Imitation vs. Innovation: Product Similarity Network in the Film Industry [pdf]

One of the reasons for similarities in Hollywood is that outright theft is quite common. Historically, it has been very difficult to prove damages to the necessary degree of certainty.

This problem would be mitigated if more jurisdictions allowed liquidated-damages clauses in contracts to be enforced on a consistent basis. Unfortunately, most attempts at liquidating damages fail because courts usually characterize them as unlawful penalties.

Judge Richard Posner, the most cited jurist of the last century, argues that sophisticated parties should be permitted to liquidate damages however they want, regardless of whether a court believes their agreement was reasonable. "The reason for the [contrary] rule is mysterious," he says, "it is one of the abiding mysteries of the common law."

In the 1970s, businesses in California persuaded the California legislature to amend the relevant statute to create a rebuttable presumption that liquidated-damages clauses were lawful. Yet the Courts were so attached to the underlying rule that it made no difference! The California Supreme Court has inexplicably extended the very lines of authority that the legislature explicitly intended to override.

I was involved in litigation that involved the red-handed theft of a pitch by the American division of a multi-billion dollar conglomerate. According to the terms of the liquidated-damages clause that the parties adopted, this theft would cost more than $120 million dollars. After years and years of litigation, with back-and-forth rulings, it was finally declared unenforceable. The firm whose pitch was stolen lost the chance for a life-changing opportunity, which may have cost it a hecto-million dollars or more.

Until there is some realistic way to protect artistic ideas, whether through contract or state fiat, imitation will prevail over innovation, because the rewards are simply inaccessible to most outsiders. The means of production are owned by the existing Hollywood elite, and they simply steal whatever novel ideas newcomers develop.

Please note: I am NOT advocating more stringent patent laws — or even stricter copyrights! People who are inspired by public ideas should be free to incorporate them in their art. But private parties also need the ability to disclose non-public ideas, without forfeiting their value. At least in Hollywood, right now they do not. And before that changes, expect the status quo.

Isn't one factor in movie similarity not just direct imitation between movies but the use of plot/script formulas[1]? As I understand, screen writing always involved hooks to emotional impact but as movies evolved, an arms race to increase the speed and impact of emotional developed where now the aim is "grad their attention in the first N seconds and never, ever let go". The situation seems akin to the loudness wars [2].

[1] http://www.slate.com/articles/arts/culturebox/2013/07/hollyw...

[2] https://en.wikipedia.org/wiki/Loudness_war

There's a slight weirdness in this paper in the way they measure similarity (which is admittedly a very hard problem). It's entirely derived from the recommendation networks of IMDB and Amazon, which are mostly built on the basis of "people who liked/watched this also liked/watched these movies". Presumably this should measure similarity on many fronts, taking into account the wide array of each user's subjective factors, including screenwriting style as mentioned in joe_the_user's comment, for example.

I think it's a pretty wobbly base for the research, though. It doesn't make it any less interesting, but even assuming that the recommendation networks perfectly represent the network of user preferences, does this really reflect movie similarity? People stick to directors or actors that they like, even across very dissimilar movies, but this still counts as a form of "similarity" in this paper.

And the Amazon network "finds items that customers tend to purchase together", which might not represent similarity at all, but complementarity!

Also, consider a film that's extremely similar to another one, but is marketed in such a way as to reach a different audience. This happens when marketers deem that "lying" about the movie in the trailers and marketing campaign will make for better box-office/video sales. Again, this might bend the recommendation networks in a way that diverges from actual movie similarity.

Despite this, the study hardly seems to question their measure of similarity at any point (I haven't read everything, though). Or rather, they redefine "similarity" to mean "proximity in the recommendation networks", which includes all these hidden factors that might overlap with other things they're comparing against. As I said, though: it's a hard problem.