The emptiness at the core of conspiracist thinking

by Joseph Shieber

Recently I was reading one of Scott Alexander’s posts about fake news and conspiracist thinking. In that post he introduces what he dubs the “North Dakota Constant”.

Alexander references a survey conducted by researchers at Chapman University, and mentioned a “control question” that the researchers included in the survey.

Here’s how the pollsters at Chapman describe that question and the responses it prompted:

Perhaps most indicative of the conspiratorial nature of Americans is the …one which, to our knowledge, we created.

Respondents to the Chapman University Survey of American Fears were asked if “The government is concealing what they know about…the North Dakota crash.” A third of Americans (33%) think the government is concealing information about this invented event.

Were the North Dakota crash added to the ranked list of conspiracies (see above), this invention would rank as number six, just under plans for a one world government.

What Alexander concludes from this is that there is a large minority of the country — the 33% willing to buy in to a conspiracy about a “North Dakota crash” that never existed — who are disposed to believe in ANY conspiracy.

Alexander suggests that the existence of the “North Dakota Constant” should make us more cautious in overemphasizing the role of “fake news” in causing conspiracist thinking. His idea is that if there is a floor of over 30% of the population disposed to believe in a made-up conspiracy, the fact that 30% of the public believe that President Obama wasn’t born in the United States is not in fact evidence of a very strong “fake news” effect. That is because, if the “North Dakota Constant” is compelling, we would expect around 30% of the public to believe ANY conspiracy about which pollsters questioned them.

The “North Dakota Constant” may well involve a genuine phenomenon. But what struck me most strongly upon initially reading about the poll Alexander cites was that it’s terribly designed. That design flaw points to an interesting underlying complication raised by discussions of conspiracy theories.

To see what I mean, consider the three versions of this exchange. It concerns salacious gossip about the D-list actor Handsome “Handy” McChiseledchin:

1 – “Did you hear about Handsome McChiseledchin? I hear that he killed his first girlfriend!”
2 – “Really?! I don’t believe it!”

1′ – “Did you hear about Handsome McChiseledchin? I hear that he killed his first girlfriend!”
2′ – “I didn’t hear that. Where did you hear it?”

1” – “Did you hear about Handsome McChiseledchin? I hear that he killed his first girlfriend!”
2” – “I happen to know that Handsome McChiseledchin has always been open about his sexual orientation, and that he’s never had a girlfriend.”

It seems to me that these three exchanges model three different ways of dealing with suspect information. Focusing on the differences between the three strategies can provide some valuable lessons for dealing with suspect information. In particular, I’ll want to look at the third version of the exchange, but it will be useful briefly to discuss the first two as well.

The first version, which I’ll term “Rejection”, involves denying the questionable information. This response is certainly sometimes called for. For example, if someone claims that illegal immigrants pose a threat of violent crime, and you know that statistics show that illegal immigrants are actually LESS likely to commit violent crimes than members of the general public, then it is perfectly appropriate to reject the inaccurate claim.

The second version, which I’ll term “Suspending”, involves suspending judgment between accepting or rejecting the claim — although it’s compatible with leaning more in one direction or the other.

In the version of the exchange above, this suspension of judgment is coupled with a request for more information about the quality of the source of the claim that speaker 1′ makes about Handsome “Handy” McChiseledchin.

What this second version of the exchange underscores is that there IS a third option between the two options available in the first version of the exchange.

Rather than merely believing or disbelieving a claim, it is also possible to refrain from making a judgment either way. If you choose this option, then you withhold judgment — either permanently, if you don’t really care about the claim or not much is riding on it for you, or until you have more or better information to form a judgment.

We’ll return to the option of withholding judgment later. But it’s worthwhile noting one exemplary case of withholding judgment that played out publicly in recent weeks.

The Vox reporter Laura McGann wrote an excellent, subtle, and detailed account of her attempt to weigh the evidence around Tara Reid’s claims that she was sexually assaulted by Joe Biden while working as a staffer in his Senate office in the early 1990’s.

