Paper Review – Conspiracy Theories and Fortuitous Data (Joel Buenting and Jason Taylor)

Conspiracy Theories and Fortuitous Data – Philosophy of the Social Sciences, iss. 40, 2010

Joel Buenting and Jason Taylor

Sometimes you read a paper and think “I wish I’d written this.” Sometimes you read a paper after finishing a PhD on a particular topic and go “Dammit! I should have thought of this.” “Conspiracy Theories and Fortuitous Data” is my “Dammit!” paper; the notion of fortuitous data which Joel Buenting and Jason Taylor present as a demarcating factor between warranted and unwarranted conspiracy theories is something I wished I had dreamt up and included in the thesis.

Now, the preceeding paragraph might suggest that I really like this paper (and I do) but it doesn’t mean that I don’t have issue with it. Whilst the notion of fortuitous data is a useful one, I don’t entirely agree with:

a) the definitions of both “conspiracy theories” and “official stories” Buenting and Taylor use which

b) means that I think there is a bit of a problem about how they use fortuitous data to demarcate between conspiracy theories and official stories and

c) I don’t think the demarcation, as they present it, tells the full story of the role of fortuitous data in conspiracy theories.

Buenting and Taylor start with an interesting distinction between views on the rationality of conspiracy theories (as found in the existing literature):

Issues surrounding the rationality of conspiracies have received increasing attention among epistemologists. Central to these discussions is the question of whether it is ever rational to accept or believe a conspiracy. Opposing views can be distinguished based on how they approach conspiracies. According to the generalist view, the rationality of conspiracy theories can be assessed without considering particular conspiracy theories. On this view, conspiratorial thinking qua conspiracy thinking is itself irrational. The particularist view about conspiratorial thinking denies that the rationality of conspiracy theories can be assessed without considering particular conspiracy theories. (p. 568-9)

They note that they are particularists (as am I, under this schema) and then note:

We suggest that the presence of fortuitous data helps distinguish at least prima facie reasons to hold some conspiracy theories as rational. Presence of fortuitous data, we claim, is good reason to dismiss a conspiracy theory as irrational. … We note that the presence of fortuitous data in an accepted conspiracy theory lends credence to the claim that fortuitous data can help to demarcate rational from irrational conspiracy theories. (p. 569)

Buenting and Taylor’s overall argument, that we should judge individual conspiracy theories with respect to the evidence rather than with respect to judgements about the general class of things called “conspiracy theories” is one that I am on record as endorsing (I wrote a doctoral thesis on this, for example). However, where Buenting, Taylor and me differ is on what gets counted as being a conspiracy theory. They make use of David Coady’s defintion, which contains in it a condition for being a conspiracy theory is that it must be in conflict with some official story (or official theory as I tend to call them).

I’ve remarked on this kind of definition before and I still find this requirement that conspiracy theories are not official stories to be a strange one because it means officialness plays a weird role in demarcating conspiracy and non-conspiracy theories. For example, it means that a conspiracy theory can suddenly become a non-conspiracy theory because it has been endorsed by some official source. Now, to be fair, Coady doesn’t think that official stories are necessarily any better, epistemically-speaking, than conspiracy theories; he is just respecting the intutition that “conspiracy theory” has pejorative connotations (an intitution I push against in my definition), but I still think this is a weird thing to respect. How do we make sense of cases like Watergate and the Moscow Show Trials if this is the operative definition in play? That being said, Coady’s definition is popular amongst my peers and the eventual book I will write will need to address, in fulsome detail, my worries with this particular condition of his definition.

Moving on:

Why do we accept a particularist approach to conspiratorial thinking? Choosing to believe a conspiracy theory is the choice to believe in one theory (the conspiracy) over another theory (the official story). Though this may seem obvious, it is crucial to note because it draws to attention the idea that a rational choice to believe any theory depends on considerations of evidence. Judging any theory to be insufficient independently of considerations regarding the evidence is irrational. Thus, a rejection of conspiracy theories simpliciter seems irrational; rational rejection or acceptance of a theory must supervene on the quality of evidence for or against that theory. (p. 569-70)

Whilst I agree with the general sentiment of the authors, that we should inspect the arguments for a conspiracy theory rather than reject them outright because they belong to the class of things called “conspiracy theories,” I think the notion that it’s a theory choice between conspiracy theories and other (non-conspiracy) theories is problematic, for at least two reasons.

