Order-of-Magnitude Popularity Estimation of Pirated Content

Abstract

Understanding the spread of information in complex networks is a key problem. Content sharing in popular online social networks such as Facebook and Twitter has been well studied, however, the future trajectory of a cascade has been shown to be inherently unpredictable. Nonetheless, cascade virality has recently been studied as a classification problem, resulting in good prediction accuracy. Herein, we address the important problem of pirated media popularity estimation in torrent applications, such as Project Free TV, Popcorn-Time, and The Pirate Bay. Although pirating software and media is illegal, the practice of pirating is actually growing in popularity. On a large sample of data acquired from The Pirate Bay, we demonstrate high accuracy in the task of identifying whether the popularity of a torrent will continue to grow in the future. Specifically, we achieve close to perfect accuracy in estimating the order-of-magnitude popularity of torrents.