A New Physics Model Is Predicting Box Office Hits Better Than You

A quick list of notable movie flops: _John Carter_ (reportedly costing $250 million), _Waterworld_ (the classic flop), _Swept Away_ (Madonna), _Battlefield Earth_ (Scientology), _Speed Racer_ (live action cartoon hype fail), _Basic instinct 2: Risk...

Jun 15 2012, 8:10pm

A quick list of notable movie flops: John Carter (reportedly costing $250 million), Waterworld (the classic flop), Swept Away (Madonna), Battlefield Earth (Scientology), Speed Racer (live action cartoon hype fail), Basic instinct 2: Risk Addiction (what?), Gigli (the even more classic flop). These are notable because they’re, mostly, movies that should have done OK by conventional Hollywood logic: heavy marketing, big names, cool effects, hype (or at least the illusion of hype). Yet they all more or less sunk. A new model created by physicists at Japan’s Tottori University claims the ability to take all the surprise out of movie flops (and smashes), so far successfully predicting the success or failure of big movies before they’re even released with excellent accuracy. Their work is out today in the New Journal of Physics.

It works by inputting two sorts of data: word-of-mouth discussion (as collected from social networks and blog posts) and daily advertising expenditures for each of the films. The output is the likelihood that a given person will go see the movie between 60 days post-release and 100 days post-release. That modeled output matched the actual output for several recent big budget films very well. As a bonus, the results also found a direct correlation between the number of blog posts discussing a particular movie and that movie’s revenue, which is perhaps a bit obvious, but still kinda interesting. If that aspect of the research holds up in the U.S., we can expect Prometheus to do fabulously.

The subject here is movies, but the researchers assert their work is general, an aspect of what’s know as many-body theory, which explains the collective behavior of interacting particles. Thus, their work should be able to describe an unlimited variety of “hit phenomena.” (There are, of course, many different kinds of hits.) The researchers are now planning on trying their model out on things like music sales, noodle cups, snacks, and “local events.” And, if you’re curious, the movies they actually looked at were The Da Vinci Code (hit novel, god, Tom Hanks), Pirates of the Caribbean: At World’s End (franchise fun), Spider-Man 3 (franchise flavor), Transformers (money plus the filmgoing demands of spoiled children), and Avatar (“profound”).