What has surprised you the most in your study of conspiracy theories?
When I joined the project I thought of conspiracy theories as part of a family of arguments that point to simple agency rather than acknowledging more complex or even random interactions. The precise ways in which conspiracy theories are different from other forms of explanation seemed (and still seem) to me more normative than substantive. In certain cases, (here Putin’s Russia is an obvious example), respectable academic discussion appears steeped in conspiracy theorising.
Building on from a point raised by John about gaussian and power distributions: we have more of a problem with randomness than we realise; in certain situations this ‘problem’ is labelled a conspiracy theory, in others not.
I would like to present an example from sport where agency-based explanations erroneously displace complexity: at around this time every year the cold-nosed businessmen running football clubs begin to cave to fan pressure and sack underperforming managers. Someone else is brought in, often results improve, and we hail the new manager as a genius. Until he gets the sack a year or two later.
Daniel Kahneman explains this phenomenon in terms of golf and reversion to the mean: to succeed at golf, Kahneman argues, you need skill + luck. At the top level of sport, everyone has a lot of skill, but luck comes and goes – hence the fluctuations. To Kahneman, pundits and experts spend their days discussing variation in luck more often than skill. The upshot is that we should not be surprised when underperforming teams (or golfers) improve from one tournament to the next, or when football teams win the premier league one year, but find themselves struggling to make the top four in the next.
For fun, I once wrote a basic simulator that would calculate what sort of points distribution we should expect to see at the end of a premier league season, depending on the relative proportions of skill and luck. Running many thousand iterations of the simulation showed me that with no skill at all, you would almost certainly see a distribution from about 35 points up to about 70 points. Because football, unlike tennis, is decided by a few, improbable events rather than many repeated tests, the skill advantage needed to begin to be assured of winning the league is very high. In tennis, a stylised form (one player serves, the other returns) is repeated across multiple games and sets. The more repetitions, the smaller the effect of luck. Consequently there are fewer upsets in tennis than in football. (This repetition may also help explain strange results in the women’s game, where grandslam matches are played as best of two rather than three sets)
The upshot of all this is we are bad at distinguishing a Gaussian distribution from genius and incompetence. With this in mind, I am surprised talk about the internet and technology companies has not been more dominated by conspiracy theorising than it has. As John notes, the power law is the norm on the Internet. If we are naturally inclined to overanalyse Gaussian distributions, the inequality created by power laws should be off the scale. Yet, the data I’ve seen so far suggests conspiracy theories are no more likely to go viral than other categories of content. And although the internet clearly has helped accelerate distribution of material, I’ve yet to find any evidence that conspiracy theories make up a disproportionately large slice of online communication.