people now must do the work to evaluate every potential news source—every social media feed, every cable channel, every newspaper—on a case-by-case basis. Doing this properly would consume enormous amounts of time and energy.

Again, I agree that it is a good thing for people to become more critical media consumers. But is this something we can reasonably demand? And is this likely to be an effective policy response? It seems to me that C&P’s proposal for combatting social media deception will only work if they can first convince millions of Fox News viewers to trust the testimony of elite professors over that of Tucker Carlson. In other words, I don’t see it happening.

That’s why I favor structural approaches. Social media platforms need to be re-designed to make more thoughtful media consumption easier. I talk about specific strategies for doing this in my original paper, so won’t rehash them here. But the key idea is to allow people to offload some of the cognitive labor of tracking epistemic reputation onto digital infrastructure. We can rebuild old norms for holding liars and fools accountable if we have better tools fitted to the vast digital public square.

Think once again about my carbon-burning car. If you want me to get rid of it, you shouldn’t try to ‘educate’ me. I’m already very aware that my choice is morally sub-optimal. But I’ve made the rational calculation that I can’t afford the personal cost to switch over. On the other hand, if you invent a new electric battery that can take me as far as a tank of gas, and if you install charging stations everywhere, then I’ll start saving up for my electric car right away. Reduce my personal costs through smart structural changes and you enable me to be as virtuous as I’d like to be. And that applies to our epistemic environment just as much as to our physical environment.

Acknowledgements

Thanks to Michel Croce and Tommaso Piazza for writing a provocative paper that has spurred me to clarify the theoretical apparatus behind my earlier work. Thanks also to Andrew Buzzell for many helpful conversations about the collective responsibility for managing our epistemic environment.

References

Baier, Annette. 1986. “Trust and Antitrust.” Ethics 96 (2): 231-260.

Croce, Michel and Tommaso Piazza. 2021. “Consuming Fake News: Can We Do Any Better?” Social Epistemology doi: 10.1080/02691728.2021.1949643.

Davis, Emmalon. 2016. “Typecasts, Tokens, and Spokespersons: A Case for Credibility Excess as Testimonial Injustice.” Hypatia 31 (3): 485-501.

Dotson, Kristie. 2011. “Tracking Epistemic Violence, Tracking Practices of Silencing.” Hypatia 26 (2): 236-257.

Hazlett, Allan. 2017. “On the Special Insult of Refusing Testimony.” Philosophical Explorations 20 (sup1): 37-51.

Joyella, Mark. 2021. “CNN Drops 68% In Prime As Fox News Channel Crushes Competition In Q2 Cable News Ratings.” Forbes.com 29 June 2021. https://www.forbes.com/sites/markjoyella/2021/06/29/cnn-drops-68-in-prime-as-fox-news-channel-crushes-competition-in-2q-cable-news-ratings/

Kitcher, Philip. 1993. “Knowledge, Society, and History.” Canadian Journal of Philosophy 23 (2): 155-177.

Mills, Charles. 2005. “‘Ideal Theory’ as Ideology.” Hypatia 20 (3): 165-183.

Moody-Adams, Michelle. 1999. “The Idea of Moral Progress.” Metaphilosophy 30 (3): 168-185.

Moran, Richard. 2005. “Getting Told and Being Believed.” Philosophers’ Imprint 5: 1-29.

Rini, Regina. 2017. “Fake News and Partisan Epistemology.” Kennedy Institute of Ethics Journal 27 (S2): 43-64.

Tanesini, Alessandra. 2016. “‘Calm Down, Dear’: Intellectual Arrogance, Silencing and Ignorance.” Aristotelian Society Supplementary Volume 90c(1): 71-92.

Categories: Critical Replies

Tags: deception, fake news, Michel Croce, non-ideal epistemology, Regina Rini, social media, Testimony, Tommaso Piazza