McGann’s personal conclusion is a paradigm example of withholding judgment:

All of this leaves me where no reporter wants to be: mired in the miasma of uncertainty. I wanted to believe Reade when she first came to me, and I worked hard to find the evidence to make certain others would believe her, too. I couldn’t find it. None of that means Reade is lying, but it leaves us in the limbo of Me Too: a story that may be true but that we can’t prove.

That wonderful turn of phrase, “mired in the miasma of uncertainty”, reveals a reason why we so often ignore the option modeled by the second version of the exchange about “Handy” McChiseledchin. It’s so much more COMFORTABLE to have arrived at a settled opinion about a claim.

We WANT to believe or disbelieve; we don’t want to have to say, “I don’t know” or “I’m just not sure”. That’s why the second option — withholding judgment — doesn’t immediately spring to mind when we think about evaluating the latest piece of internet gossip that floats to the surface of our social media feeds.

Although the withholding option is so much more difficult than the belief/disbelief option, neither of the options modeled by the first two versions of the exchange about “Handy” McChiseledchin is that difficult to grasp. The third version, however, which I’ll term “Canceling”, is likely the hardest of the three to explain.

To begin to do so, I want to speak for a moment about one of the classic discussions in 20th century analytic philosophy of language. It has to do with how to make sense of “empty” definite descriptions. To do THAT, however, let’s first talk a little about some technical terms.

The first term is “definite description”. A definite description is just a description using the definite article, “the”. It indicates that there is exactly one person, place, or thing fitting the description.

So if I say, “The inventor of the calculus”, I’m indicating that I take to be the case that exactly one person invented calculus.

As this example shows, employing a definite description isn’t always appropriate. If you believe that both Newton and Leibniz discovered calculus independently and at roughly the same time, then you’ll think that it isn’t fitting to talk of “THE inventor of the calculus.”

So one reason why it can be inappropriate to employ a definite description is because there is more than one person, place, or thing fitting the description. But there is another reason why using a definite description could be inappropriate. It could also be the case that NOTHING fits the description. In that case, we say that the definite description is “empty”.

One of the most famous examples of an empty definite description comes from Bertrand Russell’s discussion of definite descriptions.

Russell considers the sentence, “The present King of France is bald.” In that sentence, the phrase “The present King of France” involves a definite description. Since NOBODY fits that description, the phrase is a case of an EMPTY definite description.

According to Russell, although the surface structure of the sentence, “The present King of France is bald” seems quite simply, the underlying LOGICAL structure is actually more complex. Russell suggests that we analyze that sentence as actually involving two claims: There is exactly one x such that x is the present King of France AND x is bald.

On Russell’s analysis, the claim “The present King of France is bald” turns out to be FALSE — because the first of the underlying claims in the analysis (There is exactly one x such that x is the present King of France) is false. There is no such x such that it is the present King of France!

There are some very odd consequences of Russell’s analysis. Suppose, for example, that someone hears the claim, “The present King of France is bald” and, in response, exclaims, “No he isn’t; the present King of France has a thick and glorious head of hair!” According to Russell, we would have to say that BOTH of the disputants about the hirsuteness of the present King of France are saying something false.

That is, according to Russell, both “The present King of France is bald” and “The present King of France is not bald” are false. This SEEMS to contradict the logical Law of the Excluded Middle, according to which it must be the case that, for any proposition, either that proposition or its negation is true. (Russell, of course, would stress that it only SEEMS to do so, because “The present King of France is bald” actually involves, on his analysis, two claims.)

This may all seem too abstruse, so let’s bring it back to the present day. Recall that control question in the Chapman University conspiracies poll. The pollsters only gave respondents two options for each question: they could either agree or disagree.

Now, as we mentioned at the outset, 33% of the respondents agreed with the claim that “The government is concealing what they know about the North Dakota crash.” Crazy! This is what prompted Alexander to posit the “North Dakota Constant” — a threshold of “crazy” in polls about conspiracy theories.

