Brendan Nyhan’s tweet made me wonder:
“Seriously considering whether this New York Magazine article was planted as a giant field experiment to document liberal motivated reasoning.”
I am of the view that conspiracy theories are inherently natural and the wilder versions of conspiracy theories are a product of society where free exchange of idea is hampered. The problem with overreliance on data, to me, is that ALL DATA is constrained, limited, and are subject to some form of selection bias. There is no single obvious way to interpret data and certainly, data is not self-explanatory by itself. There are contexts where the data is deliberately made easy to interpret (this is why Google translate is terrific in translating business-like documents, whether from business, government, or other settings where the language is clear and straightforward), but terrible at translating literature and love letters (where nuance is the key). This is something that I know well from personal experience. My Spanish is not very good, but I have (had) little trouble reading political science and economics papers written in Spanish and can decipher history books reasonably well. But literature? Poetry, especially? Oh, boy.
The characterization of the inputs that goes into econometrics by Ed Leamer is applicable here: we never deal with data analysis devoid of “prior information.” All of prior information, every theory, every belief, every conjecture that we bring into data analysis is, in a sense, opinion, even if “reasoned and informed” opinions. And they are informed, by and large, by the people, defined broadly, that we interact with. Newton thanked “the shoulders of the giants.” It’s not the shoulders, but the informed opinions of the giants who came before us, among many others, who shape the opinions that we bring into the data analysis. Where the giants were wrong, as was the case for Ptolemy and others of the classical sciences, it can take a long time to unlearn their “prior information.”
Conspiracy theorists, like Galileo, are people who don’t buy into the “prior information.” They are “crazy,” certainly, but there is no guarantee that they are necessarily “wrong.” To the degree that the conventional wisdom, usually, tends to be right–after all, that is why the giants are, well, giants–they probably are wrong, but that cannot be taken for granted. The key, really, is that they see, for whatever reason, see something in the “data” that others do not see. Maybe it is nothing. Maybe it’s a formation of Japanese dive bombers heading to Pearl Harbor on December 7, 1941. Maybe it is worth the while to talk to them. Maybe, they will see the error of their analysis after an honest talk. Or, conversely, we will see the error of our ways and update our thinking.
To the degree that there is no unique and indisputably correct interpretation of the data, some form of “conspiracy theories” are inevitable. The extent of the conspiracy-mongering, however, is a function to which people can exchange ideas honestly and productively. Free and honest exchange of ideas allow people to learn from each other, change their thinking, and reconcile how they view the data on hand–usually for the better, although, sometimes, it can lead astray. (Stiglitz-Grossman Theorem, and the idea of beauty contest games in general, point to how people can go off the cliff, almost literally, because they learn all the wrong things from the crowd whom they think is smart, but in fact is full of copycats trying to mimic each other.)
What breeds conspiracy theories is when free and honest exchange of ideas are interfered with. The only sanctioned truth is the official version enforced through an authority that may not be questioned but the reality provided plentiful contradictions in the data seen by the people. No one, of course, has the full scope of the data so no one knows the whole picture, except that they are convinced that the official truth is wrong, or at least incomplete. Since they cannot trust the official truth, they start cooking up their own theories of how the world works, based on the limited information that they have that they feel that they can trust. These are, essentially, how conspiracy theories are born. Soviet Union and the Middle East, among other places, and for that matter, authoritarian regimes in general, were full of conspiracy stories, but that does not necessarily mean that free exchange of ideas is always precluded by state action alone. Indeed, if left to ourselves, we tend to do a pretty good job of hampering free exchange of ideas among ourselves. The story about the fake news in the social media has been the big deal, but the fake news became influential because they found a market that found it more credible than the alternatives–the “official” news from various sources. We believe the news that are compatible with our existing worldview of how the world works and the facets of the world that we do see, and, indeed, the kind of news that are trusted by people whom we trust for whatever reason–such as our Facebook friends, who may or may not be experts, but at least, whom we know to be in similar situations as we are. Thus spreads news through social media, whether they are “fake” or “real” is not relevant. To the degree that people who don’t necessarily share the same worldview do not exchange ideas freely with each other–and in social media, too many of us sanctimoniously avoid interacting and exchanging ideas with people whom we don’t understand, let alone disagree–we shall have given births to thousands of conspiracy stories subscribed to by different cohorts of people, all because they match up with the theories of reality and the limited snippets of data that we happen to have easy access to. As the aftermath of the 2016 election shows, people of all political stripes are guilty of this.
I saw this phenomenon much earlier: after 2004 election, there was a similar wave of delusional thinking among the otherwise reasonable people on the left, followed by a repeat on the right, on a grander scale, in 2012. My response after these was to forswear having opinions about politics, as they interfere with analyzing the data in a neutral and detached fashion. Unfortunately, not having opinions and preconceptions also means that the people whom I talk to have no idea what I am talking about and what context I should be placed in. Context is an amazing thing: Steinbeck was on to something when he placed the following words in Lee’s mouth in East of Eden:
Lee looked at him and the brown eyes under their rounded upper lids seemed to deepen until they weren’t foreign any more, but man’s eyes, warm and understanding. Lee chuckled. “It’s more than a convenience,” he said. “It’s even more than self-protection. Mostly we have to use it to be understood at all.”
But it is not just mannerisms of speech and ethnic stereotypes. It is the lingo developed by every group. Republicans talk one lingo, Democrats talk another lingo, data science people speak in a particular vocabulary, while stats folks have something else. They may be talking about the same thing, but they talk past each other because they don’t speak the same language. Or, in some cases, Democrats can understand the Republican lingo, but they cannot understand something that is neither Democrat or Republican, and mistake the lingo for something else. Like E. A. Poe’s Auguste Dupin might have said, if you don’t know Italian or Orangutan, they all sound the same, especially if you are prejudiced against Italians. This is a reference to Murders in the Rue Morgue. The murders were in fact committed by an escaped Orangutan, but a witness who only heard loud howls, insisted that the howls were an Italian. Needless to say that the witness really did hear loud howls, did not know any Italian, and since he didn’t like Italians, naturally came to believe that the howls must have been in Italian, leading the French police to search for a very strong red-haired Italian acrobat as the most logical suspect.
If you think the search for red-haired Italian acrobat was absurd, it was in fact the most logical conclusion given the data on hand: the suspect was obviously very strong, red hair was found on the bodies, and the suspect was seen getting away leaping from rooftop to rooftop in a manner that no ordinary human could emulate. The flaw, of course, was that there should have been no reason to believe that the witness who knew no Italian could possibly have credibly identified the howls as Italian, but then, he did not know anything about Orangutans either so correctly identifying the scream as something not human (but what exactly it was, he could not have answered–and who knows if Italian howls are like French howls anyways?) would not have helped narrow the search. Besides, an Italian murderer in Paris, logically, made far more sense than an Orangutan one since there were many Italians in Paris while, as far as the authorities knew, no Orangutan. Knowing all these, what’s the more ridiculous story? The truth, or what is more plausible–especially if you “know” from get go the Italians are no good?