Back to Prindle Institute

Can We Trust Anonymous Sources?

photograph of two silhouettes sitting down for an interview

Ben Smith’s recent article in The New York Times about Tucker Carlson’s cozy relationship with the media has caused quite a stir among media-watchers. It turns out that the man who calls the media the “Praetorian Guard for the ruling class” loves to anonymously dish to journalists about his right-wing contacts.

Missing from this discussion about Carlson’s role in the media ecosphere, however, is any exploration of the philosophically rich issue of anonymous sources. Is the practice of using such sources defensible, either from a moral or an epistemic point of view?

First, there is an issue of terminology. A truly anonymous source would be something like a phone tip, where the source remains unknown even to the journalist. In most cases, however, the identity of a source is known. These sources are not truly anonymous, but could be called “unnamed” or “confidential.” For reasons that will become apparent shortly, it is never appropriate for journalists to publish information from truly anonymous sources unless the information is capable of being independently verified, in which case there is no need to use the anonymous source in the first place. When I talk about “anonymous” sources in this column, I am referring to confidential or unnamed sources.

The basic epistemic problem with confidential sources can be summed up as follows: we really can’t assess the truth of a person’s testimony without knowing who the person is. If a shabbily-dressed stranger shuffles up to you and tells you that JFK was the victim of a conspiracy, you’re likely to discount the testimony quite a bit. On the other hand, if the head of the CIA came out and made the same claim, you’d be likely to update your beliefs about JFK’s assassination. In short, many details about a person’s identity are relevant to the reliability of their testimony. Thus, without access to these facts, it’s almost impossible to know whether the testimony is, indeed, true. But in the case of anonymous sources, the public lacks the necessary data to make these judgments. So, we are in a poor position to determine the veracity of the source’s claims. And if we can’t assess the reliability of the testimony, then we aren’t justified in relying upon it.

This epistemic trouble can often become a moral problem. Anonymous sourcing can encourage people to believe that a source’s claim is more reliable than it is, and in this way it may mislead. But surely, journalists have a moral obligation to take every precaution to guard against this. One example of the way anonymous sourcing can mislead is the anonymous essay published by The New York Times in September 2018 purporting to be written by a “senior official” within the Trump administration. This essay caused many people to believe that a cabinet-level official was helming a resistance to Trump from within the White House, but it turned out that the writer was Miles Taylor, former chief of staff to Department of Homeland Security Secretary Kirstjen Nielsen. There is a case to be made that the Times misled the public in that case, causing them to hope in vain that some sort of resistance to Trump was taking place in the upper echelons of the executive branch.

How should we go about solving this problem? How is the public to distinguish between the straight scoop and unsubstantiated rumor? How can we mitigate the harm that comes with directing public attention at a shaky story without losing the ability to speak truth to power?

Reporters’ primary answer to the problem of anonymous sourcing is to point to the reliability not of the source, but of the news publication. Call this the “vouching” solution. The reporter claims that people should believe an anonymous source because the reporter’s institution does; the source’s trustworthiness is a function of the trustworthiness of the publication. But this is like saying that you can justifiably rely on the shabbily-dressed stranger’s claim that JFK was the victim of a conspiracy because an honorable friend reports it to you, and you trust your friend to vet the stranger’s claim before presenting it. The trouble with this solution is that if we’re dealing with a truly anonymous source, our “honorable friend” – the news publication – lacks the necessary information to properly vet, and thus adequately vouch for, the stranger and their claims.

That our faith in news outlets justifies the use of unnamed or confidential sources is just one reason why it is so important for the news media to cultivate public trust. Unfortunately, however, people’s confidence in the mainstream media is at an all-time low. According to one recent poll, a majority of Americans do not have trust in traditional media. For these Americans, the vouching solution fails to even get off the ground. Moreover, for these Americans, it would arguably be irrational for them to rely on the media’s anonymous sources, given their skepticism. If one does not trust one’s friend, it would be foolish to rely on the sources for which one’s friend vouches. By the same token, if one does not trust the media, it would be irrational to rely on the anonymous sources for which the media vouches.

What does journalistic vetting of anonymous sources involve? One thing it does not entail is securing independent verification of an anonymous source’s information. If this were possible, then it would be unnecessary to grant a source confidentiality at all — journalists could just settle for the independent evidence. Thus, journalistic vetting usually involves scrutinizing the motives and behavior of the source. Is the source eager or reluctant to share information? Is she in a position of power or vulnerability? What is her agenda?

Which brings us back to Carlson, who seems like a signally poor candidate for confidentiality. Smith’s article makes clear that Carlson likes to portray himself in a flattering light to reporters, and that he is eager to share information. He is also, of course, in a position of great power and influence, and surely uses his effusions strategically to further his own agenda. For these reasons, using Carlson as a confidential source seems to be an epistemically and, because of the potential for misleading the public, ethically dubious practice.

When Is It Rational to Trust a Stranger?

This post originally appeared October 13, 2015.

A father hands over the keys to his house to a stranger, his children fast asleep upstairs. Two grandparents share their living room with a traveling salesman in town for the week. A young woman falls asleep in the guest room of a man she has never met before that night. While such scenarios may sound like the beginning of a horror film, it is now a fact that millions of individuals in over one-hundred-and-ninety countries rely on online services to rent lodgings, most often in private homes. The broader sharing economy encompasses, among other things, the sharing of workspace, the sharing of automobiles, and even the sharing of home-cooked meals. In some cases what is shared is publicly owned, such as in most bicycle sharing schemes. But typically one party owns what is shared in an exchange between strangers.

