If the Internet Movie Database is to be believed, The Shawshank Redemption is the greatest American film ever made. The American Film Institute’s list, meanwhile, lists it way down at number 74, while Citizen Kane takes the top spot. The actual greatest American film, Samuel Fuller’s Pickup on South Street, doesn’t make either cut.
OK, that’s just one lowly reporter’s opinion.
Ranking is subjective business, yet still we feel the need to preserve or otherwise laud particularly good cultural artifacts, like when the National Film Preservation Board decides which movies to protect for future generations—Ferris Bueller’s Day Off, for example. Though it’s difficult to measure, it would be nice if the Preservation Board had some sort of objective measure to use when making these decisions. Now, network science researchers at Northwestern University have a system to do exactly that: citations.
Ranking is subjective business, yet still we feel the need to preserve or otherwise laud particularly good cultural artifacts, like when the National Film Preservation Board decides which movies to protect for future generations—Ferris Bueller’s Day Off, for example.
In developing a citation-based measure, Max Wasserman, Xiao Han Zeng, and Luis Nunes Amaral took inspiration from two sources: academic citation networks, which track both research-to-research referrals, and Google’s PageRank, which follows links from one webpage to another.The basic idea is that when someone cites or links to someone else, that indicates the cultural importance of the latter’s work. Just as in academia and the Web, movie directors allude to each others’ work all the time, but there’s just one tiny problem: Directors rarely, if ever, note those allusions. There’s no credit at the end of The Departed, for example, paying tribute to 1930s cinema like The Public Enemy or Scarface, both of which Scorsese’s film borrows from.
Fortunately, there’s now a great resource for uncovering those links: IMDb, which includes a “connections” section for the movies in its database. Using that as a foundation, Wasserman, Zeng, and Amaral constructed two measures: a simple count of the number of times movies made reference to a given film, and the original PageRank algorithm, re-purposed for movie allusions instead of webpages.
The movie-buff scientists next pulled in more traditional variables, such as Roger Ebert’s starred reviews and Metacritic averages, to predict whether a film was included in the NFPB’s National Film Registry. Across a variety of prediction algorithms, PageRank kept pace with Metacritic scores and generally outperformed Ebert reviews, IMDb ranking, and citation count.
Interestingly, a version of the citation count that included only references made 25 years after the original film came out—The Departed‘s allusion to Scarface would count, but the Spaceballs call back to Chinatown wouldn’t—outperformed both Metacritic and PageRank.
“Twenty-five years may seem like a long time to wait to begin quantifying film significance. However, significance by definition may not be readily apparent,” the authors write in Proceedings of the National Academy of Sciences. “A film’s significance”—and, they point out, the significance of other art or even scientific research—”should ultimately be judged on how its ideas influence filmmaking and culture in the long term.”