Tuesday, March 1, 2011

.torrent = .crime

By Terry Hart
Copyhype | Article Link

I sometimes see the phrase “.torrent = .crime” used online in discussions about enforcing copyright online. It is considered by copyright critics as a dig against efforts to enforce the widespread copyright infringement occurring within the bittorrent ecosystem1 — the idea being that content producers have mistakenly declared torrent technology categorically unlawful.

Most recently, it has popped up in response to the US government’s seizure of domain names as part of Operation in Our Sites. A number of the sites targeted were part of the bittorrent ecosystem, like meta-search engine Torrent Finder. And with the pending appeal of Isohunt in the Ninth Circuit, the meme will likely persist — the torrent search engine’s appellate brief devotes considerable space to rebutting the association between torrents and infringement. Proponents of the meme believe the association is unfair and are quick to point out the many legitimate uses of the bittorrent protocol as evidence.

The snappy soundbite, however, glosses over the distinction between a technology and uses of a technology. It also relies on a fundamentally flawed premise: the fact that there are some legitimate uses of a technology does not make all uses of that technology legitimate. And within the general bittorrent ecosystem, there are a lot of illegitimate uses of the technology — so much so that the association between “torrent” and “crime” is not entirely unfair.

The Supreme Court, through its decisions in Sony Corp of America v. Universal City Studios, 464 US 417, and MGM v. Grokster, 545 US 913, maintained what is really a common sense distinction between technology and uses of that technology. Simply put, you can’t be liable simply for making something that can be used for copyright infringement, but you can be liable if you intend or encourage infringing uses.

Put another way, while direct infringement is a strict liability tort, meaning the intent of an infringer is irrelevant, secondary liability requires some form of culpability. Grokster reads Sony as saying you can’t impute this intent based solely on the characteristics of a device. But the rule in Sony doesn’t mean a manufacturer or service is completely precluded from liability if it satisfies the Sony safe harbor — a manufacturer or service provider can still be liable if it is found to be culpable in other ways, including whether it induced infringement.

The Supreme Court in Grokster noted three features of the evidence of intent that most showed culpability. Satisfying a known source of demand for copyright infringement, no attempt to filter or develop mechanisms for diminishing infringing activities, and selling advertising based on high-volume use — revenues increase the more people use it, and the overwhelming amount of people use it for infringing purposes. Any one of these alone is not enough to show culpability, but taken together, the intent to encourage infringement is, in the words of the Court, “unmistakeable.”

The torrent ecosystem facilitates piracy — not by design, but by use.

I’m talking about just the torrent ecosystem here — the trackers, hosts, and search engines that do little if anything to ensure their services are not used to aid in distributing unauthorized content. BitTorrent as a protocol is in no way a problem. Facebook, Twitter, and other large-scale systems use the BitTorrent protocol to distribute software updates to their servers; the same is true of gaming companies like Blizzard Entertainment, makers of Worlds of Warcraft. Many Linux distros are promulgated via bittorrent. The technology offers several advantages over other methods of distributing large amounts of data.

But outside these specific uses is a world of public torrent hosts that allow anyone to upload torrent files of any type of content, no questions asked, and search engines that can help users locate whatever they want. And it is here where, any way you look at it, “.torrent” actually does equal “.crime”. The overwhelming majority of what is being distributed in the general bittorrent ecosystem is infringing.

How overwhelming? According to the available research, anywhere from 90-99% of files available in the torrent ecosystem are infringing.2 A study released January 2011 discovered only 1 legitimately offered file in the top 10,000 torrents offered by one of the largest public bittorrent trackers.

In the face of such overwhelming infringement, it’s absurd to think you can design a site or service that facilitates access to this universe of material, without any attempt at filtering or mitigating infringement, and escape liability. The fact that there are some legitimate uses for the technology does not mean that all uses of the technology are legitimate. Search engines, trackers, and torrent hosts exist for verified and authorized content. But that fact doesn’t save the general and open-for-all torrent sites; neither does the fact that the protocol is used by large-scale systems for distributing software updates. It’s not like Facebook and Twitter are going to Torrent Finder to find new patches.

The distinction between legitimate and illegitimate uses of bittorrent is a distinction between services where the use of the protocol is secondary to the service and the ability to pirate through the service is diminished and minimized, and services that tap into the general bittorrent ecosystem. The latter services may claim to have legitimate uses, but these claims are drowned out by the evidence.

The business model of these sites depends on popular — i.e., copyrighted — content. It’s the major draw: Torrent Finder wasn’t making $15,000 a year advertising to people looking for the latest Linux distros or searching for unsigned bands. Isohunt’s Gary Fung admitted as much during the course of that litigation.3 Courts looking at nearly every other major file-sharing service that has found itself subject to a lawsuit have come to the same conclusion: ”While there is doubtless some demand for free Shakespeare,” said the Supreme Court in Grokster, ”the evidence shows that substantive volume is a function of free access to copyrighted work.”4

Sites that have gone legit have discovered just how true this is. Search engine YouTorrent reported that it lost over half its traffic when it stopped indexing files from sites like PirateBay and only returned verified authorized and open-license results.

Some torrent site operators have adopted a “would if they could” attitude toward copyright infringement: they’d like to stop it, but it’s impossible to know what files are authorized and which are not. This argument doesn’t hold up against the significant percentage of infringing files available.

Some theories of secondary liability, like contributory infringement, require that the secondary party has some knowledge of the underlying misconduct, with the caveat that the secondary party cannot escape liability by being “willfully blind.” You can’t, like an ostrich with his head in the sand, deliberately ignore bad behavior and expect to stay out of trouble. While the question of willful blindness can often be a tough one to answer, courts haven’t had trouble in cases of file-sharing services. With such a high percentage of infringing files present on all these services — 89%, 95%, 99% — you can’t claim ignorance of infringement.

Anyone who bristles at the association of bittorrent with copyright infringement has to admit, if honest, that within the general bittorrent ecosystem, torrents have largely become synonymous with distributing unauthorized content. The technology itself is neutral, it can be put to legitimate or illegitimate uses, but any service wanting to become a component part of the unregulated, open, and public torrent universe has a high burden for establishing legitimacy.



  1. The bittorrent “ecosystem” describes the multitude of independent components — from torrent file hosts, to torrent search engines and meta search engines, to trackers — that together make the bittorrent protocol work.
  2. Robert Layton & Paul Watters, Investigation into the extent of infringing content on BitTorrent networks, Internet Commerce Security Laboratory (2010), confirms 89% of all torrents in sample infringing; Sauhard Sahi & Ed Felten, Census of files available via BitTorrent, (2010), concluded 99% of files likely infringing; Columbia Pictures v. Fung, Order granting Plaintiffs’ motion for summary judgment on liability (2009), evidence showed approximately 95% of files available on defendant’s torrent search engines infringing.
  3. Columbia Pictures v. Fung, p. 34. 
  4. 545 US at 926.