Spam Software Picks the Sundance Winners

Seems like picking the winners ahead of time isn't that tough after all. As David Miller reports on internetnews.com, an anti-spam consulting company modified an open source application to pick the eventual winners of the 2006 Sundance Film Festival.

The folks at Unspam Technologies, based in Park City, Utah, fed the descriptions and credits of each film at the festival into a modified version of POPFile. POPFile is an open-source spam filtering application that uses Bayesian methods to sort through spam on mail servers. The Unspam Technologies team used Sundance film guides from years past to indentify the most common characteristics of previous winning films, then used that data to sort out this year's winners:

Unspam came up with a list of those films most likely to fly or flop based on more than 200 variables, including the time of day that the movie premiered, the location of the theater and the author of the description in the film guide.

Some data points turned out to be meaningless, like the number of adjectives found in a particular film description. Others, like the number of producers working on a film, proved significant.

These guys should consider taking that POPFile box to Vegas for March Madness...