1. As Charles Pigden has argued, sometimes the choice of theory is restricted to different conspiracy theories (any explanation of 9/11 seems to involve choosing a conspiracy theory, for example).

2. It isn’t clear from Buenting and Taylor’s paper what work “official” is doing with respect to these things they call “official stories.” Unless we finesse what an official story is and what beliefe in such a story entails, it’s hard to know why they run this particular distinction. Since Buenting and Taylor do not define official stories, this is a problem. Are they using Coady’s definition of “official story” or do they have some other defintion in play?

This worry about this lack of a definition for “official stories” is compounded by what Buenting and Taylor list as examples of official stories which are also examples of conspiratorial thinking which they then refer to as “uncovered conspiracy theories” (p. 570). They talk about Watergate, for example, as a reason for adopting the particularist approach (an approach, as I said before, I have no issue with) but if Watergate is an uncovered conspiracy theory, does that mean that official stories can be uncovered conspiracy theories (the approach Coady takes) or are they operating with some other notion of “official story”? Once again, the lack of a definition for this crucial term is frustrating, because, by-and-large, I like this paper (and I wish I had written most of it) and it would be quite simple to fix this issue; either state explicitly that the notion of “official story” in use here is Coady’s or state how they differ.

Buenting and Taylor go on to say:

Uncovered conspiracy theories suggest that it is sometimes rational to believe conspiracy theories and that sometimes conspiracies occur. These cases strongly suggest that assessing the rationality of a conspiracy theory should be done on a case-by-case basis. Thus, these cases suggest that if we are to take conspiratorial thinking seriously, we ought to adopt the particularist approach.

This is a victory of sorts for the conspiracy theorist, but a minor one. A conspiracy theorist wants something more than the (backhanded) compliment that her theory type is not unequivocally irrational; a conspiracy theorist wants recognition that her theory token reflects the true state of affairs, the actual explanation of the event in question. (p. 570)

Hear hear, I say (and that last paragraph is beautiful; I wish to steal it and make it my own). That being said, surely it depends on what your definition of “conspiracy theorist,” is? I take it that anyone who believes any conspiracy theory turns out to be a conspiracy theorist. The authors could be argued to be claiming conspiracy theorists are what my supervisors and me coined as being “conspired world theorists” (because of their emphasis on both the type and the token) and I think it would be a mistake to think that the former (conspired world theorists) are necessarily the latter (people who hold to specific conspiracy theories). This is a problem of definitions, again, and to a certain extent it follows from the authors’ use of Coady’s definition, which has it that conspiracy theories are contrary to official stories.

Their reliance on the Coady definition, however, does get them their demarcating criteria between rational and irrational conspiracy theories:

“We take the difference between a rational and an irrational conspiracy theory to hinge not only on its explanatory power, but also on one feature of the evidence for the official story to which it is opposed, called fortuitous data.

What is fortuitous data? It is data that:
i) supports the official story; but
ii) fits the official story too well; is “too good to be true.” Finally,
iii) the “lucky” nature of the data is left unexplained by the official story. (p. 572)

Buenting and Taylor take it that fortuitous data is a property of evidence that supports some official story and is part-and-parcel of the contrast between conspiracy theories and official stories. As we have to chose which story or theory to be believe, we have to consider the evidence. Now, I kind of agree here that we often when considering conspiracy theories we are talking about contrasting one explanation with another, but this is a feature of any situation where there are competing explanations, rather than one that is unique to conspiracy theories (fortuitous data may well be found in non-conspiracy theories).

