Counsel for Technology, Entertainment, and Business
This past Thursday I gave a talk called “From Pages to Platforms” about the ways that Web 2.0 has affected Internet Law. Here it is in webcast form.
I recommend downloading (instead of allowing it to play in the browser) for a full-screen picture.
On this week’s Singularity Law Podcast, Michael and I briefly discussed some of the conflicts that exist between various copyleft licenses. Since then, I’ve had a few conversations with friends and attorneys (one at a job interview, in fact) in which I’ve detected some significant misunderstandings about the way these licenses relate to each other. Among many people–even some IP law practitioners–there’s a notion that “open source is open source.” This couldn’t be more wrong.
Copyleft licenses all have one big thing in common: they remove particular common copyright law restrictions on copies of works. For most works protected by U.S. copyrights, the owner of a copyright uses the law to prohibit others from reproducing, adapting, or distributing copies of his work. But copyleft licenses grant some or all of those rights to every person who receives a copy of the work. It’s wrong to think of this as the copyright owner surrendering one or more of his exclusive rights. The copyright owner is simply offering a license to use the work in one of a certain number of ways. Subsequent uses of the work are then “licensed uses” so long as they comply with the terms of the license selected by the copyright owner. But if someone uses the work without complying with the license, then his use is an infringing one.1
So, copyleft isn’t an absence of copyright. In fact, copyleft licenses rely on copyright law to enforce their provisions, which are usually intended to ensure that subsequent users of the work will always have the same rights to a work as did the generations before them.
The problem here is that these licenses are not created equal.
One common family of licenses that is used for Web content is the Creative Commons series. These are designed to encourage creativity among Web users by making content available with only “some rights reserved.” But there isn’t just one license called “Creative Commons;” even in this single “family” of licenses there are a number of versions available. For example, a copyright owner could decide to require attribution on each copy made, or require that the work and derivatives only be used for non-commercial purposes. The copyright owner could even prohibit modification of the work entirely. One provision allows licensors to require that licensees “share-alike,” meaning that copies and derivatives of the licensed work must be licensed under identical terms.2
Compare the Creative Commons licenses to the GNU General Public License (GPL) or the GNU Free Documentation License (GFDL) and you’ll find that there are stark differences that often make it impossible to create “mixed” works that incorporate content licensed under different terms. Michael Scott, my co-host on the Podcast, learned this the hard way when he sought to use material from the EFF’s website (licensed under a Creative Commons license that requires attribution and that all uses be noncommercial) on his IT Law Wiki, a reference project that makes all material available under the GFDL (which does not require attribution and permits commercial uses). This kind of “mixed” work couldn’t be done without asking the copyright owner of the original work to waive certain rights so that the material can be relicensed under new terms.
Similar incompatibilities exist between the GPL and the Apple Public Source License (APSL), the Mozilla Public License (MPL), and the OpenSSL license. Does this mean that a distribution of GNU Linux (GPL) can’t be bundled with a copy of the Firefox Web browser (MPL)? Of course not. But it does mean that a developer couldn’t use code from the Linux kernel’s TCP/IP stack to improve OpenSSL and then distribute copies of that improved software under the OpenSSL license.
Luckily, not all of these licenses are at war with each other. For reference materials and documentation, one terrific new development is section 11 of the new GFDL version 1.3, which permits most existing material that was previously licensed under the GFDL to be relicensed under the Creative Commons 3.0 Attribution-ShareAlike license. This means that a work that was published under a GFDL license prior to the start of this month can be used in many Creative Commons projects. Wikipedia, the driving force behind this revision, is licensed under the GFDL, which means that all content that was posted to that site prior to November 1 is now more compatible.3
On the software side of things, the Free Software Foundation has also made things simpler by making the GPL–by far the most popular open source license–compatible with a variety of other copyleft licenses. With other popular licenses like the two-clause FreeBSD license on the list, you might be surprised by just how usable GPL code really is.
