Read our COVID-19 research and news.

A convenient way for sharing files, peer-to-peer networks such as BitTorrent have gained notoriety for allowing users to illegally download copyright protected music and movies.

A convenient way for sharing files, peer-to-peer networks such as BitTorrent have gained notoriety for allowing users to illegally download copyright protected music and movies.


A peek inside the Internet's favorite file-sharing network

More than a quarter of all Internet traffic belongs to BitTorrent, a file-sharing system that allows users to swap everything from music to movies. Now, for the first time, researchers have revealed a link between a country’s economy and the type of files its residents download from BitTorrent. The findings are shedding new light on online behavior and could help law enforcement track down Internet pirates.

“[The researchers] found an intriguing and creative approach,” says Johan Pouwelse, a computer scientist at Delft University of Technology in the Netherlands, who was not involved in the work. “I’ve never seen a study done like that.”

BitTorrent is a so-called peer-to-peer (P2P) communication protocol. Unlike traditional networks that store information on a central server, which power well-known services like Google and Facebook, P2P relies on users to make their resources directly available to other users. The technology supports everything from Skype calls to the U.S. military’s intelligence-sharing network. Though people use BitTorrent to swap legal material such as free software, the network has proved particularly attractive to those sharing copyrighted music, movies, and games illegally. Just last month, the parent company of the Ultimate Fighting Championship settled a multimillion-dollar lawsuit against a New York resident who confessed to uploading pay-per-view events to popular BitTorrent repositories like the Pirate Bay.

The decentralized nature of BitTorrent not only presents a challenge to law enforcement officials, but also to researchers seeking large-scale data on user behavior. Investigators can’t conveniently retrieve information from a central server, but instead must monitor individual users to gain a global picture.

So computer scientist Jordi Duch at Rovira i Virgili University, Tarragona, in Spain and his colleagues collaborated with computer scientists at Northwestern University, who developed a software plug-in called Ono. The software accelerates BitTorrent download speed. In exchange for this benefit, Ono’s 1.4 million users can choose to release information on their file-sharing behavior, such as the timing of downloads and the size of the shared files, for research purposes. To protect the users’ privacy and entice cooperation, the researchers did not collect data on the content of the shared files. They also used an algorithm to anonymize the data, so that the researchers themselves did not know the IP addresses of the users, which can be used to track them down.

Without information about the file content, the researchers invented a creative way to gauge what exactly users were sharing. Sampling files shared on the Pirate Bay, they found that different types of content correspond to specific file sizes. A 100-megabyte file is more likely to be a digital music CD, whereas a 2-gigabyte file tends to be a high-definition movie.

After more than 4 years tracking user behavior, the researchers analyzed the activities of about 10,000 active BitTorrent users during a typical month. The users—at least those who agreed to use the Ono plug-in—behaved in a highly predictable manner, the team reports online today in the Proceedings of the National Academy of Sciences. A majority downloaded files on a weekly basis, and most focused on sharing one or two types of files. By looking at a user’s first 100 downloads, the team could predict with 80% accuracy what type of file the user would download next, Duch says.

But what really stood out is that users from the same country displayed a tendency to download the same types of content. After factoring socioeconomic indicators such as a country’s gross domestic product per capita and Internet access, analysis showed that users from rich countries such as the United States tended to download more music files, whereas users from poorer countries such as Spain favored movie files.

Alex Kigerl, a cybercrime researcher at Washington State University, Pullman, who was not involved in the study, says he finds the results surprising, as he would expect users from richer countries to take advantage of better Internet infrastructure to download larger files. Duch suggests that the availability of online streaming services such as Netflix may have curbed the tendency for Americans to download movies and TV shows. In contrast, in countries like Spain, where no such services are available, downloading became the prevalent way to access such content.

The study could help engineers design better file-sharing algorithms that optimize download speed by matching users who share the same interests, Duch says. The data could also be used to evaluate the efficacy of antipiracy policies, he says. In an unpublished study using the same data, the researchers found that regulations on online downloading tend to produce only short-term effects. For instance, when the New Zealand government passed a three strikes law against pirating in 2011, BitTorrent user activities dampened for a few months, only to climb back up later, Duch says. “Law enforcement agencies … are trying to put walls in a field where there are several ways to evade them.”

*Correction, 7 October, 12:15 p.m.: This article has been corrected to note that BitTorrent user activity dampened when the New Zealand government passed a three strikes law against piracy, not when the U.S. Department of Justice shut down Megaupload.