All this cooperative activity between strangers is taking place in an age when parents feel the need to install “nanny-cams” in their children’s rooms, companies monitor their employee’s web surfing, and a proliferation of online services allow anyone to order a background check on anyone else. Do these apparently divergent cultural trends point to a more fundamental polarization of values? Or do they simply represent differential responses to varying social circumstances?

To the skeptic, the trustful enthusiasm of the sharing economy is a symptom of naïveté. The only cure is a painful experience with cynical breach of faith. Recent cases like the alleged sexual assault of an airbnb guest are the canaries in the coalmine. To the optimist, these sensational cases are remarkable precisely because of their paucity. What the amazingly rapid growth of the sharing economy teaches us is that human beings, in aggregate, are much more trustworthy than previously imagined.

I think that both the skeptic and the optimist have got it wrong. On the one hand, it’s silly to think that involvement in the sharing economy confers upon its participants the esteemed moral character trait of trustworthiness. On the other hand, the trusting attitudes manifested in many corners of the sharing economy are both rational and prudent, under the right conditions.

Borrowing a term from Princeton philosopher Philip Pettit, I will refer to these as the conditions for trust-responsiveness. In a paper delightfully entitled “The Cunning of Trust”, Pettit makes the case that you can have reason to trust others even if you have no antecedent knowledge about their reliability. This is because you can make them responsive to your trust simply by communicating that you trust them. This might seem like pulling a rabbit out of hat. But on reflection, the dynamic is not unfamiliar.

Pettit’s analysis rests on a relatively uncontroversial psychological claim: human beings care very much about their standing in the eyes of others, and are often moved by love of regard. In his Theory of Moral Sentiments, Adam Smith, the father of modern economics and a keen psychologist, went so far as to say that Nature has endowed Man “with an original desire to please, and an original aversion to offend his brethren. She taught him to feel pleasure in their favorable, and pain in their unfavorable regard. She rendered their approbation most flattering and most agreeable to him for its own sake; and their disapprobation most mortifying and offensive.”

What other people think of us matters to us a great deal. In particular, a reputation for trustworthiness has both intrinsic and instrumental value. Pettit notices that when people decide to rely on us, they signal to others that they regard us as trustworthy. We are motivated to be responsive to their trust because we want that signal to be broadcast. As Pettit puts it, “The act of trust will communicate in the most credible currency available to human beings – in the gold currency of action, not the paper money of words – that the trustor believes the trustee to be truly trustworthy, or is prepared to act on the presumption that he is.”

When a person is widely perceived to be trusted, he or she gains a highly valuable status. When we manifest trusting reliance, we give the person we rely on an incentive to do the very thing we rely on them to do, because they want to cultivate and maintain that status. This is why the trust of strangers can be a rational gamble. It is never a sure bet, but it is a good bet more often than one might imagine. And this is why the skeptic is wrong. The dramatic growth of the sharing economy is predicated on fundamental facts about of human psychology.

But the optimist gets something wrong as well. There is no necessary connection between ubiquitous sharing and the dawn of a new age of trustworthiness. Trust-responsiveness and trustworthiness are altogether different animals. A trustworthy person will do what he or she is trusted to do regardless of whether anyone else is watching. This is why we hold trustworthy people in esteem and think that trustworthiness is a morally desirable trait of character. In contrast, trust-responsiveness is predicated on a desire for good opinion and is therefore, at best, morally neutral. Moreover, trust-responsiveness will only survive under certain institutional conditions.

It’s worth noting that these conditions exist par excellence in many corners of the sharing economy. The oxygen in which this economy exists is the collection and dissemination of reviews. On airbnb, for example, hosts who meet certain criteria of responsiveness, commitment, and five-star reviews are granted the coveted status of “superhost” which is signified by a red and yellow badge of approval on their profile. This status may increase demand for booking, thereby providing a financial incentive to hosts looking to juice their profits. It also works because it flatters people who self-identify as open, warm, and hospitable.

But we shouldn’t be too cynical about all this. Aristotle noticed that moral virtue could be acquired by cultivating good habits. It may be that exercise of the dispositions of trust-responsiveness can help cultivate the morally desirable trait of genuine trustworthiness. Maybe. I think the jury is still out on that one.

Our judgments about whether to expose ourselves to the hazard of trust are influenced both by our beliefs as by arational factors. Sometimes we just have a bad feeling about someone – we don’t like the cut of their jib. These kinds of knee-jerk responses can be wiser than our reflective selves, which are prone to rationalization. But just as often our “intuitive” judgements reflect unexamined biases. A 2014 Harvard Business School study found that “non-black hosts are able to charge approximately 12% more than black hosts, holding location, rental characteristics, and quality constant. Moreover, black hosts receive a larger price penalty for having a poor location score relative to non-black hosts.” Clearly we have a long way to go in learning how to trust well and without prejudice.

My own family rents out a garden-level apartment in our house on airbnb. We’ve met many interesting people, including one guest who eventually became a co-author of mine. And the money we earn helps to pay a hefty daycare bill. When we tell our friends and family that we have lent our keys to scores of people, they sometimes respond with horror and disbelief. And to be honest, in some frames of mind, we feel pretty nervous ourselves. But overall I think we are making a rational bet, and not one that presupposes a Pollyannaish faith in humanity. Of course, a truly malicious person can always rack up sterling reviews with the express purpose of lowering his victim’s defenses. But this kind of evil, like moral virtue, is rare.

My other work on trust and promises can be here.