The focus on conspiracy vs. official theories seems to be making this more complex than it need be; what if the contrast is between two clear conspiracy theories (for example, you might have a case where its obvious a conspiracy was responsible for some event and in both cases you might also reasonably assume that the conspirators are putting out disinformation)? I don’t think this analysis needs to talk about official stories because I don’t think fortuitous data needs to be restricted simply to official stories.

[T]he lucky nature of fortuitous data is crucial. It tries to capture the idea that the nature of some evidence is just so lucky that it cannot be shrugged off as mere chance; the luckiness of events requires explanation.” (p. 573)

One example the authors use to illustrate a case of (seemingly) fortuitous data is American Airlines Flight 77’s crash into the only section of the Pentagon reinforced to withstand such an impact.

[T]he official story explains why the plane crashed into the Pentagon rather than into some random neighborhood (for example), but it does not explain why the terrorists choose that particular part of the Pentagon rather than any other. But, the plane hitting the Pentagon where it did—rather than anywhere else — is highly fortuitous — given that the Pentagon was reinforced only in that section. Moreover, that the plane hit the refurbished section of the Pentagon rather than anywhere else suggests that the pilots intended to hit just that spot. (p. 573)

The convenient location of the impact is a question 9/11 Truthers continue to hammer home; it looks fortuitous for the accepted explanation of the attack on the Pentagon that it hit there (rather than in a section which had not been reinforced to withstand such an impact).

Supporters of an official story do not (normally) find the “lucky” nature of fortuitous data problematic; conspiracy theorists do, emphasizing the fact that the “lucky” nature of the data points toward a different explanation. Is the conspiracy theorist justified in holding this? Consider the Pentagon Case again. How does the official story account for the location of the plane crash? In so far as it is merely a plane hitting the Pentagon, the official story can account for this by pointing to the (supposed) terrorists who hijacked Flight 77. Period. But, the official story is silent about why the plane hit the Pentagon where it actually hit the Pentagon. (p. 574)

I like this example because it provides for a nice illustration of how some piece of evidence might be considered by some as being fortuitous data and by others as what I will call “fortunate data”, evidence which could be construed as fortuitous data but really is just an example of chance.

Buenting and Taylor, having come up with an example of a very interesting demarcation between conspiracy theories and other (non-conspiracy) theories ask the obvious question: Is there an example of fortuitous data we can point to to show this demarcation works? They answer “Yes”, writing:

[T]he Watergate scandal is both an accepted conspiracy theory, and it also contains (at least) one instance of fortuitous data. (p. 576)

As previously mentioned, Buenting and Taylor refer to Watergate as “an accepted conspiracy theory”. Under the definition they have borrowed from Coady, they should be referring to this as an official story. Once again, the lack of a definition for what they take it are “official stories” is, I think, problematic.

That being said, Watergate is a good example of the way in which some data ends up being fortuitous rather than just fortunate because of the infamous eighteen and an half minute gap in the recordings of conversations Nixon had with officials about the Watergate Hotel break-in.

That the tapes had such a gap, while showing no involvement on the part of Nixon:
i) supports Nixon’s story (i.e., the official story); but
ii) the “gap” in recording is “too good to be true,” which suggests that
Nixon is hiding something; and, finally,
iii) the tape’s gap goes unexplained by the official story. (p. 577)

Contrast this example with the Pentagon example. In the Watergate case I think it’s fair to say that the eighteen and an half minute gap really is “too good to be true” and thus counts as fortuitous data, whilst the Pentagon case might be an example of fortuitous or fortunate data with respect to the official story. I like this paper and should the book version of my thesis ever get commissioned I’m going to supplement my coverage of selective evidence (chapter 5) with discussion of Buenting and Taylor’s notion and analysis of fortuitous data. I’m particularly interested in the Watergate example as they present it because it seems to be that it’s a case where fortuitous data and selective evidence combine; the eighteen and an half minute gap isn’t just fortuitous but it is also an example of evidence selection in that Nixon and his cronies massaged the evidence to support the claim he knew nothing about the break-in to the Watergate Hotel.

