Tuesday, May 8, 2012

'Fair use' cases cited by Google don't serve to justify Android's infringement of Java APIs

Within a matter of hours, Google will file its motion to declare the copyright trial a "mistrial" because of the partial verdict rendered yesterday. Even before the motion is filed, the intention is transparent: Google wants to avoid that the only missing piece -- "fair use" -- is decided in Oracle's favor as a matter of law. Ideally, Google wants it decided in its own favor, but if not, Google wants the chance to confuse another jury about it.

I'll comment on the "mistrial" motion as soon as I see it. The very fact that Google announced its intent to bring such a motion should give all those pause who think yesterday's partial verdict was particularly favorable to Google. If the answers that the jury actually gave served Google's purposes, why wouldn't it just accept this as a basis for the next steps?

The fact of the matter is that the judge could provide the missing link, "fair use", and convincing a judge of "fair use" is different from getting a jury to the point where a majority probably also wanted to vote against Google's defense but a minority somehow felt bad about ruling against Google. A jury gets some basic instructions from the court, and it can get specific legal questions answered by sending jury notes to the courtroom, but it doesn't research all of the case law on "fair use" to acquire a thorough understanding of where the law stands on this issue. But a judge does read the relevant rulings, and the rulings they cite, and possibly some of the rulings those rulings cite, and listens to all of the parties' argument as to why a certain case is apposite or inapposite.

I believe it's generally harder to convince a judge of "fair use" than to persuade or mislead a jury because a legal professional will typically be much more analytical, and especially much less emotional, about the relevant factors than ordinary citizens. Also, legal professionals are, on average, more likely to be in favor of strong intellectual property rights than the average person out there. But for Oracle and Google it doesn't matter what the average outcome would be -- they care about only this case. And once we're talking about a particular case, a lot (sometimes everything) depends on who can point to applicable cases that really support his argument. Parties often misrepresent (usually by quoting out of context) what certain decisions actually said -- and by simply looking up the actual decisions cited, or at least the context of a given quote, it's often possible to identify weaknesses in this kind of argument.

Since the close of evidence (i.e., more than a week ago), the parties have exchanged a fair amount of legal argument about issues that either one would like the judge to decide in its favor regardless of whether or not those questions, or issues relevant to them, were put before the jury. Google's relevant motion for judgment as a matter of law (JMOL), the most recent version of which was filed last Wednesday, didn't try to convince the judge that fair use can be decided in its favor. There are page limits, and Google focused instead on issues such as copyright ownership and copyrightability. A defendant can afford the luxury of focusing on a very few issues. Even one issue can be enough to defeat a whole claim, while a plaintiff must overcome all defenses. Nevertheless, it's interesting that Google thought it was a better use of the available space to focus on different things than asking for a JMOL in favor of "fair use". It's yet another indication of Google being uncomfortable with this being decided by the judge as a matter of law. As I'll mention further below, Google tried this last summer and failed.

Google finally made a JMOL-related "fair use" argument in one of yesterday's filings: four of the 15 pages of Google's response to Oracle's motion for JMOL are devoted to Google's argument that "Oracle failed to establish that no reasonable jury could find fair use". This means Google contents itself with opposing Oracle's request for JMOL. By contrast, there are other items on which Google clearly claims that "Google, not Oracle, is entitled" to a ruling in its favor -- but "fair use" isn't one of them.

It's also worth noting that Google takes a defensive position on the third factor (quantity and quality of the work taken), arguing that this factor "either favors fair use, or is at least neutral". For the other three factors, Google claims that those weigh in its favor, so the third factor is one on which Google is most defensive. We'll never know whether the jury could have reached a consensus had it known about Google's soft stance on this factor.