I’m an open source advocate because copyleft licenses enable people to create tools that can change the world,4 while providing future generations with code and text that can be used to solve absolutely any problem for which it is suited, including new problems that the original author did not foresee. But I also recognize that tangled and competing licenses don’t provide any benefit to anyone. In the future I hope to see an organization like the Free Software Foundation take more of a leadership role in this area, bringing the community together so that creative people can make the most of their work.
It’s not easy being a search engine these days.
The job of a search engine is to organize information on the Web and present that information to users in a way that is meaningful and skimmable. Modern search engines like Google and Yahoo! work by scouring Web sites for information and then indexing the contents of each page in a database from which it can draw information to respond to queries. This process involves creating a cache, which is a copy of the text or other content of each Web page added to the index.
It’s certainly not a new story to hear about a copyright owner complaining about this cache. Perfect 10 litigated the hell out of this issue1 two years ago when they demanded that a court enjoin Google from creating “thumbnail” (small low-resolution) images for its Google Image Search. Perfect 10 prevailed in their initial action but then lost on appeal when the 9th Circuit found that Google’s fair use defense trumped Perfect 10’s infringement claim. This appellate decision followed the tradition of Kelly v. Arriba Soft,2 an earlier case which established that the use of thumbnail images on a search engine is a fair use.
The Kelly decision gave us a doctrine that I support, even though I have to admit I’m not entirely comfortable with it. It’s not hard to see why the Ninth Circuit found that a cached thumbnail image is a fair use. The four-pronged fair use test in § 107 of the Copyright Act weighed heavily in favor of Arriba Soft, with the all-important fourth prong–the effect of the use upon the potential market for the original–solidly on the search engine’s side because, if anything, a low-quality thumbnail preview will guide consumers to the original copyrighted works.
Fair enough. This means that we need to consider whether a copy accomplishes the goals of the original, whatever those may be. Arriba Soft’s thumbnail copies didn’t, because they were low-resolution versions of Kelly’s epic photos. But what about Perfect 10’s images? Perfect 10 is a pornographic magazine. In fact, their business model includes selling small cell-phone-sized versions of their images that are similar to the thumbnail images cached by Google, Arriba Soft, and other search engines. Clearly the thumbnails can substitute for the originals of similar size. Luckily, this isn’t why Perfect 10 lost; Google prevailed largely because the Web sites supplying the source for its thumbnails were not even owned by Perfect 10. The Court found that those third-party sites might be infringing, but that was an issue that would need to be litigated independently of Google.3
But what if a copyright owner brought suit over something like those Perfect 10 thumbnail images without the messy third-parties involved? I don’t think that wouldn’t be a fair use under § 107. It would mean that search engines are making full copies that can completely fulfill the purpose of the source images. And when a search engine copies a Web page in its entirety and makes it available in something like the Google cache, that’s also not a fair use for the same reason. That means that search engines either needed (1) another defense or (2) a license.
This is the situation that presented in Parker v. Yahoo!, Inc.4 Parker claimed that Yahoo! infringed by copying text from his Web site into its cache, but the Court rejected his argument because, according to a well-known and well-accepted Internet convention, Parker could have prevented the indexing and caching of his page by using a robots.txt file. Parker even admitted that he was aware of this. The Court reasoned that, by not including a robots.txt file, Parker granted Yahoo! and other search engines an implied license for that use.
This is a great precedent for search engines. Decisions like Parker and others have gone a long way toward making me feel comfortable with Kelly. I’d still like to see the courts square off this concept in future cases. For example, one of the characteristics of the Web is that information is released it’s very hard to contain. One of the questions that survived Parker was whether the implied license is revocable. Saying “yes” to that might be like telling a newspaper that it can recall old editions of its paper at will. But judging by the trend we’re seeing in these search engine caching cases, I’m not terribly worried.
I used to work for a video game publisher here in LA, so it’s not unusual for video game-related news and cases to come across my desk. But over the last few days I’ve seen an unusually high amount of commentary from all different corners of the industry about used game sales. I don’t think this is as big of a problem as those commentators have suggested.