However, I have an issue with this paper which is more than the lack of a definition for “official stories” (although it is related). I think Buenting and Taylor’s analysis would be better focussed on patterns of fortuitous data rather than just the existence of fortuitous data. One piece of data which appears “too good to be true” might just be lucky-cum-fortunate, but more than one piece, a pattern of such data, might well suggest it is fortuitous-cum-designed/conspiratorial.

Now, to be fair, Buenting and Taylor do get into this issue when they talk about luck and fortuitous data but I still think there is more to be said. Some data will be fortunate (in the sense of lucky) whilst not being fortuitous (in the sense of being “too good to be true”). Separating out data which is truly fortuitous from merely fortunate data is going to be difficult, given that a relative judgement of the data as being either fortuitous or fortunate will likely affect the way you interpret other “lucky” pieces of evidence.

Now, I freely admit that talking about a pattern of fortuitous data is going to be problematic because, as I said, some data really will be fortunate (with respect to some theory) and not examples of selective evidence or fortuitous data; talking about patterns might well end up mixing in fortunate data with fortuitous data (and counting one as being an example of the other; this is a problem which cuts both ways). However, any piece of data on its own could look fortuitous; you need context to judge the likelihood that the data is more than chancy but, rather, too good to be true. Buenting and Taylor give an example of one of the 9/11 hijacker’s passport being found close to Ground Zero, which is an example of data which is fortunate rather than fortuitous but, and this is the problem with my notion of patterns again, it could be construed as fortuitous with respect to other pieces of the evidential record.

So, what distinguishes fortunate from fortuitous data is my major question. There is, I think, going to be a certain presumption in any talk of such a distinction acting as a demarcating factor between conspiracy and other theories that we can actually tell the difference. Watergate is a nice example because it’s clear the fortuitous data was really an example of evidence manipulation; the story about the gap in the evidence was too good to be true. With respect to the impact of American Airlines Flight 77 into the Pentagon, well, there are a host of factors which weigh in on whether it’s merely fortunate data for the accepted conspiracy theory or, as 9/11 Truthers would have it, fortuitous data.

Now, to stress that my worry is really just one about how to develop this material further (and how to integrate it with my own work, which uses a slightly different definition of both “conspiracy theory” and “conspiracy theorist”, I’m going to end this paper review/set of lengthy notes with one of Buenting and Taylor’s footnotes (which I think shows they are quite aware of the kinds of issues I’m also concerned with):

Here’s an interesting wrinkle with our account of fortuitous data: aren’t cases of fortuitous data “too suspicious to be untrue”? This can have a strong and a weak reading. The weak reading claims that fortuitous data points to only poorly concocted conspiracies. Consider the Passport Case again: finding the passport at the base of the towers seems too lucky, suggesting that the conspirators should have known better. Thus, fortuitous data points to poorly thought out conspiracies. Our response, of course, is that a poorly concocted conspiracy is still a conspiracy that could be rationally believed in. Thus, this weak reading can hardly count against our account, given that we want to show that sometimes it’s rational to believe in some conspiracies.

The stronger reading is potentially more troubling. Suppose, for the moment, that the CIA orchestrated the attacks of 9/11. We know that the members of the CIA are highly intelligent, so it seems safe to assume that they know better than to plant a passport. If so, then the passport has to be from the hijackers on the plane; its presence is “too suspicious to be untrue”: it just can’t be planted, because no CIA operative would be that careless. Thus, the objection runs, believing a conspiracy theory in this case would be irrational. We have two responses. First, the strong reading is not possible for all instances of fortuitous data. The lotto case is a prime example of data that is not too suspicious to be untrue. Second, for those cases where the strong reading is possible, this merely highlights the importance of other features of conspiracy theories. We do not claim that fortuitous data is sufficient for picking out rational theories—and this example illus- trates why. Thus, in cases where the strong reading is possible, the conspiracy theorist will have to defend her pet theory by referencing fortuitous data in conjunction with other features, like errant data, explanatory power, simplicity, etc. (p. 575-6)