After having considered Russell on empty definite descriptions, however, you might be ready to note something else. “The North Dakota crash” is an empty definite description. So according to Russell’s analysis, the 67% of people who claimed that the government ISN’T “concealing what they know about the North Dakota crash” would ALSO be placed in the position of claiming something false! They would be like a person who is forced to choose between the claim that “The present King of France is bald” and the claim that “The present King of France is not bald”. Neither claim, on Russell’s analysis, is acceptable!

Thinking about this strange result in the Chapman University poll got me thinking about the response to Russell’s analysis of definite descriptions advanced by the philosopher P. F. Strawson.

Strawson suggested that we shouldn’t simply acquiesce to the odd consequences of Russell’s analysis. Instead, Strawson thought that the phenomenon of empty definite descriptions points to the fact that some claims PRESUPPOSE the truth of other claims.

In practice, what this means is that a claim like “The government is concealing what it knows about the North Dakota crash” can only be true OR false — can only be evaluated for its truth AT ALL — if we first presuppose that “the North Dakota crash” is not empty, that it in fact refers to a real event. If “the North Dakota crash” IS empty — if it doesn’t refer to anything — then Strawson suggests that the claim “The government is concealing what it knows about the North Dakota crash” is NEITHER true nor false. This is where Strawson dissents from Russell’s analysis.

And this, after a long detour, finally brings us to the third version of the exchange about “Handy” McChiseledchin. I termed it the “Canceling” option, and now we’re in a position to understand what I mean.

To refresh your memory, here’s the third version of the exchange:

1” – “Did you hear about Handsome McChiseledchin? I hear that he killed his first girlfriend!”
2” – “I happen to know that Handsome McChiseledchin has always been open about his sexual orientation, and that he’s never had a girlfriend.”

Now you’re in a position to see what’s going on. “His first girlfriend” is a definite description. Speaker 2’’ points out that the description is in fact EMPTY. In doing so, speaker 2’’ CANCELS the presupposition on which the truth evaluability of the claim “Handsome McChiseledchin killed his first girlfriend” rests. In this way, speaker 2’’ rejects the claim as not even subject to evaluation.

How could we design a poll to reflect the Canceling option — and Strawson’s insight? Would adding an option like “Neither agree nor disagree” or “Not sure” help? This would be an attempt to build the second version of the exchange — the one that modeled the withholding of belief/disbelief — into the poll.

The problem with this is that it’s not fine-grained enough to take account of WHAT it is that we’re doing when we refuse to affirm or deny a given claim. In the second version of the exchange, speaker 2’ treats the accusation against “Handy” McChiseledchin as worthy of examination, but as needing more evidence before reaching judgment.

This, however, is NOT what Strawson’s example involves. In a case like that, the refusal to affirm or deny a claim doesn’t reflect a lack of evidence. Instead, we refuse to affirm or deny the claim because we KNOW that the very basis of the claim is faulty. It lacks the presupposition that makes it worthy of affirming or denying.

Why is this distinction significant? Why make such a fuss about it?

Here’s why. When you argue whether or not “The government is concealing what it knows about the North Dakota crash”, you tacitly accept the reality of the North Dakota crash. You’ve already taken a step into the world of conspiracist thinking.

The challenge of the present age is that we’re constantly faced with cases in which we need to be vigilant about adopting the Canceling option. Examples include manufactured outrage about unmasking or “Obamagate” or a massacre that never occurred.

Such cases threaten to break the traditional journalistic model of claim and counterclaim, because that model ACCEPTS the presupposition of the reality of the underlying phenomenon at the center of the debate. In dealing with the disinformation rampant in contemporary right-wing messaging, we need finally to be more sophisticated about understanding the conditions that need to be met before a good-faith discussion of policy disagreements can begin.

In so many of the discussions of the manufactured crises emanating from the contemporary right, the correct response is to starve those discussions of oxygen, rather than to fuel the flames. When dealing with controversies revolving around empty definite descriptions, the only winning move is not to play.

Like what you're reading? Don't keep it to yourself!
Share on Facebook
Facebook
Tweet about this on Twitter
Twitter
Share on Reddit
Reddit
Share on LinkedIn
Linkedin
Email this to someone
email