The First Sale Doctrine as codified in the Copyright Act1 permits a purchaser to transfer ownership of a particular copy of a copyrighted work as long as no additional copies are retained. Game discs qualify for this,2 and so they can be freely sold. Historically, this has never been much of a problem for the industry. Unlike novels and videos which are often used once and then archived, consumers have traditionally resisted selling their used games because it’s a different kind of product. But now, with companies like Gamestop building retail businesses that aggressively promote the sale of used game discs, the situation has changed significantly. Some publishers and developers fear that an overly powerful secondhand games market could strangle the industry by cutting off consumers’ demand for new game discs. With new games costing as much as $30 million to develop in an industry that employs around 100,000 Americans, it’s not hard to understand that fear. On the other hand, novelists and movie-makers have been dealing with the First Sale Doctrine for ages, but those industries have managed to survive even with the proliferation of lending libraries and video stores.
One potential solution would be for the industry to move more toward digital downloads. Some of the most successful games in recent weeks, such as Braid and the highly anticipated Capcom title Mega Man 9, have found tremendous success as downloadable games on platforms like Nintendo’s WiiWare, Sony’s PlayStation Network, and Microsoft’s Xbox Live Marketplace. Many developers have also found success on Apple’s App Store, a service that puts application and game downloads at the fingertips of iPhone and iPod users. The use of digital downloads has a number of advantages to the industry, one of which is that downloaded games are typically locked to a single device through the use of DRM, cutting off the First Sale Doctrine at its knees.
But moving further into downloadable games also has serious consequences for the industry. While mainstream gamers have generally been accepting of DRM, with services like Nintendo’s Virtual Console reporting sales on the order of 10 million games sold per year,3 critical demographics such as the 18-35 year-old male group have expressed dissatisfaction with the DRM restrictions placed on popular games like EA’s Spore. With the entertainment sector moving increasingly away from DRM, it’s not clear that the video games industry would benefit in the long run from increased reliance on a dying technology.
Another more important problem is that the storage and bandwidth limitations of game consoles don’t always meet the needs of today’s cutting edge games. For example, most games for the PlayStation 3 ship on Blu-Ray discs that can hold up to 50 GB each. Some recent games such as Konami’s Metal Gear Solid 4: Guns of the Patriots have already managed to fill these discs. Delivering games this large over the Internet is impractical with current technology because they would take far too long for players to download and would occupy too much space on game consoles’ hard disks.
There’s no denying that this situation puts the industry in a tough spot.
I think that the best solution here would be to make some small changes to the nature of mainstream disc-based games. One of the things that distinguishes a game disc from a novel or a video is that, in many cases, once a novel or video has been used (i.e. read or viewed) once, there might not be a lot of perceived value to the consumer in using the product again. That’s why there will always be a sizable block of consumers who rent many DVDs but don’t buy any. But the video game industry can be different by crafting products that retain their perceived value even after they have been enjoyed once.
One way to do this is by providing gamers with experiences that go beyond a typical “single player campaign” experience. Nintendo’s Metroid series was a pioneer in this area; they contain items and areas that are not essential to finishing the game, but that greatly enhance and change the nature of the game if the player chooses to explore. Vivendi’s Diablo series is similarly innovative, with dungeon layouts and discoverable items that are randomly generated each time the game is played, presenting the player with a slightly different experience on each playthrough. Some other recent games have begun to emphasize Internet multiplayer capabilities that pit players with (or against) other players from around the world. This creates communities in which players build perceived value in the game over time. It also encourages players to push sales of those games through word-of-mouth advertising because players know that the more people have the game, the better it will be. Finally, another solution is to make available new content in the form of downloadables. These can be offered to players for free simply to drive sales even as the game gets older, or they can be sold to players in the form of revenue-generating micro-transactions. It’s worth mentioning that a gamer who has invested even a few dollars in game-related micro-transactions is probably far less likely to ever sell his copy than someone who has invested nothing other than the purchase price.