In the following, I'll take a look at the cases Google primarily cites in its argument against Oracle's JMOL motion. It's an exercise that I'm sure will help all those of us who don't deal with "fair use" issues on a daily basis to get a better idea of what the legal framework for "fair use" really is. Google has excellent lawyers, so we can assume that they've researched all of the potentially applicable cases and found the ones that, relatively speaking, best support Google's case. As a side note on the quality of Google's litigators, I told a legal reporter on Twitter today that the ones representing Google there probably weren't Google's first choice because I assume Google would have preferred Quinn Emanuel, an absolutely amazing firm that handles most of the Android defense effort in high-profile cases and presumably wasn't available because it also lists Oracle as a client on its website, but even Google's second choice is undoubtedly so very good that it identified the cases most favorable to Google's case. But are those best cases good enough that Google can prevail on "fair use" if the judge rules on it himself (or if not this judge, then maybe an appeals court)? Let's look at those citations one by one.

What's important to keep in mind here is that "fair use" is subject to a multifactorial analysis, so it generally isn't decided based on just one of the four factors. That's why appellate decisions don't have very clear-cut rules of the kind that "if criterion A is met, decision B follows". The appeals courts primarily dealt with whether certain findings cut against or support "fair use". In other words, those are generally decisions about the eligibility of certain kinds of fair-use-related arguments. But many such decisions can collectively form a legal framework.

Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569, 584 (1994) -- the "Pretty Woman" case

Both parties cite this case a whole lot. Oracle's motion calls it "[t]he leading case on 'transformative' use", and the first fair use factor, which is about commercial and transformative use, is indeed the context in which both parties cite it for the most part. They also cite it in connection with the fourth factor (effect on market potential or commercial value of the work taken).

The copyright holder here was Acuff-Rose Music, Inc., and claimed that the members of the rap music group 2 Live Crew and their record company infringed, with the 2 Live Crew's song "Pretty Woman", the plaintiff's copyright in Roy Orbison's rock ballad, "Oh, Pretty Woman".

The district court considered this kind of parody fair use. The appeals court (Sixth Circuit) viewed this as unfair, mostly in light of the amount of original material taken, the commercial nature of the parody, and a presumption of harm because of the commercial nature of the use. The Supreme Court held that "2 Live Crew's commercial parody may be a fair use", mostly because it thought that a fairly transformative work can be fair use even if it is, at the same, very commercial, and also didn't feel that the theory of harm (lost sales for the copyright holder) was convincing. The case was remanded, but the parties settled prior to any subsequent ruling. According to Wikipedia, the defendants took a license and they paid for it, but it's not known how much. It might have been a face-saving exit for the right holder, but most of the time, such settlements do come with significant payment obligations, even if usually at a considerable "discount" from the right holder's demand. The reason why I tend to believe the latter is that most of the commercial opportunity for a rap song is really when it's new, so by the time the parties settled, the 2 Live Crew didn't really have to fear an injunction too much, but it had to think about the risk of being liable for damages for past infringement. I doubt that the courts would have decided that the right holder wasn't entitled to anything.

This "Pretty Woman" case has already had an impact on Oracle v. Google because Judge Alsup adopted a part of its reasoning for the definition of "transformative" use: the Supreme Court said that transformative use means the new creation "adds something new, with a further purpose or different character, altering the first with new expression, meaning, or message". I said before that I don't like a wording like "something new" -- in a Supreme Court ruling, which will be applied by professional judges, that's not a problem because it will be understood reasonably and applied responsibly, but if it's given to a jury, I think it's too vague to be helpful. It can lead a jury to think that "anything goes" as long as it's "new" in whatever way imaginable.

Having looked at how the parties cite the case, I think it provides purely defensive ammunition to Google. Google can point to this in order to avoid that it loses the first fair use factor only because Android is a totally commercial project. But if properly applied to Oracle v. Google, Android's embrace-extend-extinguish approach to Java cannot be justified with the "Pretty Woman" logic. To "embrace" and "extend" a software platform involves adding "something new", but not in a way that is in the public interest with a view to the concept of "transformative" use. And the "extinguish" part cuts against the fourth factor. I said further above that Google tried last summer to persuade Judge Alsup of "fair use" and failed. Here's a link to a summary of what the judge wrote about fair use in his denial of a sweeping summary judgment motion Google had brought about Oracle's copyright claims. Look at this quote from the ruling:

"On the present record, a reasonable fact finder could disagree with Google's rosy depiction of Android's impact on the Java market."

Theoretically, the "present" record as of last summer and the one as of today can differ, and at the time the judge didn't take firm position but merely denied that this matter was ripe for summary judgment. Still, chances are that the judge is still aware of the "extinguish" part of "embrace-extend-extinguish".

Getting back to the "Pretty Woman" case, a "parody" (the word appears countless times in the Supreme Court ruling) really is very much in line with what statutory lists as examples of "fair use" (which the jury instructions recited): "criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research". But the Campbell ruling doesn't really give Google ammunition to have Android's hijacking of Java likened to any of those examples, while the word "parody" could be added to that list without being the "odd man out".

With a view to the fourth factor, it's easy to understand that a court may believe a rap parody of an older song of a different style doesn't harm the commercialization of the original song on the bottom line. I'm not a music industry expert, but I'm an observant consumer and I ran marketing campaigns for consumer software, a somewhat similar market. From that vantage point, I believe such parodies and remakes of a fairly different style actually drive sales of the original song -- or at least they don't hurt too much on the bottom line. But if a remake is very similar to the original, it's sometimes hard to even make a distinction when a song is played on the radio. With Android and Java, Google also claims that there are similar dynamics, but looking at how Android's market share has grown in recent years at the expense of actual or potential Java platforms, the substitutive effect should be pretty clear. It's easy to imagine that someone who bought the 2 Live Crew's version of "Pretty Woman" additionally purchased the original, or vice versa -- but how many people use two smartphones in parallel?

Harper & Row, Publishers, Inc. v. Nation Enters., 471 U.S. 539, 561 (1985) -- the "President Ford biography" case

In the "Pretty Woman" case, I saw strong indications that the right holder got a significant payment. In this "President Ford biography" case here, the Supreme Court clearly ruled that a certain magazine article, which quoted 300-400 words from that biography, did not constitute fair use. Three of the justices dissented because they thought this kind of quoting was in the public interest. It's not surprising that there was significant dissent. This case was really in a gray area, and the outcome could have been a different one. But based on what the outcome was, it was certainly the opposite of what supports Google's case.

Google quotes only one very broad and general sentence from that decision, while Oracle cites it in connection with three of the four fair use factors. Here's the sentence Google points to:

"Congress resisted attempts to narrow the ambit of this traditional enquiry by adopting categories of presumptively fair use, and it urged courts to preserve the breadth of their traditionally ample view of the universe of relevant evidence." [emphasis added by Google in its filing]

Basically, Google tries to use this reasoning in order to give relevance to arguments that stray from the examples of and criteria for fair use the law provides. But that's about it. Oracle gets far more mileage out of the "President Ford biography" case. One of various examples is that the limited quantity of quoted words in there really lowered the bar for right holders trying to overcome a fair use defense, and Oracle doesn't even need to lower the bar in that regard because the judge agreed that all those names, which he thinks aren't protectable on their own, are protectable as part of the structure, sequence and organization of the Java APIs. Also, the fact that the jury apparently agreed rather quickly on infringement shows that those copied elements are quite recognizable, even to non-technical people.

Sony Computer Enter. v. Connectix Corp., 203 F.3d 596, 606 (9th Cir. 2000) -- the "video game emulator" case

This was a video-game-emulator-friendly Ninth Circuit ruling. Sony sued the maker of a software capable of making PlayStation games run on PCs. Sony won a preliminary injunction. Here's how the Ninth Circuit summarized the core of its appellate ruling:

"We reverse and remand with instructions to dissolve the injunction.  The intermediate copies made and used by Connectix during the course of its reverse engineering of the Sony BIOS were protected fair use, necessary to permit Connectix to make its non-infringing Virtual Game Station function with PlayStation games.  Any other intermediate copies made by Connectix do not support injunctive relief, even if those copies were infringing."

That summary is enough to show that this ruling, even though unusually defendant-friendly, really dealt with something that's fundamentally different from the Android/Java situation. What happened here is that an injunction was granted against software that was "nonfringing". If there was an infringement, or at least an infringement that had to be defended based on "fair use", it related to "intermedia copies", a term that appears twice in the passage quoted above. In Oracle v. Google, the issue is not whether Google made any "intermediate copies" for the purpose of "reverse engineering" (other than that any of this is relevance to the "clean-room implementation" theory). The issue is that 750,000+ Android devices are activated every day. That's massive redistribution of material the jury clearly found to be infringing. In the "video game emulator" case, there wasn't massive redistribution -- just a few internal copies.

While the context of this case is lightyears away from the Android/Java issue, there are certain statements in that ruling that Google can use if it quotes them out of context. For example, Google points to the Ninth Circuit's finding in favor of "transformative" use -- but this quote shows that it was only deemed "modestly transformative" (since the games running on the emulator were the same that one could also play on the PlayStation itself):

"We find that Connectix's Virtual Game Station is modestly transformative."

Let's put this into context again: if something is only "modestly transformative" but any potential infringement is limited to a few internal copies made for interoperability and reverse engineering purposes, then the infringement might be proportionate to the modest transformative value.

Google argues, somewhat defensively, that "a reasonable jury could find that Android is, at a minimum, 'modestly transformative.'" But if "modestly transformative" was enough in a case in which the actually distributed code wasn't at issue, that certainly doesn't mean it's enough to justify large-scale redistribution of infringing material. That's where Google's references to this case face a fundamental problem.

Google furthermore points to the fact that "the use was for the purpose of achieving compatibility, a legitimate use under the first factor". But it's an established fact that Android is not compatible. Even if it was made fully compatible, the issue would still be that "the use" Google refers to in this case related only to a few internal copies, not to massive redistribution of infringing material.

The Sony decision also points out that copyright law is not supposed to confer a monopoly over a certain market. However, the ruling then points to another decision I'll address, Sega v. Accolade, which attaches great importance to "the statutory purpose of promoting creative expression" -- a concept that's difficult to apply to Android and Java. At any rate, the decision merely said that this factor favored the defendant, but the overall "fair use" analysis depends on multiple factors, and in this context, let's never forget that infringement was limited to a few internal copies. It took a few internal copies to open up a whole market. That's a different value proposition for society than shipping 750,000+ infringing devices every day.

Google's proposal to apply the logic for a few internal copies to its huge Android business is not only a giant leap but it would ultimately reduce copyright law to absurdity. Even though copyright law should not confer a monopoly in a given market, all intellectual property rights are time-limited monopolies by definition. That's why right holders can generally prohibit unlicensed use.

Sega Enters. Ltd. v. Accolade, Inc., 977 F.2d 1510, 1524 (9th Cir. 1992) -- the "20-25 bytes" case

This is another video game-related case decided by the Ninth Circuit. It raised a couple of copyright issues and a trademark issue. The trademark part is undisputedly irrelevant to Oracle v. Google. On the copyright side, there's the "intermediate copying" issue (for reverse engineering/interoperability) I just addressed in connection with the Sony case (no need to discuss that one in too much detail again) as well as the following issue of 20 to 25 bytes being used in a configuration file to make games run on Sega game consoles:

"During the reverse engineering process, Accolade engineers had discovered a small segment of code – the TMSS initialization code – that was included in the “power-up” sequence of every Sega game, but that had no identifiable function. The games would operate on the original Genesis console even if the code segment was removed. Mike Lorenzen, <977 F.2d 1516&> the Accolade engineer with primary responsibility for reverse engineering the interface procedures for the Genesis console, sent a memo regarding the code segment to Alan Miller, his supervisor and the current president of Accolade, in which he noted that 'it is possible that some future Sega peripheral device might require it for proper initialization.'"

