Wednesday, August 4, 2010 - The Progress & Freedom Foundation Blog

More Data Confirms: File-Sharing Is Really About Piracy

Recently Nate Anderson of Ars Technica published a story entitled Only 0.3% of the Files on BitTorrent Confirmed to be Legal. The title tells the tale: Australia's Internet Commerce Security Laboratory just released another study of the content of a "decentralized" file-sharing network; it confirms, yet again, that these networks and programs are used almost exclusively for copyright piracy. Here is a summary of the finding of existing studies on how decentralized file-sharing programs are actually used:


Protocol(s) Studied

% of files found infringing

MGM Studios Inc. v. Grokster, Ltd., 454 F. Supp. 2d 966, 985 (C.D. Cal. 2006)


90% of available files

97% of files selected for downloading

Columbia Pictures v. Fung, 2009 U.S. Dist. LEXIS 122661 (C.D. Cal. 2009)


95% of files downloaded

Sahi/Felton study


99% of available files

Arista Records, Inc. v. Lime Group, 2010 U.S. Dist. LEXIS 46638 (S.D.N.Y. 2010)


98.8% of files selected for downloading

Internet Commerce Security Laboratory


99.7% of the most actively seeded torrents

The consistency of these results is telling, as is Illinois State University's "Digital Citizen Project," which collected empirical data on many uses of programs and protocols. See Alexandre M. Mateus & Jon M. Pena, Dimensions of P2P and Digital Piracy in a College Campus, 1, 21, 29 (TPRC 2008). It did not compile percentage-of-infringing-use statistics, but its results were clear enough:

Collectively, the findings of the Grokster, Digital Citizen, Fung, Sahi-Felten, Lime Group, and Internet Commerce Security Laboratory studies send a simple message with broad implications: "Decentralized" file-sharing networks, programs and indices have never really been "neutral" or "dual-purpose" technologies: They are so optimized for piracy that they are almost never used for lawful purposes because they are unsuited to the needs of lawful exchange. To see why, review the following hypothetical and then answer the question posed.

Hypothetical: Your friend Perry tells you that he is going out to buy some marijuana so he can get stoned, and some oregano so he can season his soup. You then see Perry slink into a dark alley where anonymous strangers wearing ski masks lurk in the shadows. Perry whispers to a masked stranger, and then receives from him a bag containing some dried, greenish herb that Perry immediately stuffs into his pocket before leaving the dark alley. Question: Do you think it more likely that Perry just tried to buy (A) the illegal marijuana or (B) the legal oregano? The circumstances of the exchange make the answer to this question really obvious.

Piracy-adapted "decentralized" file-sharing programs, networks, and tracker sites are much like Perry's dark alley: For example, a program like LimeWire program lets tweens, teens, and college students play Russian roulette with a notoriously deceptive program in hopes of getting files that may not be what they seem from anonymous strangers who can almost immediately become untraceable. The 98.8% infringement rate in Lime Group thus merely confirms what our common sense would already suggest: if some "innovator" like LimeWire's Mark Gorton creates an online analog of a dark alley full of masked strangers, people will studiously avoid it—when they are engaged in lawful exchange.

In light of these studies, and a parade of cases like Napster, Aimster, Grokster, UseNet, Fung, and Lime Group, litigating inducement cases against similar bad actors should get much simpler. There should be less need for complex discovery, because evidence of intent will be far easier to infer from simple data about how the service and its operator behave. The law infers that people intend the obvious consequences of their actions, and the consequences of distributing a program like LimeWire or operating a site like Isohunt are now really obvious.

Fortunately, the now-obvious consequences of dark-alley designs will usually be only one of many signs that a given file-sharing program or tracker site is a specialized piracy machine. For example, a program distributor's or site operator's efforts to affect what EFF called "plausible deniability" will provide a clear signal of intent. In other words, "innovators" who think that they are building lawful businesses should try hard to understand when, why, and how consumers are using their programs or services. Consequently, when "innovators" are instead trying hard to avoid knowing when, why, and how consumers are using their programs or services, they must know that they really, really don't want to know this information. The law infers intent from such conduct. Nevertheless, we just saw the nitwit CEO of LimeWire LLC try to convince us that throughout the last decade, his company never had a clue as to why its program became so popular, particularly after Grokster caused so many other distributors of functionally similar piracy machines to fold or filter:

Whatever the numbers of files authorized for sharing versus those that are not, LimeWire does not know those numbers. It did not in 2000 and it does not now. LimeWire's searching and sharing functions are entirely decentralized. After downloading and installing LimeWire on their computers, we currently have no visibility into what types of content users seek, send and receive with the software.

Not even EFF could still find that decade-long affectation of "deniability" to be "plausible." The good thing about intent-based inducement liability is that you cannot "design around it": The harder you try, the more clearly you reveal your malign intent.

N.B.: As always, the preceding discusses particular implementations of peer-to-peer networking technologies. That does not mean that ,em>all implementations of such technologies are suspect. For example, the growth of IPTV might very well result in the implementation of some sort of distributed or peer-to-peer distribution system--perhaps even one implementing the BitTorrent protocol. If that happens, I predict that we would have no difficulty distinguishing that productive implementation from some of the piracy-driven implementations discussed above.

posted by Thomas Sydnor @ 11:14 AM |