Sunday, April 21, 2013

Explaining The Twin Movie Observation

Photo Credit: CircadianHour reddit user


Reddit this weekend discovered the strange phenomenon of Hollywood releasing similar movies at the same time. Brushing aside the methodological ambiguities of what constitutes a movie pair, the best explanation seems to be some combination of 1) informal knowledge sharing due to agglomeration effects (social relationships between studio employees, industrial espionage, etc.) and 2) many studios responding to the same exogenous factors (public demand, current events, etc.).

While that's probably correct, I can't help but offer a skeptical take. It's entirely possible that this effect is actually an illusion. If you assume a heavy dose of randomness regarding which movies end up getting made and which ones don't (a safe assumption) and when, then by pure coincidence you'd expect, occasionally, to see similar movies get released close to each other. It might seem like this happens a lot, but we probably tend to notice two similar movies released together more than two different movies released together (because it's funny). And of course there are many, many more unrelated movie pairs than there are related movie pairs.

To accurately verify the twin movie hypothesis, you'd need to set up some stricter measurement guidelines (time categories, theme similarity, etc.). Using Google keyword associations would probably be a good start (Search: "asteroid movie" etc.). A richer approach would also look at different industries with similar observed outcomes. Both music and fashion, which also churn out many artistic innovations, rarely end up with product pairs: if an idea takes off, everyone copies it. But industries where we do see product pairs also tend to have only two major competitors--Hollywood has dozens of major studios. The twin movie hypothesis resembles a duopoly nested within the product differentiation strategy of firms competing in a crowded field. It would be interesting to see if any other economic models (game theory, Cournot competition, etc.) could shed some light on this interesting phenomenon.