IE 11 is not supported. For an optimal experience visit our site on another browser.

Horrors! Software Predicts the Best (and Baddest) Movies

Researchers say the number of times a movie is referenced by other movies is a better predictor of its importance than Oscars or critical acclaim.

An automated system that counts how many times movies are referenced by later movies can outdo film critics and the Oscars when it comes to predicting which films will stand the test of time, researchers report. But wait, there's more: The system suggests that more appreciation should be given to horror films — including the one that some consider the worst movie ever made, "Plan 9 From Outer Space."

Researchers at Northwestern University lay out their case in the Proceedings of the National Academy of Sciences. They looked at various metrics for judging more than 15,000 movies — including critical reviews, awards and box-office sales — and studied how well each of them matched up with a film's inclusion on the Library of Congress' National Film Registry. The movies are selected for showcasing by the National Film Preservation Board and the director of the Library of Congress.

The best predictor for inclusion on the registry turned out to be the number of times a movie that's at least 25 years old is referenced by other movies.

So if the movie-citation metric is so good, which movies are the highest-ranking candidates yet to be added to the registry? The research team came up with this list:

  • "The Texas Chain Saw Massacre" (1974)
  • "The Seven Year Itch" (1955)
  • "The Wolf Man" (1941)
  • "The Shining (1980)
  • "Creature of the Black Lagoon" (1954)
  • "Dumbo" (1941)
  • "Spartacus" (1960)
  • "Friday the 13th" (1980)
  • "Plan 9 From Outer Space" (1959)

"If one were to look for a pattern in these movies, probably the strongest one is the over-representation of horror movies," Northwestern engineering professor Luis A. Nunes Amaral told NBC News in an email. "One could hypothesize that many people are biased against horror movies and that it is harder for such movies to make it into the National Film Registry."

Amaral said the big-data experiment was aimed at testing techniques for deciding which innovations in research, technology and medicine should be adopted over others.

"Unlike determining whether a thermometer is more accurate than another, which can be trivially done, there are many situations in which comparing the merits of two competing approaches is not at all trivial," he said. "Films provided a good model system with which to study and compare the performance of different measures of quality/significance/merit."

And now, judge the quality/significance/merit of "Plan 9 From Outer Space" for yourself.

— Alan Boyle

In addition to Amaral, authors of the research paper, titled "Cross-Evaluation of Metrics to Estimate the Significance of Creative Works," include lead author Max Wasserman and Xiao Han T. Zeng. The U.S. Army Research Office supported the research.