Category: Copyright

Google v. Oracle III – Popularity Does Not Overturn Copyright

Looking at Google v. Oracle as a consumer and citizen, common sense insists upon a measure of skepticism in response to the premise that the “future of all software development” depends on Google prevailing in this case.  Many of those who say so are the same folks who tend to omit the fact that licensing—especially in B2B relationships—spawns innovation all the time.  The underlying bias that copyright makes works inaccessible, rather than licensable, is a connotation that should not be overlooked in this case, particularly when Oracle’s Java licensing agreement is specifically designed to demand interoperability and promote development.   

In Part II, I said would begin to address the core matter of contention in Oracle—the copyrightability of the “declaring code” that Google copied for development of the Android mobile platform.  Google’s main legal arguments boil down to the following:  1) the “declaring code” they copied is a “method” and, therefore, an unprotectable idea under Section 102(b) of the copyright law; but 2) the Court could more narrowly find that copyright is denied under the merger doctrine; and 3) failing all that, Google’s use was a fair use.  

In this post, I want to focus on the second of those assertions, partly because the Section 102(b) argument against copyrightability is largely being made by supporting amici, and although Oracle filed its SCOTUS brief on February 12, supporting parties have not yet filed.  But the other reason I want to focus on the merger claim in this case is that the rationale being applied comes uncomfortably close to asserting that once a work becomes popular or important enough in the market, its copyright protection is somehow diminished.  This is simply wrong as a matter of law and is an issue of no small concern for creators working in any protectable medium.

“Had To” vs.  Had To

To begin with an analogy, if I ran a design firm that relied on a variety of staff and freelance artists who most commonly know how to use Photoshop, my dependence—even existential survival—on people with those skills using that software does not relieve me of the obligation to obtain a license from Adobe. (In fact, firms like this are Adobe’s bread-and-butter.)  Yet, Google appears to be arguing that its “need” use the Java code was justifiable without a license because the popularity of Java denies copyrightability in the copied lines of code under the merger doctrine.  

To review what I prefaced in Part II, merger denies copyright in works when there are so few ways to express an idea that the expression and idea are said to be “merged.”  With software, merger can seem rather tricky because, unlike other forms of creative expression, software always performs a function, which is not protected by copyright law.  For instance, no software author can protect the concept of having a computer perform the function of regulating the fuel efficiency in your car, but he can protect the code he writes to perform that function—unless there is only one way to write that code, in which case merger denies copyright.  

Google’s merger argument with respect to the Java “declaring code” is founded on the premise that they “had to use it.”  But to read their brief (and others), it is not at all clear that they “had to” in the strict, legal sense of “only one way to do it” so much as they “had to” in the business sense that it was the best way to enable Java-fluent developers to write for Android.  For instance, the Google brief states, “… the declarations can only be written one way to perform their function of responding to the calls already known to Java developers.” (Emphasis added)

That rationale, akin to my Adobe Photoshop analogy, may have been exigent to Google’s ambition to enter the mobile market and quickly attain the dominance in that space that it now enjoys, but it does not read as a correct appeal to merger.  In particular, we have two undisputed facts in this case:  at the time of authorship, Sun engineers could have taken many different creative paths to produce the “declaring code” at issue; and at the time of infringement, Google engineers also had many creative options to achieve the same functions.

Google even admitted as much, stating in its brief that its engineers could have written their own version of the desired “packages” but for the fact that this would have been a barrier to enabling Java-savvy developers to write for Android. But just as with my design firm example, this is grounds for licensing an urgently-needed work, not a rejection of copyright under merger.

Google Up to its Usual Tricks?

Assuming the above is a fair criticism, it is a clever sleight of hand for Google to color the business reason they “had to” copy Oracle’s code by advancing a legal theory that could more broadly weaken copyright for all types of creators.  Hardly novel territory for Google (or indeed many of its supporting amici), it is quite common to encounter anti-copyright rhetoric alleging that the more a work is considered essential (i.e. very popular), it steadily migrates into the commons; and this legal fallacy was a concern raised by the Solicitor General in its brief recommending the Supreme Court deny cert in this case … 

“Petitioner [Google] also argues that the court of appeals erred by focusing on the choices available to respondent ex ante when it created Java, rather than on the choices available to petitioner when it sought to devise a way for programmers familiar with “industry standard Java shorthand commands” to use those same commands in Android. Petitioner’s approach would treat the current popularity of respondent’s work among developers as retroactively divesting the work of copyright protection.  (Emphasis added. Citations omitted.)

Note that the word retroactively underscores the point that Google’s allegedly urgent “need” to copy the Java code in 2005—now that it has been synthesized through various legal defenses over the ensuing 15 years—attempts to reach back and deny copyright circa 1995 by applying a doctrine that would not have denied copyright at the time of authorship.  As stated in Oracle’s brief, filed on February 12, “Google invokes merger—a narrow judge-made doctrine that does not apply unless the original author had very few ways to express the idea.  It does not apply here because, as Google concedes, Java SE’s authors had countless options.”

Consequently, Google’s merger argument begins to read like an attempt to draw attention away from the creative options available to both original author and alleged infringer—a distraction that is perhaps camouflaged by the fact that computer code is generally invisible to most observers and is functional in nature, which makes it appear more susceptible to the merger exception.  

I will admit that while combing through the briefs filed so far in this case, I find myself wondering whether Google engaged in unscrupulous business conduct and is managing to thread the needle of a plausible legal defense—or whether Google engaged in bad-faith conduct and then proceeded to cobble together a Jenga tower of flawed defenses that, if upheld, are more holistically harmful to copyright law than might at first appear. 

Certainly, the latter conclusion would be consistent with the “infringe now, litigate after we own the market” strategy that has come to define Silicon Valley culture;  but this conclusion is also implied by the legal arguments presented when we begin to unpack them one at a time, just as the courts do.  For the sake of copyright principles writ large, creators should hope the Supreme Court takes a very cautious approach when considering the logical foundation of Google’s appeal to merger in this case.  The relative value of a work does not diminish its protection.  On the contrary, we protect works under copyright with the hope that they will become very valuable indeed.    

DMCA Review Begins. Watch the Red Flag.

Early last week, the Senate Judiciary Committee held the first in what will be a year-long series of hearings (roughly one per month) to review the Digital Millennium Copyright Act.  Almost as old as the publicly-available internet itself, the 1998 DMCA expressed the best efforts of Congress to predict how the digital market might evolve and to, therefore, strike a balance between the interests of internet service providers (ISPs) and copyright owners.

Over the intervening twenty-two years, much—MUCH—has been written, debated, shouted, flung, wailed, opined, and scorned about the DMCA, specifically Titles I and II of the five-title statute.  If we ask the tech-centric/copyright-skeptics, they are likely to say that Title I (§1201) is a disaster and that Title II (§512) is working just fine; while the creator/copyright proponent will tell us exactly the opposite. I cannot condense the number of issues raised in this first hearing alone into a single post—especially when §1201 and §512 address very different legal regimes—and it is far too early in the review process to respond to any specific proposals being made. 

What I will reiterate in this post is that the greatest concern to creators of every size is the conditional liability shield (“safe harbor”) provided to web platforms by §512.  It is the foundation of the oft-described “whack-a-mole” problem whereby the independent author attempts to remove infringing uses of her works one-by-one, only to have them reappear on the same platform(s) faster than she can prepare new notices.  (And “whack-a-mole” can be just as big a problem for a small business like an apparel maker as it is for a traditional artist like a musician.)  

In response to this futile battle with online infringement, authors often give up enforcement via the DMCA takedown process (resigned to donating even more revenue to billion-dollar corporations) while they ask as a community why the major platforms in particular cannot do a better job of preventing protected works from being chronically re-uploaded without license.  This second question is where we step into a BIG policy kerfuffle with regard to §512, and I imagine it is a topic about which we are going to hear a lot of ideas and a lot of noise.  

This week’s hearing hosted two panels of witnesses, the first of which provided an overview as to how the DMCA came to be; while the second panel, comprising IP academics, provided some insight as to where the DMCA debate may be heading.  In the interest of keeping this post containable, I will focus on the testimonies of Professor Sandra Aistars of the George Mason School of Law and Professor Rebecca Tushnet of Harvard Law School, and the subject of “red flag” knowledge under the DMCA.  

What is “Red Flag” Knowledge?

Unfortunately, you will get different answers depending on whom you ask, including a court split on the matter if you ask either the Second or the Ninth Circuit Court of Appeals.  But in everyday life, “red flag” knowledge is a reasonable, common-sense inference that one can draw from a modest amount of empirical evidence and experience.  If you enter the house to find trash strewn across the floor and a chagrined puppy in the corner, you will not need training in forensic science to have “red flag” knowledge that either the dog has committed a misdemeanor, or he has been artfully framed by the cat.  

That roughly describes the degree of analysis Congress intended ISPs to perform when encountering evidence of copyright infringement on their platforms.  As Professor Aistars noted, “Although Congress did not obligate service providers to actively seek out infringements, it did require them to act expeditiously to remove infringing materials once they have knowledge or awareness of infringing activity on their networks.” (See companion Appendix describing basic ISP Conditions.)

For example, let us imagine that the users of a web platform we’ll call Vimeo are making videos using some famous music we’ll call Beatles songs.  Any ordinary observer can reasonably assume that these users probably did not license these sound recordings; yet in the case Capitol Records v. Vimeo, the Second Circuit held, on the issue of “red flag” knowledge, that the platform’s operators would have needed either legal or music-industry expertise in order to discover infringement.

Keeping in mind that voluntary removal of material based on “red flag” knowledge of infringement is a condition of an ISP’s “safe harbor,” decisions like Vimeo do more than erase this part of the statute—they exacerbate a culture of infringement through court-sanctioned willful blindness.  And as Aistars added in her testimony, “Pointedly, this occurred in a case where discovery had revealed emails from managers to employees winkingly encouraging infringement.”  Thus, Aistars is among those who would advocate clarifying the meaning of “red flag” to restore the intent of §512.

The Vimeo emails Aistars mentions are typical of the shoulder shrugs and middle fingers creators are used to receiving from many platform operators, and application of the DMCA to date has unquestionably fostered cultural attitudes anathema to the kind of cooperation between ISPs and rightsholders Congress specifically intended to promote two decades ago.  Further, unintended endorsement of this culture among site operators may be exacerbating a persistent misunderstanding among individual and commercial users that the internet is a realm of automatic immunity.  As I have described in several posts, this misconception can cause unnecessary trouble for both creators and users of protected works.

Responses to Fixing “Red Flag” 

Anticipating the likelihood that, if there is to be any revision to §512 at all, “red flag” will be a major point of debate, Professor Tushnet warned against what she and others see as throwing out the proverbial baby with the bathwater.  “If there is one message I would ask the members of the Committee to take away today,” she stated in her opening testimony, “it is that most beneficiaries of §512 are not Google or Facebook.”  Tushnet cautions that if we were to amend §512 solely as a response to the challenges creators face on very large, commercial platforms like YouTube, we risk simultaneously putting compliant, smaller platforms out of operation and facilitating even greater monopolization by the largest entities.

As a statistical matter, Tushnet is making a “few bad apples” argument, except for the fact that some of the baddest apples in the bunch happen to be the most powerful, wealthiest internet companies in the world.  So, even if we take her premise and data at face value (i.e. that millions of compliant sites rely on §512 to exist), this does not recommend ignoring the catalog of evidence that application of the DMCA has promoted willful blindness among the operators of major ISPs.  Simply put, if twenty-million sites operate without harm while one site does harm to twenty-million creators, we still have a problem if the law shields that one site from liability.  So, the status quo cannot be the final answer.  

As a practical consideration, Tushnet’s argument is based on the assumption that a more clearly defined restoration of the intent of “red flag” knowledge can only be implemented by technological measures, which only the largest ISPs can afford.  Hence, her argument that this will result in entrenching, for instance, YouTube’s monopoly position, notably glossing over the fact that there are other forces entrenching online monopolies.  While this technology-investment argument is worthy of discussion, the aforementioned Vimeo case is just one example in which the principle of “red flag” knowledge was obliterated in a purely human paradigm (i.e. human managers choosing not to see what was right in front of them).

Post Hoc Ergo Propter Hoc? (or not all good things come from §512.)

As Tushnet testified, her own Organization for Transformative Works site hosts over “four-million works” yielding 1.2 billion page views per month, while the site receives takedown notices at a rate of less than one per month, most of which are invalid.  Assuming these data are correct, the site to which she refers seems barely relevant as an example. It is a large fanfic platform with what appears to be a vast amount of material—mainly short works of written text—that is highly unlikely to infringe.  No sound recordings.  No photographs.  No film clips.  At most, some fanfic writer could maybe—and I mean maybe—run afoul of a derivative works right. 

From a cursory review of OTW, it is not at all evident that adopting a clearer, statutory definition of “red flag” (in order to hold the majors accountable) would force a site like this one to invest in prohibitively expensive technology in order to remain complaint.  If the platform is indeed receiving takedown notices at a rate of less than one valid notice per month, this is most likely evidence that the site hosts little to no infringing material—and that when notices are received, human review is sufficient to the task.  Further, the fact that the site hosts “fandoms” for a long list of works owned by major motion picture studios indicates that infringement must be very low to near zero if it has not invited the attention of an industry with the resources to send notices in volume.  

As is often the case, defenders of the status quo (the same is true for Section 230 of the CDA) will say “look at all the benefits this law has yielded” and then point to examples that, under scrutiny, do not necessarily rely on the liability shield so substantially as may be asserted.  In this vein, Tushnet’s testimony includes several references to all manner of good news about the creative industries—more movies, TV, music, etc. than ever before—but it would be a logical stretch to assert that, for instance, Billie Eilish’s YouTube-to-Grammy-Awards success story owes much at all to §512—let alone the collapse of the “red flag” principle. 

As Chairman Tillis noted, “this is a very wonky subject,” and that last description of mine was very wonky indeed; but DMCA review will be a devil-in-the-details story to watch.  Despite the hyperbole that will inevitably seep onto social media about these hearings, it is neither practical nor desirable for rightsholders to seek obliteration of the safe harbor altogether—that is not the goal.  But at the same time, it cannot be acceptable that a statute designed to mitigate copyright infringement and incentivize cooperation has served to reward infringement and position ISPs and rightsholders at permanent loggerheads.  

*This case is further complicated by a conflict between state and federal law over the use of sound recordings made prior to 1972, but that’s a whole other bowl of noodles. 

Photo source by Robertobinetti70

Appendix I to DMCA 2020: Section 512 “Safe Harbor” Conditions

May people know that online service providers are shielded from liability for copyright infringement by their users, meaning that a court will, on summary judgment, often excuse a web platform as a named defendant when an infringement has been committed by its customers.  Many people are not aware, however, that a service provider must meet certain conditions in order to remain protected by this “safe harbor.”  

These conditions are voluntary, and although failure to meet them does not automatically make a provider liable for infringement; non-compliance will—or is meant to—void the automatic protection in a potential litigation.  Below is a list of several–but not all–of the key conditions a service provider must meet under the DMCA statute Section 512, including an explanation of “red flag” knowledge:

THIRD-PARTY INFRINGEMENT — The infringing material must have been made available by users/customers.  

This is the foundation of 512—the very reasonable assumption by early online service providers (e.g. the Baby Bells) that users will inevitably transmit infringing material online.  If the platforms were held liable for infringement by its users, this would have stifled investment in developing many platforms that host User Generated Content (UGC).  Infringing material may not be made available by a service provider.  If a site operator directly uploads or transmits infringing material of its own volition, the “safe harbor” does not shield it from liability.

NOTICE & TAKEDOWN — The service provider must expeditiously remove infringing material upon receipt of a valid takedown notice sent by the copyright owner or their agent.  

Often simply called notice-and-takedown (or just takedown), sites that wish to maintain the protection of the “safe harbor” generally comply with this provision, though it is a subject of controversy on all sides.  For creators, sending takedown notices, one infringing use at a time, is the source of the “whack-a-mole” complaint.  For ISPs and some users, the takedown regime is often described as rife with abuse and error.  See post here responding to one “abuse” study cited by Professor Tushnet in her testimony at the first DMCA hearing 2020. 

REPEAT INFRINGERS — Develop and maintain a policy that includes account termination as a final step for repeat-infringers.  

This issue made big news when ISP Cox Communications lost two substantial lawsuits for failure to maintain such a policy.  While the DMC does not clearly define “repeat infringer” or dictate the design of a “repeat infringer policy,” it is usually some type scaled warning process (e.g. six-strikes), but which must result in account termination if the repeat infringer refuses to stop.  For instance, the courts found that Cox’s 14-strikes-and-they-will-eventually-reset-your-account process voided their “safe harbor” in court. 

KNOWLEDGE — Site operators are not required to search for infringement, but they must remove material upon obtaining knowledge that it is infringing.   

Because the knowledge conditions have largely been ignored in practice over the past 22 years, many people do not know they exist or what they are.  Codified in 512(c), the statute expressly states that the service provider “shall not be liable” if (1) its operators do not have actual knowledge of infringement; (2) its operators are not aware of facts or circumstances from which infringing activity is apparent; and (3) upon obtaining knowledge of infringement, expeditiously removes the relevant material.  

That second condition describing “facts or circumstances” is what we often refer to as “red flag” knowledge under the DMCA.  This is the “walks like a duck,” common-sense knowledge standard that has generally been erased from practice.  For instance, many cases and complaints involve famous works used in ways that any layperson could assume is unlicensed; or several famous litigations have found evidence of internal communications indicating that site operators had a pretty good idea that infringement was taking place.