Americans spent nearly $19 billion on video games in 2007, and that number is set to look even better once 2008 comes to a close. Over the past two decades we’ve watched this industry adapt dynamically to a number of consumer and technological trends such as Internet piracy and the emergence of the “casual” games market. In every case the industry has emerged stronger than it was when it went in. I wouldn’t lose sleep over the emergence of a used games market.
Save your fear for Capcom’s upcoming Resident Evil 5 launch instead. If it’s anywhere near as good as its predecessor, when that hits in March 2009 I think we’ll all be having a few zombie nightmares.
Last Wednesday a federal judge declared a mistrial in Capitol v. Thomas and set aside the judgment for the plaintiffs when he found that the theory of “making available” may not have been sufficient to constitute infringement.1 The core of the RIAA’s position here is that Thomas infringed by merely offering to distribute copies of a work from her computer, even if no one ever actually downloaded a single copy. The problem with this sort of a theory lies of course in the Copyright Act. While § 106(3) provides recovery for “distribution,” nowhere does the Act permit recovery for “attempted” copyright infringement of any kind.
If you haven’t been following this case, the solution might seem obvious. Why doesn’t the copyright owner simply present evidence that the defendant actually distributed some copies? But it’s not that simple, and I don’t think the RIAA can win this time. Here’s why.
Recording companies identify plaintiffs through a fairly crude process.2 When an Internet user runs a file-sharing application that uses the peer-to-peer model, the file-sharing application typically searches the user’s computer for media files and indexes all that it finds. The index will normally contain a list of the audio files sorted by artist, album, genre, and some other criteria. It then uploads this index to another computer on the peer-to-peer network with a fast connection and large storage capacity. The computers that store indexes of other users’ files are commonly called “supernodes,” while other users are called simply “nodes.”
When one node on the network wants to find a particular recording, it submits a search query to a supernode. The supernode returns a response that indicates which computers on the file-sharing network have copies of the file along with the IP addresses of those computers. When a copyright owner (or one of its contractors such as MediaSentry, a company that provides online copyright enforcement services) wants to identify infringers, it connects to a file-sharing network as a node and searches for the name of a recording for which it owns the copyright. When the supernode returns a list of computers that are offering the file for download, the copyright owner sends another query to the supernode called “find more from this user” that returns a list of all of the files being offered from one of those computers. The copyright owner then brings suit against the Internet subscriber using that IP address for damages arising from infringement of the copyright owner’s exclusive distribution right under 17 USCS § 106(3).
Assuming that the indices generated by the defendants’ computers accurately convey the media files that are hosted by that node, the copyright owner now has an accurate list of files that the defendant is making available to the public–nothing less and nothing more. The node is offering these files for download and, if a third party wants to initiate a file transfer, such a transfer will likely occur. But neither the supernode nor the node contains any evidence that such a transfer actually took place. In other words, the copyright owner has plenty of evidence that the defendant made an offer to distribute infringing copies but no evidence that anyone actually took him up on that offer. Worse yet, given the number of nodes available on most file-sharing networks at any one time (typically millions or more), for popular sound recordings and movies, it is actually exceedingly unlikely that most nodes have ever consummated an outgoing transfer because there are so many other nodes with copies of the same content.
In a court filing, an RIAA attorney contended that ”requiring proof of actual transfers would cripple efforts to enforce copyright owners’ rights online – and would solely benefit those who seek to freeload off plaintiff’s investment.”3 If the plaintiffs continue to limit themselves to the methods I described above, I don’t doubt this. But there’s no question in my mind that “making available” is not equivalent to any of the exclusive rights enumerated in the Copyright Act.4 To win with the evidence they have, the plaintiff would need to in effect fashion a new kind of claim for attempted copyright infringement. In the Ninth Circuit, judges have already resisted such attempts. In Perfect 10 v. Amazon.com, the Court in dicta blocked what it called “the proposition that merely making images ‘available’ violates the copyright owner’s distribution right.”5 A separate line of cases has also evolved out of an Eighth Circuit case6 that has been applied to file-sharing actions. These Judges consistently find that “record companies must show that an unlawful copy was disseminated ‘to the public.’”
The problem here is that, because of the way the Internet works, we don’t really have those kinds of records in an easily accessible form anywhere. The RIAA seems to understand this. It’s one thing to use the discovery process to find out what a particular computer has downloaded. That’s trivial. But finding out what other computers have downloaded from a defendant, while not impossible, could be extremely difficult.
For this reason, I don’t think Capitol and MediaSentry can win this one.
This weblog is an informational resource only. It is not designed to offer legal advice.
Does a software developer waive his right to sue for copyright infringement by releasing his code under an open source license? This is the question to which the appellate court answered “no” today in Jacobsen v. Katzer.1
The plaintiff in this case is the manager of an open source software project whose code was improperly used by the defendant in a commercial software product that did not comply with the plaintiff’s license. But when the plaintiff brought an action for copyright infringement and moved for a preliminary injunction, the defendant argued that by releasing his code under an open source license the plaintiff waived his right to sue for copyright infringement and could now only sue for breach of contract under the license. The defendants relied on language from Sun Microsystems v. Microsoft Corp.2 that limited the rights of copyright owners to sue licensees for copyright infringement. The problem here is that the defendant — and the lower court — seems to have misunderstood an important element of how a license works.
When a copyright owner issues a nonexclusive license to a licensee, he grants the licensee the rights to use the work for some purpose. Here, the plaintiff wanted to ensure that his work remains available to the public in source code form so that anyone can edit and improve upon his work. That’s why he chose to release the software under an open source license rather than some other kind. In doing this, the plaintiff limited the scope of the license he granted the defendant as an end-user, and the defendant exceeded that scope. Legally, this means that by using the plaintiff’s code in a way that was repugnant to the open source license, the defendant’s use was not a licensed use at all.
Consequently, the central issue in this case was whether the relevant requirements in the license were conditions for the license to be valid or merely covenants to which the receiver of source code contractually agreed. A breach of a condition points toward copyright infringement liability because conditional language defines the scope of a contract. On the other hand, promissory language merely serves to create covenants that are actionable only in contract law.3 Luckily for the plaintiff, the appellate court seems to have had an easy time finding that the relevant language in the license was conditional in nature. Provisions such as “the intent of this document is to state the conditions under which [the program] may be copied” and that rights are granted “provided that” certain conditions are met weighed heavily in favor of the plaintiff in this case. This should serve as an important reminder to anyone drafting a license agreement who wants to preserve his client’s ability to sue for copyright infringement.
Practical matters aside, one of the best things about this case is that the court reaffirmed that “copyright holders who engage in open source licensing have the right to control the modification and distribution of copyrighted material.” All too often people seem to confuse “open source” with “public domain” when in fact they are very different. In fact, open source relies heavily upon copyright law to exact its restrictions. The peculiar thing about open source is that, while traditional software uses copyright as an offensive tool to thwart pirates, the copyleft movement uses the same body of law as a defensive shield to ensure that its work remains freely available and modifiable.
This weblog is an informational resource only. It is not designed to offer legal advice.
Here are some of the issues I’m following this week:
The Internet has created interesting copyright law questions for many different kinds of media, from the photographs in Kelly v. Arriba Soft to the audio recordings in A&M Records v. Napster. Typically these cases turn on whether the defendant–typically the trailblazing operator of some new Internet-based service–has distributed the plaintiff’s copyrighted material without authorization.