In the second round of reverse engineering, Accolade engineers focused on the code segment identified by Lorenzen. After further study, Accolade added the code to its development manual in the form of a standard header file to be used in all games. The file contains approximately twenty to twenty-five bytes of data. Each of Accolade’s games contains a total of 500,000 to 1,500,000 bytes. According to Accolade employees, the header file is the only portion of Sega’s Code that Accolade copied into its own game programs."

On a completely hypothetical, fictitious basis, this issue would mean for Java that apps are only executed by the virtual machine if the initialization part of the code contains 20 to 25 bytes, sort of an identifier, and in order to make apps run, someone reverse-engineered this and created an alternative Java compiler that also adds that identifier sequence of 20 to 25 bytes to its initialization data.

So besides the reverse engineering part that is not about massive redistribution, there's the issue of large-scale distribution of software containing those 20 to 25 bytes. Even if the distribution volume was massive (it apparently was because Accolade sold significant quantities of games), the scope of this is minuscule on a per-unit basis. It's hard to imagine that those 20 to 25 bytes constituted any creative achievement. They just served the purpose of an identifier, like a password. By comparison, numerous Java method names are longer than that.

Google relies on this decision strongly in connection with the second "fair use" factor, arguing that "the Copyright Act does not protect 'functional or factual aspects of the work'" and claiming that "[w]orks having 'strong functional elements' are entitled to less protection than, for example, works of fiction." However, it's easy to agree that an identifier sequence of 20 to 25 bytes is purely functional or factual -- it just doesn't mean anything at all for 37 APIs consisting of 400 classes and many thousands of methods.

The argument about how "functional" the APIs are will continue to play a role in this case, particularly in connection with copyrightability. The judge has asked the parties questions about it. We'll all have opportunities to look into that question in more detail in that context.

Kelly v. Arriba Soft Corp., 336 F.3d 811, 820-21 (9th Cir. 2003) -- the "search engine thumbnail images" case

For this case (the final one discussed in this post), let's start with Google's (only) reference to it, which is all about the third "fair use" factor (quantity and quality of work taken):

"Moreover, even if the jury found that the portion of the work that Google used is significant, '[i]f the secondary user only copies as much as is necessary for his or her intended use, then this factor will not weigh against him or her.'"

At first sight this looks like a great excuse for a lot of things. But it takes only one example to show why this isn't the kind of sweeping and permissive rule it appears to be. If I wanted to watch a movie without paying, that would be an "intended use". Assuming the movie has 95 minutes, I may not need the credits part at the end, so I'll copy 92 minutes. If I get sued, I'll explain what my "intended use" was and point to the fact that I only copied what was necessary for this purpose. Obviously that wouldn't work: the third "fair use" factor would become useless. In fact, it would even become useless if this kind of "intended use" justified, with respect to the third factor, an unlicensed copy of even one minute.

Google says that "by using only the SSO and not the implementing code, and by limiting its use of the SSO to the API packages that developers are most likely to expect to be able to use to write programs in the Java language for a smartphone platform, Google used only as much of Oracle’s work as necessary." But Google doesn't believe in this argument too much, which is why it says about this "fair use" factor that it may simply be "neutral".

What I just quoted from Google is arbitrary: who determines that those 37 APIs are the ones "developers are most likely to expect"? Did they run a poll? Did they perform a reliable statistical analysis? But even if they had done so, it comes down to the "embrace-extend-extinguish" strategy: attract all those developers by providing some initial familiarity. It's all about taking, not about giving back to the Java community. It's a one-way street.

The Kelly case was also very different from the Android/Java situation. It was about thumbnail images displayed by a search engine. If I see a thumbnail image on Google, it may invite me to click through to the actual webpage, or maybe not, but at any rate, it won't be a substitute for the high-resolution image. Android isn't some kind of "low-resolution Java" serving a distinct purpose.

If you'd like to be updated on the smartphone patent disputes and other intellectual property matters I cover, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents and Google+.

Share with other professionals via LinkedIn: