An analysis of the RIAA's complaint against Dan Peng '05

As the Daily Princetonian reported on the 4th, Dan Peng '05 is being sued by several RIAA members for allegedly operating "wake," a site which, according to the plaintiffs, facilitated copyright infringement.

I recommend reading the Daily Princetonian article for background. Briefly, wake.princeton.edu was Dan Peng's computer. He allegedly ran a service called "wake" that periodically scanned Princeton's Microsoft Windows File Sharing (SMB) network and indexed the names of and sizes of guest-accessible files in a database. Interested parties could go to http://wake.princeton.edu and search this database for words in the titles of files they were interested in. Wake did not store the contents of the indexed files on its system. Rather, it returned the addresses of the files on the campus network. If a user wanted to look at the files their search turned up, he or she would click the address supplied, and his or her web browser would take them there. Wake's role was confined to the indexing -- once the results page appeared on the user's screen, all of the other communication was carried out between the user's computer and the computer holding the files they wanted.

Throughout this article, references to "Wake-like" systems should be taken to mean SMB (Windows File Sharing) indexing services as described above. All of the systems with which this author is familiar follow that pattern.

According to the Daily Princetonian article, the Wake webserver was visible "outside" the campus network (in other words, it could be accessed from machines outside of the Princeton University community). The actual files linked to in Wake search results were not, as Princeton blocks SMB traffic at the campus's Internet gateway.

With the preliminaries established, we'll begin to analyze the complaint.

[Header material omitted.]

NATURE OF THE ACTION

1. This is an action for direct and contributory copyright infringement arising out of the knowing and willful conduct of Defendant. Defendant has hijacked an academic computer network and installed on it a marketplace for copyright piracy that is used by others to copy and distribute music illegally. In addition to operating this piracy marketplace that facilitates direct copyright infringement by others, Defendant is committing direct copyright infringement himself by copying and distributing hundreds of sound recordings over his system without the authorization of the copyright owners.

We can take the paragraph's suggestion and split the complaint into two pieces. First, plaintiffs allege that the defendant has committed contributory infringement by running a search service. Second, plaintiffs allege that the defendant committed direct infringement by "copying and distributing hundreds of sound recordings over his system without the authorization of the copyright owners."

The former complaint shall be the focus of this analysis. The latter complaint (of direct infringement) is less legally interesting: precedent in cases of direct infringement is much more clear than it is for contributory infringement. Determining its presence is largely a matter of fact-finding, not one of legal reasoning. As this is an analytical, rather than a factual article, we will not consider the claims of direct infringement.

2. Defendant has established, maintains, and controls this self-contained system that uses the resources of the Princeton University computer network and is accessible to all of the users of the University network. This system operates similarly to the Napster system, which was shut down by a federal court injunction due to its blatant facilitation of copyright infringement.

The Wake case shares some elements with the Napster case, if only because both services enable users to search for shared music. But they diverge in several legally significant respects:

• Napster supplied file-serving as well as file-searching-and-indexing capability to its users via its software and servers. Wake-like systems possesses only the searching-and-indexing capability. The file-serving was provided by third-party SMB client software, which is part of the Windows operating system used by many of Wake's alleged users.
• Napster permitted users to share only MP3-encoded sound files. Wake-like systems index all files, regardless of format. This gives those systems a substantially larger set of non-infringing uses.
• Napster's clientele could share files only in the context of the Napster network. Wake's alleged clientele could share files over the Princeton network without even knowing that Wake existed. In fact, file-sharing on the Princeton network was widespread well before Wake's alleged author arrived on the Princeton campus as a freshman. Even the name "wake.princeton.edu" speaks to this: it is an allusion to the better-known "sleep.princeton.edu", an indexing service described in the Daily Princetonian that began operating before the defendant even came to Princeton.

Thus, Defendant has taken a network created for higher learning and academic pursuits and converted it into an emporium of music piracy where copyright infringement is simplified down to the click of a computer mouse. Like Napster, Defendant's system provides users of the University network with infrastructure and facilities for committing copyright infringement. Those facilities include access to a central computer server, a continuously updated index of infringing recordings available on Defendant's own system, and a method that facilitates the rapid and efficient copying and distribution of those recordings.

The plaintiffs' rhetorician was in rare form, but their logician was asleep at the switch. Unlike Napster, as discussed above, the facilities supplied by the allegedly-infringing Wake system are neither necessary nor sufficient for file-sharing. Unlike Napster, Wake-like systems only provide an catalog of available files. All of the actual file sharing is handled by Microsoft Windows File Sharing programs, which run, can be run, and are often run independently of Wake-like software. Napster users, on the other hand, were dependent on the Napster company's MusicShare software to merely share files, much less search shared files. Without a Wake-like system, users can still find music by word-of-mouth, or by browsing the machines on the network using Microsoft's Network Neighborhood tool. They can search the files on individual machines using the "Start->Find..." command in Microsoft Windows. All that Wake could do was speed the process --- it allegedly performed the equivalent function of searching all of the computers simultaneously, saving the user a lot of pointing, clicking, and typing. But it did nothing that a dedicated user could not do with time, or that a savvy user could not do with her own software. In modern versions of Microsoft Windows, a user does not even need copious spare time or technical know-how: Microsoft incorporated a network-wide search tool directly into Windows XP, which functions as a personal Wake-like system.

The defendant cannot be described as having converted Princeton's file-sharing network into an "emporium of music piracy" because for his alleged search service to even function, people must have been sharing files in advance. (They were; file sharing at Princeton started long before the defendant arrived in Fall 2001, as did the first Wake-like search service.) The Web would be useful in the absence of Google.com, but Google.com would be useless if no one placed web pages online. Likewise, the alleged file sharing on the Princeton network is independent of the existence of Wake. It existed before Wake, and we've no reason to presume that it stopped now that Wake was voluntarily shut down.

Users connect to the computer server operated by Defendant at the Internet site, "http://wake.princeton.edu/," to search for, locate, and make copies of Plaintiffs' copyrighted sound recordings that are being made available by other users of the network; those copies are then distributed further for copying by other users, all without the authorization of the copyright owners.

This is a mix of half-truths and utter falsehoods. First, the allegedly-infringing Wake indexed not merely sound recordings. It indexed every publicly-shared document on the network.

Secondly, while most of the allegedly-shared content was presumably copyrighted, plaintiffs have a steep burden in demonstrating that it was all shared "without the authorization of the copyright owners". Students on the network shared directories on their hard drives containing their own documents, in which (barring egregious academic violations) they hold copyright. I assume that most Princeton students are not so schizophrenic as to simultaneously place their documents in a shared folder and decline to grant themselves the right do so. Other students might share works with authorization: Linux ISOs or other free software, music with relaxed licensing terms (Etree.org catalogs a good deal of that music), personally-created photographs, videos, and sound recordings of student groups. All of these works are shared with the copyright owner's permission. The allegation that files are shared "all without the authorization of the copyright owners" statement is false.

Furthermore, there exist fair uses of the plaintiffs' copyrighted material for which authorization need not be obtained. A student who leaves his CD collection at home can legally space-shift his media after finding it with a Wake-like system. A student who owns a CDs but no working CD drive, or has all of his music on LPs or cassette tapes but has no player at school, or prefers to listen on her computer rather than on a tape deck or turntable may legally do the same.

The second-to-last clause "those copies are then distributed further for copying by other users" is also incorrect, and even more insidiously so. Unlike Napster, KaZaA, and other peer-to-peer applications, Wake-like systems do not control what happens to downloaded files. Wake-like indexing services have no control over downloads, or where files go after they are downloaded. If a user places a file in a directory that they have explicitly shared, then, yes, it is indeed shared after they download it. But this is the user's choice, not the defendant's. Nothing about the alleged Wake system ensures that "copies are then distributed further" --- if they are, it is through the actions of the downloading user, not the Defendant or his software.

3. All of the most popular sound recordings, as well as many up and coming artists, are being infringed through Defendant's system. Schedule A to this Complaint, which is incorporated herein, provides a list of some of the songs that are unlawfully being copied over the Defendant's system ("the Copyrighted Recordings"). Among the well known artists are Bruce Springsteen, Santana, Billy Joel, Norah Jones, Madonna, Christina Aguilera, U2 and Eminem.

4. Defendant is acutely aware of the infringing activity occurring through his system since he himself is copying and distributing hundreds of sound recordings over his system without the authorization of the copyright owners. The sound recordings being distributed by Defendant himself include works by Bruce Springsteen, Everclear, Green Day, blink-182, U2, and Smash Mouth.

The plaintiffs conflate alleged contributory and alleged direct infringement. ¶ 3 meanders about works "infringed through defendant's system" without marking its claims as those of direct or of (tenuous) contributory infringement. ¶ 4 then discusses works that "he himself [the defendant] is copying and distributing hundreds of sound recordings over his system" --- which presumably means direct infringement. No indication is given as to why the alleged direct infringement in ¶ 4 should make the defendant "acutely aware" of the alleged contributory infringement in ¶ 3 --- as discussed earlier, the searching and serving elements of Wake-like systems are independent. The defendant can only see the terms for which users search, not what is actually downloaded. Even if the defendant indeed committed direct infringement by serving files without authorization from his own computer, simply being a direct infringer does not grant one the clairvoyance needed to tell if the queries coming into a search engine are ill-intentioned.

5. Defendant's conduct has caused and continues to cause Plaintiffs grave and irreparable harm.

Direct infringement carries a presumption of harm, but we have to wait for the findings of fact to be issued to make a determination of the extent of that harm (and this paper does not consider those direct infringement issues.) Proving harm via contributory infringement is a far more involved matter -- and this assertion alone doesn't cut it. Plaintiffs must demonstrate how "wake" is actually the sole cause of this harm in a network with other search engines, downloadable search tools, and the ability to search manually.

[Discussion of jurisdiction and identification of parties omitted.]

UNIVERSITY LOCAL AREA NETWORKS

31. Every university provides their faculty, students, and staff with access to computer services and resources. Such computer resources are provided to assist faculty and students in their academic endeavors, and to assist staff in the administration of university affairs. Two such computer resources are access to the Internet and provision of and access to a university Local Area Network, or "LAN."

Campus computer systems are provided to assist with academics, but are hardly restricted to academic use. According to the University's policy on the matter:

As a member of the University community, you are provided with the use
of scholarly and/or work-related tools, including (but not confined
to) access to the Library, to certain computer systems, servers,
software, and databases, to the campus telephone and voice mail
systems, and to the Internet. You have reasonable expectation of
unobstructed use of these tools, of certain degrees of privacy (which
may vary depending on whether you are a University employee or a
matriculated student), and of protection from abuse and intrusion by
others sharing these resources. You can expect your right to access
information and to express your opinion to be protected as it is for
paper and other forms of non-electronic communication.

The University grants students a "reasonable expectation of unobstructed use of these tools [telephones, the Internet]," knowing that the former will be used to call both classmates (academic use) and parents (nonacademic use), and the latter will be used both for class research (academic use) and reading Slashdot (nonacademic use). The University does not restrict use of the network for legal, non-academic file-sharing, so long as its bandwidth use is not excessive. (See the discussion of ¶34.)

32. Access to the Internet by university Internet users is typically done through a T1 or T3 connection that provides for very high speed access, allowing use of the Internet in a very quick and efficient manner that is far superior to much slower "dial up" access over telephone lines. Many university Internet users use their high speed Internet access not for academic pursuits, but to connect to and use certain Internet "peer-to-peer" networks for the purpose of unlawfully copying and distributing copyrighted sound recordings and other copyrighted content to and from their personal computers. The most notorious of such systems was the Napster system, which was shut down by a federal court injunction. Other similar systems such as the KaZaA system are the subject of pending litigation in the U.S. District Court for the Central District of California brought by copyright holders including Plaintiffs here.

33. Copyright infringement over university networks has become a serious problem for Plaintiffs and other copyright owners, as well as for universities. Recent statistics have indicated that, at some universities, nearly 50 percent of the available Internet computer resources (known as "bandwidth") are being used for the unauthorized copying and distribution of copyrighted material. Such conduct has greatly taxed the capacity of university computer network resources and has greatly increased the cost of providing and maintaining those resources. Additionally, these file-trading systems raise liability and security concerns for universities. In response, many universities have restricted or banned entirely access to such infringing peer-to-peer systems from computers that are part of the universities' networks.

"Many universities" have indeed banned peer-to-peer systems. Princeton is not one of them, perhaps because they prefer not to expose themselves to vicarious liability by demonstrating that they, as a matter of policy, can control network traffic. Princeton's policy on the use of peer-to-peer technology is best summed up by an excerpt from a Help Desk solution from the Office of Information Technology:

Synopsis:
MP3's: Are they illegal to have on my computer?

Solution:
Please be aware that Tools like KaZaA and eDonkey are not illegal, but
they can be used for illegal activity. Do you have permission from the
rights-holders for all the material to take and/or give copies of it?
If not, OIT cannot assist you in your endeavors. To do so would be to
participate in violating federal law, and in acting as agents of the
University the Help Desk is forbidden to do so.

Princeton recognizes that peer-to-peer clients have both legal and non-legal uses, but does not prohibit them outright. As for restrictions, see the next paragraph.

34. For example, in or around September 2002, administrators at Princeton University warned students that those who continued to connect to and use these Internet peer-to-peer systems could be placed on a slower part of the university's network.

Not quite. The University has no restrictions on peer-to-peer downloading. The policy to which plaintiffs allude is only a restriction on uploading, not "use" in general. Another Help Desk solution explains the policy:

All University computers are required to restrict file serving to one
upload at a time.

To confirm that they are unconcerned with downloading, they note:

Please note that limiting the number of uploads does not affect
your ability to download files. OIT strongly recommends that you do
not share files from your computer. This is not a requirement, but it
will help protect you from viruses and it will improve the quality of
internet service for all Princeton users. To do this, follow the
instructions below and change all simultaneous uploads to zero.

The solution goes on to explain how to disable uploads in Morpheus, KaZaA, BearShare, AudioGalaxy, and Gnucleus. The University acknowledges and permits legal file sharing, and even explains how to configure file sharing clients to be good citizens of the Princeton network. Only students who ignore this advice risk being "placed on a slower part of the university's network."

35. In part, at least, because such restrictions limit the ability of university network users to illegally copy music and other copyrighted materials from outside the university network (over Internet peer-to-peer systems), some individuals have sought to replicate such peer-to-peer systems (i.e., Napster equivalent systems) within the university network. This is done by creating, maintaining, and operating the equivalent of such systems on LANs provided by universities to their faculty, students, and staff. A LAN is a computer network that is dispersed over a limited, defined geographic area, such as a university campus, through which a user computer can interact with all other user computers on the network. In many respects, a university LAN operates as a "miniinternet" for university users only.

36. Without a Napster equivalent system, LAN users cannot effectively search for and transfer song recordings over the network. On a typical LAN, there is no centralized index of all sound recordings that users have designated for copying and transfer over the network. As a result, there is no efficient way for a user to know which sound recordings are available on which computers, or for users to gain access to available sound recordings.

¶ 36 is yet another half-true, half-false paragraph. It is partially true that "LAN users cannot effectively search" a Windows File Sharing network without a network-trawling search system in place. Searching without one is inefficient, but only for users of older operating systems. With Microsoft Windows XP, "LAN users" can "effectively search" without a centralized "Napster equivalent system": the XP operating system includes LAN search software.

Furthermore, the ability to "effectively...transfer song recordings" (or any other file, for that matter) over the network is unaffected by (and indeed, completely independent from) network search software. If one is informed of the address of a file via an email, a web page, a phone call, a flier, an advertisement, or a word-of-mouth communication, one can easily access it directly. If one learns of a computer's name or discovers it with the Network Neighborhood tool, one can search it with built-in Windows directory search functions. (The Daily Princetonian article to which defendants refer later mentions a student whose network scouring involves "manually search[ing] through files." It's not impossible to search the network by hand -- it's just time-consuming if one doesn't have Windows XP.)

The punch line is that search and transfer are distinct. Transfer, which is necessary for indexing and searching, (the converse is emphatically not true) was supplied by the network's users and their computers long before Wake was established. This case differs fundamentally from Napster's, as Napster also supplied a transfer conduit. Wake-like systems merely catalog the information made available in the conduit constructed by Princeton University and populated by Windows-File-Sharing users with software from Microsoft Corporation. It indexes a preexisting network, much as Google.com rides atop the World Wide Web, indexing the documents therein.

DEFENDANT'S INFRINGING CONDUCT

37. Defendant created a system that is designed to enable and facilitate the widespread unauthorized copying and distribution of sound recordings over Princeton University's LAN. Defendant has installed, operates, and maintains a computer server that provides indexing and search processing functions for users of that LAN. Defendant's server actively scours the network for files that others have designated for copying and distribution, and indexes the names of those files even without the knowledge or acquiescence of network users who have so designated those files, and without the consent of the copyright owners of the works embodied in those files.

Wake-like services operate under the same presumption that Web search engines (Google.com, for instance) do --- they assume that if a file is shared globally (are "designated for copying and distribution"), it is fair game for indexing. If a user does not want to share her files, she simply declines to "designat[e] [them] for copying and distribution." If a user wants to share her files but does not want them indexed, she takes standard, readily-agreed-upon measures to prevent it. On the Web, he or she creates a "robots.txt" file: a set of instructions for "robots" (computers that search the Web) telling it which files it may and may not access. On a Windows File Sharing network, there are informal conventions. Those who wish to shut out Wake-like search engines but leave access open to humans might create a world-readable "humans.txt" file with the password to access the non-world-readable files inside. Humans could read the file, Wake-like engines would not. Alternatively, server operators might simply ask search engine operators to stop indexing their sites. Plaintiffs provide no evidence that the defendant received any such requests.

The Internet operates on the presumption that publicly-readable information is intended to be publicly-accessible and publicly-searchable, unless otherwise noted. Princeton's Office of Information Technology provides instructions in designating shared directories and restricting access to non-shared directories.

In regards the "without the consent of the copyright owners" provision, it is first of all incumbent upon he who serves the files to restrict access to them if the copyright owner declines to authorize distribution. Secondly, Wake-like engines do not actually copy the files themselves, just the metadata (names and sizes.) The name of a file is an uncopyrightable fact, regardless of the file's contents. It is at most a trademarked term, but as Wake-like engines undeniably use those terms descriptively, there is nothing in the law to enjoin Wake from indexing those names. (We leave aside the fact that file names are often assigned by the server operator herself, not the copyright owner.)

If the plaintiffs wanted to stop the defendant from indexing infringing material, they might have told the users of the network to stop providing it illegally.

Defendant's server inventories the music files each user has designated for copying and distribution, maintains a centralized index of the names of those music files, and makes that index available to users of the LAN. In this manner, files that a user maintains on his or her hard drive are made available for copying and distribution by all users of the LAN regardless of the intention of the users who initially designated those files, and often without their knowledge.

Again, standard and settled practice on the Internet is to take people at their word: if a file is marked as publicly accessible, it is meant to be publicly accessible. If it should not be publicly accessible, it is incumbent on its server's operator to not designate it as accessible.

38. Defendant further has established and maintains an Internet site containing a copyright notice that is accessible over the World Wide Web at the URL >http://wake.princeton.edu/<. By accessing that web site, users of the LAN search for and locate sound recordings that Defendant has indexed for copying and distribution by typing in search terms into a search window provided by Defendant. Results of the search are then returned to the user. These results include the file names of the sound recordings that match the search term and the location on the LAN of users' computers that are making those sound recordings available for copying and distribution over the LAN.

That almost describes a Wake-like system from the user's point of view. Plaintiffs neglect one fact: Wake-like systems, like any good search engine, index not merely sound recordings, but all shared files.

A user need only click on a particular search result, and the file containing the sound recording is automatically downloaded - i.e., copied and saved - directly from the offering user's computer to the hard drive of the requesting user's computer.

With appropriate software, the file is downloaded with a single click. This is due more to Microsoft than to the Wake-like system -- once Microsoft's Internet Explorer is given an address of a file or directory on a LAN, it handles the work of downloading and saving the file. All the search system does is supply the address. The same can be said of any standard Web search engine, like AltaVista.com or Google.com.

39. Through the centralized indexing, searching, and file transfer functions provided and maintained by Defendant, Defendant has functionally replicated the unlawful systems that facilitate and enable the massive infringement of copyrighted sound recordings on the Internet, and that have been enjoined by courts in other cases, such as Napster, Aimster, and AudioGalaxy. The centralized index, search, transfer, and other functions provided by Defendant are equivalent to the functions that were provided by the notorious Napster system, which was enjoined and ultimately shut down as a consequence of the copyright infringement that it enabled and facilitated.

Some elements of ¶ 39 are true, but others are false, which renders the conclusion invalid. Wake-like systems provide indexing and search, certainly. They most certainly do not provide file transfer capability (which is needed to perform indexing). That is provided by Windows File Sharing. In that respect, the defendants' system is radically different from the other, monolithic systems they discuss.

In at least one way, however, Defendant's system is even a more efficient and effective infringement system than was Napster's. In Napster, a user affirmatively was required to log on to the system in order to make available MP3 files for copying and distribution. As averred above, however, Defendant's system actively searches for, inventories, and indexes all filenames that are being made available on the network regardless of whether the user offering the files has intended them to be made available to all other users on the network or not.

Incorrect. A grossly misconfigured system will make files available without its owner's knowledge. Likewise, a grossly misconfigured machine on the Internet could expose its every secret to the Internet at large. Again, settled convention presumes that files are shared only intentionally. Princeton's Office of Information Technology explains how to restrict access to files -- and only a truly incompetent computer manufacturer will ship a workstation with all files shared by default. A user must take affirmative steps to share them. Even the plaintiffs admit this in ¶ 37 when they discuss "files that others have designated for copying and distribution" (emphasis mine.)

40. Defendant is well aware of the widespread piracy of copyrighted recordings over his system. Defendant maintains on his web site the "top 20 searches" that have been carried out on his system, and that list shows that a search for the popular artist, Eminem, is one of the more frequent searches executed on Defendant's system. Also, Defendant's system is even promoted as a source for "MP3" files in an online publication called "The Princeton Pauper." Defendant should be aware of this promotion as his own web site provides a direct link to this online publication. Most telling of Defendant's knowledge of the widespread piracy conducted over his system is that Defendant himself uses this system to distribute MP3 files, many of which contain songs owned by Plaintiffs. Defendant has downloaded over his system hundreds of songs owned by Plaintiffs and many of the files that Defendant has made available for widespread distribution on his system are copyrighted recordings.

We'll lay aside the direct infringement claims, as they rest entirely on factual conclusions. Unfortunately, they're intermixed with the contributory infringement claims here here. Plaintiffs claim that the defendant's alleged personal direct infringement somehow gives him knowledge that the search system is used for "widespread piracy." Again,

• Once a search returns its results, a Wake-like system's role in finding the file is complete. The system operator does not know if the user actually downloaded any of the files listed in the result.
• Even if the system operator did know about actual downloads, he or she would not be able to distinguish legal from illegal downloads. Even copyrighted songs are available under the fair use doctrine to a user who owns the CD and who wishes to space-shift her collection.

The Pauper's claim that "wake" offers "MP3" files is no more an indictment of the site than a statement that mp3.com provides "MP3" files is an indictment of mp3.com. Although the RIAA has done much to conflate the two, the term "MP3" is not a synonym for "copyright infringement". Even if the Pauper had described wake.princeton.edu as an "emporium of music piracy", the fact that the endorsement existed would not provide the defendant with the knowledge of "actual" individual infringing actions that section (1)(C) of the DMCA (see below) requires.

Finally, note that plaintiffs carefully avoided saying that the defendant linked to the advertisement in the Pauper, but only that he linked to the magazine itself. They merely note that "his own web site provides a direct link to this online publication". He may have provided the link on his web page simply because he likes the magazine, rather than because of a negotiated reciprocal-linking agreement.

41. Defendant's web site also has been the subject of an article in the university newspaper, "The Daily Princetonian." In November 2002, the newspaper highlighted the centralized indexing and searching functions provided by Defendant's web site that enable students to more efficiently "get music" on the university's network.

Plaintiffs are presumably referring to this article which appeared on November 20. The article mentions "wake" among several other sites that index the Princeton network. The article supports viewing Wake as a filename-indexing engine.

42. The system that Defendant has created, controls, and maintains thus enables and facilitates the large-scale unauthorized copying and distribution of the copyrighted recordings over Princeton University's LAN. Such infringements would not occur on the scale or the manner in which they do without the conduct and participation of Defendant, who provides the infrastructure and technological means to accomplish these infringements. Schedule A provides only a subset of the recordings that are being unlawfully copied through Defendant's system. As detailed in Schedule A, the Copyrighted Recordings copied and distributed without authorization over Defendant's system include many of the most popular recordings by the most popular recording artists, including, but in no way limited to, Bruce Springsteen, Santana, Billy Joel, Norah Jones, Madonna, Christina Aguilera, U2 and Eminem.

Even if the obvious inaccuracy is laid aside (the assertion that the defendant "provides the infrastructure and technological means to accomplish these infringements" which ignores the fact that the infrastructure existed long before he arrived on the campus in the Fall of 2001), the defendant's "conduct and participation" are hardly essential to file-sharing at Princeton. First, wake is (allegedly) one among many Wake-like services: a Daily Princetonian mentions article several such sites, one of which (sleep.princeton.edu) existed long before wake. (The name "wake" is a play on "sleep.") Second, the means to create one's own private or public index server are readily available for download and installation. An enterprising student could easily set up such a server in an afternoon. The plaintiffs' assertion that "Such infringements would not occur on the scale or the manner in which they do without the conduct and participation of Defendant" is most emphatically false.

A quick web search reveals no less than eight SMB-indexing engines. Zaval File Search, Fast File Search, FemFind, mp3spider, PySMBSearch, Samba Database, Seek42, and UntzUntz LAN Scan are listed on Freshmeat.net, which bills itself as (among other things) "the Web's largest index of Unix and cross-platform software". A colleague suggested five more: Netropolis, Strangesearch, Celery/Stalk, Phynd, and Dormspider. (Phynd and Dormspider both depend on Celery/Stalk.) Finding more such programs is left as an exercise to the reader. Incidentally, if they use Microsoft Windows, readers need not look very far: Windows XP includes such a tool.

43. Defendant's unlawful conduct does not end there, however. Defendant himself is distributing hundreds of copyrighted sound recordings without authorization that he unlawfully downloaded and thus copied. From a web site he maintains, Defendant has distributed and continues to distribute, for unauthorized copying by others, hundreds of copyrighted sound recordings, including works by Bruce Springsteen, Everclear, Green Day, blink-182, U2, and Smash Mouth. Through this conduct, Defendant is distributing, without authorization, Plaintiffs' Copyrighted Recordings for copying by anyone on the LAN who accesses his web site.

Again, this allegation of direct infringement is a factual matter that does not touch on any unresolved legal questions. This paper does not discuss it.

[First claim for relief (regarding direct infringement) omitted; see above.]

SECOND CLAIM FOR RELIEF

CONTRIBUTORY INFRINGEMENT OF COPYRIGHTS [By All Plaintiffs Against Defendant]

52. Plaintiffs incorporate herein by this reference each and every averment contained in Paragraphs 1 through 43 inclusive.

53. Defendant has knowingly and systematically induced, caused, and materially contributed to the above-described unauthorized copying and distribution of the Copyrighted Recordings and thus to the infringement of Plaintiffs' copyrights and exclusive rights under copyright in the Copyrighted Recordings.

54. The infringement of each of Plaintiffs' rights in and to the Copyrighted Recordings constitutes a separate and distinct act of infringement.

55. The foregoing acts of infringement by Defendant have been willful, intentional, and purposeful, in disregard of and indifference to the rights of Plaintiffs.

56. Defendant's conduct constitutes contributory infringement of Plaintiffs' copyrights and Plaintiffs' exclusive rights under copyright in violation of 17 U.S.C. sections 106 and 501.

Contributory infringement requires three factors: that an act of direct infringement took place, that the alleged contributor knew about the act, and that the alleged contributor facilitated the act. (See ChillingEffects for a high-level description. See also part IV of the Napster judgment: "Traditionally, 'one who, with knowledge of the infringing activity, induces, causes or materially contributes to the infringing conduct of another, may be held liable as a `contributory` infringer.' Gershwin Publ'g Corp. v. Columbia Artists Mgmt., Inc., 443 F.2d 1159, 1162 (2d Cir. 1971)").

In this complaint, plaintiffs have not identified any of the supposed direct infringers beyond the defendant. Without such identification, their case rests on conjectural infringements by wake's alleged clients.

The Digital Millennium Copyright Act provides a limitation on contributory-infringement liability for search engines if they follow certain rules, listed below.

TITLE II--ONLINE COPYRIGHT INFRINGEMENT LIABILITY LIMITATION
SEC. 201. SHORT TITLE.
     This title may be cited as the ``Online Copyright Infringement
Liability Limitation Act''.
SEC. 202. LIMITATIONS ON LIABILITY FOR COPYRIGHT INFRINGEMENT.
     (a) IN GENERAL.--Chapter 5 of title 17, United States Code,
is amended by adding after section 511 the following new section:
``§ 512. Limitations on liability relating to material online

     --intervening text omitted--

      ``(d) INFORMATION LOCATION TOOLS.--A service provider shall
not be liable for monetary relief, or, except as provided in subsection
( j), for injunctive or other equitable relief, for infringement of copy-
right by reason of the provider referring or linking users to an
online location containing infringing material or infringing activity,
by using information location tools, including a directory, index,
reference, pointer, or hypertext link, if the service provider--
           ``(1)(A) does not have actual knowledge that the material
      or activity is infringing;
           ``(B) in the absence of such actual knowledge, is not aware
      of facts or circumstances from which infringing activity is
      apparent; or
           ``(C) upon obtaining such knowledge or awareness, acts
      expeditiously to remove, or disable access to, the material;
           ``(2) does not receive a financial benefit directly attributable
      to the infringing activity, in a case in which the service provider
      has the right and ability to control such activity; and
           ``(3) upon notification of claimed infringement as described
      in subsection (c)(3), responds expeditiously to remove, or disable
      access to, the material that is claimed to be infringing or
      to be the subject of infringing activity, except that, for purposes
      of this paragraph, the information described in subsection
      (c)(3)(A)(iii) shall be identification of the reference or link, to
      material or activity claimed to be infringing, that is to be
      removed or access to which is to be disabled, and information
      reasonably sufficient to permit the service provider to locate
      that reference or link.

There are two ways to gain the search-engine protection of the DMCA: one can either take the no-knowledge-of-violations route in (A)-(B) or the will-remove-violations route in (C). One must also meet parts (2) (no profits from infringement) and (3) (compliance with takedown letters.)

The plaintiffs have claimed to have evidence the defendant cannot meet (1)(A) and (1)(B) -- namely, they cite a "top 20 searches" feature on the cite which ranks one of their artists (¶40). It would be reasonable to assume that the defendant saw that list, and did not think that all of the files in his database that matched that keyword were non-infringing.

Google.com has a top searches page, which (by no mere coincidence) also features Eminem. In their 2002 year-end listings, they note that Eminem was the most-searched-for man on Google, under both the Web Search and Image Search categories. Obviously, Google knows that all images of the man are copyrighted under the Berne Convention, and that many of them (for instance, those published in magazines or on music-industry web sites) are probably not authorized for free distribution.

Google.com is not prosecuted for contributory infringement for linking to infringing images of Eminem because of the DMCA. First of all, there are simply too many results to vet: the Image Search lists 14,900 image hits for "eminem" (verified on 5 April 2003). Under this section of the DMCA, they do not have (A) "does not have actual knowledge that the material...is infringing" (in other words, any particular piece of material is infringing) but under (B), they know that there are countless Eminem fan-sites, many of which have infringing photos of the pop idol. Google's defense is part (C): if they are informed by the copyright holder that the material to which they link is infringing, they remove it. They provide a detailed description of this policy on their web site. Removals of material from Google by DMCA request are well documented, and indexed at ChillingEffects.

A SMB search engine that covered the undergraduate computers alone would have to search 230 computers, if we assume that 5% of the undergraduate body makes files available over SMB (a modest estimate). If each one of those made available 100 files (another modest estimate), that leaves the defendant with 23,000 files to review, many of which are ever-changing as people add or remove computers from the network, change their sharing preferences, or add or remove files from their computers. As with Google's images of Eminem, no single human being (much less a busy student) can be expected to vet all of these files for copyright violations --- even though the defendant knows that, statistically, some of them are shared in violation of copyright. He is protected by part (C), which indicates that if he is made aware of a copyright violation (for instance, by a DMCA takedown notice served to him or to the University DMCA officer in accordance with part (3)), he "acts expeditiously to remove, or disable access to, the material."

(1)(C): Plaintiffs attempt to establish defendant's knowledge of infringement (which could be applied to (1)(C)) by accusing the defendant of direct infringement in ¶40. If plaintiffs' direct-infringement claim is rejected, defendant will easily meet (1)(C). On the other hand, if the defendant is found to have committed direct infringement, he will have failed to satisfy (1)(C) with respect to his own files, which were presumably indexed by the engine, and which he knew were directly infringing.

Whether knowledge that his own directly-infringing files were indexed in the search engine is fatal to the defendant's search-engine protection is ultimately a matter of how the factors ((1)(C), (2), and (3)) are weighted by the court. If the factors are interpreted as simple Boolean tests, then defendant will lose protection, for failing to meet (1)(C). If, on the other hand, they are weighted factors (as are the four factors that determine whether a given use is "fair use"), defendant may still be able to gain search-engine protection. Given that the assumed (1)(C) violation was already covered by a direct-infringement complaint, given that factors (2) and (3) are decidedly in his favor (see below), given that the plaintiffs made no effort to go through proper, established channels to stop the infringement, defendant stands a good chance of being able to claim (at least partial) search-engine immunity if the factors are weighted rather than Boolean. Result: If direct infringement is assumed, depends on method of weighing (1)(C), (2), and (3). If factors are Boolean, favors plaintiffs. If factors have weights, favors defendant.

(2): The defendant clearly receives no financial benefit from his alleged operation of "wake," so (2) is satisfied. (This also relaxes the burden of (1)(C) --- absent revenues, the defendant has substantially fewer resources with which to investigate possible infringement.) Nor does he have the ability to control the "activity" --- his control ends when the search results are returned. He has no control over actual acts of infringement (downloads). Contrast this to Napster, where the Napster MusicShare program handled downloads. Result: Clearly in the defendant's favor.

It is the copyright owners' right to serve complaints as described in (3). However, the copyright owners never contacted the defendant to register their complaints in the manner described in (3). (If they had, they would have no reason to leave that out of the complaint, for if they had contact him and he declined to act, his site would lose this safe-harbor protection.) Nor did plaintiffs contact the University's DMCA agent regarding "wake," for if they had contacted her, she would not have called the suit a "terrible surprise". Instead, the plaintiffs decided to ignore the DMCA, and have failed to make a good-faith effort to resolve their complaint without involving the court. Result: Clearly in the defendant's favor.

As an aside, the defendant's (alleged) search engine would have made an excellent tool for the recording industry to detect the violations for which the DMCA authorizes them to issue complaints. It was, as the article notes, the only SMB search engine visible from outside the Princeton campus. The plaintiffs could have easily searched it for the files listed in Schedule A, and issued takedown notices to the University, for both the defendant, and for every directly infringing student on the campus. To further the irony, without Wake, the plaintiffs would find it more difficult to view Princeton SMB network. Had they left the service alone, it would have provided them with enough information to issue takedown notices to all of the direct infringers on the network.

The DMCA requires service providers to undertake a number of bookkeeping measures to qualify for the safe-harbor liability limitation: service providers must designate an agent to receive official complaints, establish termination policies, and accommodate and refrain from interfering with "standard technical measures" (meaning DRM). This paper does not discuss agent designation and termination policies because Wake-like systems at Princeton operated on top of a network (the University network) that already has these policies in place. The plaintiffs, having served such complaints to Princeton before, knew that complaints submitted to the designated University official would be transmitted to the alleged "wake" operator. The final requirement, non-interference with DRM, is always satisfied by Wake-like engines: contemporary DRM acts on the contents of files, Wake-like engines index only their names and sizes. Those engines have nothing to do with DRM. (Defendant may assume "standard technological measures" means only contemporary DRM, because the law defines "standard technological measures" as systems "developed pursuant to a broad consensus of copyright owners and service providers in an open, fair, voluntary, multi-industry standards process.")

In sum, the defendant, much like Google.com, could not be expected to investigate every item in his site's database, so he was not in violation of part (1)(C) for third parties. If he provided directly-infringing files, he does not meet a strict interpretation of (1)(C), but depending on how the factors are weighted, he may still retain search-engine protection. He did not profit from his site, so he satisfies part (2). Finally, he was never informed of a complaint under part (3) (which also militates in favor of a finding of lack of knowledge of specific violations in (1)(C)), so he cannot be accused of shirking the burdens associated with those complaints. The "wake" site consequently is eligible for safe harbor protection under part (C).

Given their failure to take the DMCA-required steps to inform "wake" of copyright violations, until plaintiffs demonstrate why Wake-like search engines do not meet the DMCA safe-harbor proposals, any contributory-infringement claims must be treated with extreme skepticism.

57. As a direct and proximate result of Defendant's infringement of Plaintiffs' copyrights and exclusive rights under copyright, Plaintiffs are entitled to the maximum statutory damages, pursuant to 17 U.S.C. section 504(c), in the amount of $150,000 with respect to each work infringed, or such other amounts as may be proper under 17 U.S.C. section 504(c).

58. Plaintiffs further are entitled to their attorneys' fees and full costs pursuant to 17 U.S.C. section 505.

59. Defendant's conduct, as hereinabove averred, is causing and, unless enjoined and restrained by this Court, will continue to cause Plaintiffs great and irreparable injury that cannot fully be compensated or measured in money. Plaintiffs have no adequate remedy at law. Pursuant to 17 U.S.C. section 502, Plaintiffs are entitled to preliminary and permanent injunctions prohibiting further contributory infringements of Plaintiffs' copyrights and exclusive rights under copyright.

WHEREFORE, Plaintiffs pray for judgment against Defendant as follows:

l. For a preliminary and a permanent injunction enjoining Defendant, and all persons acting in concert or participation with him, from: (i) directly or indirectly infringing in any manner any of Plaintiffs' respective copyrights (whether now in existence or hereafter created), and exclusive rights under copyright, including without limitation, the Copyrighted Recordings listed on Schedule A, and (ii) from causing, contributing to, enabling, facilitating, or participating in the infringement of any of Plaintiffs' respective copyrights, and exclusive rights under copyright, including without limitation, the Copyrighted Recordings listed on Schedule A.

2. For maximum statutory damages in the amount of $150,000 with respect to each copyrighted work infringed, or for such other amounts as may be proper pursuant to 17 U.S.C. section 504(c);

3. For Plaintiffs' attorneys' fees, full costs, and disbursements in this action; and

4. For such other and further relief as the Court may deem just and proper.

This raises the question: why do the plaintiffs deserve a preliminary injunction if they haven't even bothered to contact the University's DMCA agent (who called the suit a "terrible surprise") to tell the people actually sharing the files to stop?

Conclusion

This case fails to answer one critical question: How is a content-neutral search engine to be held liable for contributory infringement? The best that the plaintiffs did was point to the Napster case. But Napster was not merely a search engine --- it was also a file transfer utility and file sharing server. Wake-like tools provide only search functions. Plaintiffs must explain why they chose to ignore the alleged direct infringers who used the engine, to bypass University's DMCA-standard complaint-filing system, and go after the engine's alleged operator on the grounds of contributory infringement. Plaintiffs must further explain how Wake's alleged contributory infringement falls outside the "safe harbor" provided to search engines by the DMCA.

I extend my thanks to Niraj Bhatt for (among other contributions) his pointers on contributory infringement, to Joshua Tauberer and David Robinson their critiques of my reasoning, and to Joshua Stratton for pointing me to other relevant sections of the DMCA (which were also mentioned in a Slashdot comment.)

joseph@barillari.org [barillari.org]

$Id: peng.html,v 0.41 2007/01/05 06:12:06 jdb Exp $
$Log: peng.html,v $
Revision 0.41  2007/01/05 06:12:06  jdb
added link

Revision 0.40  2003/04/12 20:33:42  jdb
Clarified DMCA-related language

Revision 0.39  2003/04/12 20:13:53  jdb
updated to include /. comment on DMCA eligibility

Revision 0.38  2003/04/12 20:05:37  jdb
Strengthened DMCA analysis against DGR's critique, added material
suggested by JS.

Revision 0.37  2003/04/08 22:23:10  jdb
Updated to include Zack Rosen's research on WinXP

Revision 0.36  2003/04/08 19:29:28  jdb
s/systems systems/systems (thanks Amit)

Revision 0.35  2003/04/08 14:52:46  jdb
Omitted remark about the $4.2 billion figure on
http://www.riaa.org/Protect-Campaign-3.cfm figure being incompatible
with "Plaintiffs great and irreparable injury that cannot fully be
compensated or measured in money". (Thanks to A.C. for pointing out
that the comment might irritate some audiences unnecessarily.) AFAIK,
(c) law/precedent assumes that damage is irreparable/immeasurable for
the purpose of C&D injunctions. (Please correct me if I'm wrong.)

Revision 0.34  2003/04/08 14:47:07  jdb
s/least/lest (thanks F.M.)

Revision 0.33  2003/04/08 14:44:22  jdb
s/begs the question/raises the question/ (thanks M.C.)

Revision 0.32  2003/04/08 14:39:04  jdb
Fixed grammar in paragraph on fair use (thanks A.A.)

Revision 0.31  2003/04/07 22:10:12  jdb
Punctuation fix

Revision 0.30  2003/04/07 22:03:03  jdb
Fixed some ambiguities

Revision 0.29  2003/04/07 21:56:44  jdb
Added more info about University policy, fixed some links

Revision 0.28  2003/04/07 04:49:56  jdb
Inserted paragraph on fair use