As Kelly illustrated, web pages are absolutely fair game for copyright infringement actions when they contain infringing materials. Take a closer look at this very post as it appears in your computer’s web browser. There are a few small graphics here and there such as the small “Linkedin” icon to the right, but the vast majority of this page is text. As the author of the substance of this text–the “words” of it–I own the copyright to it and, by publishing it on this blog, I’m granting you a license to download, read, and use it under certain license terms that are outlined in the Creative Commons License that I’ve indicated at the bottom of the page. But how about the “design” of the words? I don’t own the copyright to the typography or font. I certainly don’t have a right to distribute the font, and I’m certainly not authorized to convey any rights to you in that regard. Luckily, thanks to the basic design of HTML and the Web, this has traditionally never been a problem. No web page actually contains or distributes a copy of a font. Web pages instead rely on tags written into the HTML which simply tell the receiving party’s computer’s web browser (in this case your web browser) which font or fonts to use for rendering the web page’s text.
So, for example, if I want the text of this post to display in the font called “Helvetica,” I insert a tag that tells your web browser to display that text in “Helvetica” and rely on your computer already having that font and using it to display the text. If for some reason your computer does not have that font, your web browser will probably substitute the typeface with some other one, perhaps “Times New Roman.” There’s no potential for copyright infringement here because the web standards in place simply don’t provide a mechanism for the distribution of a font; we’re merely using ones that you (the viewer) already have.
As you can imagine, this results in a frustrating situation for many web designers. The inability to “push” fonts to visitors’ computers means that web designers are typically limited to a very small selection of “web-safe fonts,” or typefaces that are likely to be installed by default on wide variety of computers and operating systems. Apple and others in the software business have recently been promoting a new standard for the web called CSS Level 3 (or “CSS3″) which provides, among many other things, a way for web designers to easily embed a copy of a font into a web page. In theory, this would mean that a web designer could use any font he likes in his designs instead of being limited to only a handful of “web-safe” fonts. The latest version of Safari, Apple’s freeware web browser application for Windows and Mac OS X, is among the first to support CSS3 and Apple has begun promoting this. Specifically, the Safari website contains the following text as of this writing:
With CSS3 web fonts in Safari 3.1, web designers can go beyond web-safe fonts and use any font they want to create stunning new websites using standards-based technology. Safari automatically recognizes websites that use custom fonts and downloads them as they’re needed.
Typography designers are understandably disturbed by this sort of marketing because it could give some web designers the mistaken impression that it’s permissible to distribute copies of a font without authorization from the owner of the copyright in that font. This sort of marketing language bears a scary resemblance to the U.S. Supreme Court’s holding in MGM v. Grokster, in which the Court unanimously declared that one who “actively induces” copyright infringement is secondarily liable for that infringement. I don’t think it would be a stretch to call Apple’s “use any font they want to create stunning new websites” text an “active inducement” of that infringing activity.
Of course, there are obvious differences between the Grokster application and Safari. Grokster was designed specifically to attract former users of Napster’s peer-to-peer network following that service’s annihilation, for the same purpose as Napster. The Safari application is merely a cutting-edge web browser. Indeed, Safari itself doesn’t provide for file distribution at all; it only provides a way to download information that has been posted. There’s also a weak argument that Apple’s Safari website is aimed at end-users (communicating what they can expect from the application) and not web designers (telling them what they should do). But, for the purposes of secondary liability, none of this matters. At the end of the day, Apple’s Safari website is actively telling visitors that web designers can distribute any fonts they want for users to download through CSS3, and this sort of marketing language is dangerous. There’s nothing wrong with offering CSS3 compatibility and advertising that fact, but in a post-Grokster world, Apple and other software companies with cutting-edge technology need to be more careful about publishing information that seems to promote use without regard to infringement.
Disclaimer: This weblog is an informational resource only. It is not designed to offer legal advice.
Update: I remembered this morning that PDFs can also include copies of the fonts used therein, depending on the settings of the software used to make them. Some PDF creation software, like the one built into Mac OS X, seems to embed font files for every font used in a document. Unless the PDF format provides some way to prevent end-users from accessing and using embedded fonts, this poses the same question as CSS3.
Here’s a summary of the stories I’ve been following this week:
