On Monday I attended Judge Alsup’s courtroom to hear the closing arguments for Oracle’s copyright infringement case against Google. Before the jury reaches their verdict for phase one of the trial, I am providing my decision based on what I know about the case and what I witnessed in court. While I do not have ready access to the court record and trial exhibits as the jury does, I have access to news and commentary related to the trial and hopefully some knowledge of copyright law. To help guide me I will be using the Final Charge to the Jury (Phase One) and Special Verdict Form the judge gave the jury yesterday. I am reprinting the content of the verdict form before giving my answers and explaining them.
1. As to the compilable code for the 37 Java [Application Programming Interface (API)] packages in question taken as a group:
A. Has Oracle proven that Google has infringed the overall structure, sequence and organization of copyrighted works?
B. (Skip if you answered No to 1A) Has Google proven that its use of the overall structure, sequence and organization constituted “fair use?”
2. As to the documentation for the 37 Java API packages in question taken as a group:
A. Has Oracle proven that Google has infringed?
B. (Skip if you answered No to 2A) Has Google proven that its use of Oracle’s Java documentation constituted “fair use?”
3. Has Oracle proven that Google’s conceded use of the following was infringing, the only issue being whether such use was de minimis:
A. The rangeCheck method in TimSort.java and ComparableTimSort.Java
B. Source code in seven “Impl.java” files and the one “ACL” file
C. The English-language comments in CodeSourceTest.java and CollectionCertStoreParametersTest.java
4. Answer the following special interrogatories only if you answered yes to Question 1A:
A. Has Google proven that Sun and/or Oracle engaged in conduct Sun and/or Oracle knew or should have known would reasonably lead Google to believe that it would not need a license to use the structure, sequence, and organization of the copyrighted compilable code?
B. If so, has Google proven that it in fact reasonably relied on such conduct by Sun and/or Oracle in deciding to use the structure, sequence, and organization of the copyrighted compilable code without obtaining a license?
1A. Yes, Oracle proved that Google infringed the overall structure, sequence and organization (SSO) of copyrighted works. According to Jury Instruction 19, Google agrees that it uses the name names and declarations in 37 API packages Android has with respect to 37 Java API packages. Google also agrees that the SSO of the 37 API packages in Android is “substantially the same” as the SSO of the corresponding API packages Java has. Oracle concedes that the line-by-line implementations are different, but the issue is the copying of the SSO, not the actual code. As Jury Instruction 28 mentions, an infringer does not escape liability by adding new material. Google cannot argue the other material in the Android platform overshadows taking the SSO. Professor Astrachen, Google’s expert witness, pointed out there was no question that Google copied the SSO of the API packages, just that Google had an excuse to copy. This brings us to the next part, the fair use determination.
1B. No, Google has not proven that its use of the overall structure, sequence and organization constituted fair use. Jury Instruction 26 briefly explains to the jury how to answer this question. Four factors determine fair use: the purpose and character of the use, including whether such use is of a commercial nature and whether such work is transformative; the nature of the copyrighted work, including whether the work is creative, functional, or factual; the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and the effect of the use upon the potential market for or value of the copyrighted work.
The purpose and character of Google’s use of the SSO is of a commercial nature. Google contends that Android is not a commercial use because it is an open source project and freely gives the platform to other companies for them to modify and use. However, Google still developed Android to promote Google products. Google does not earn money distributing Android, but Google earns money through Android’s promotion of Google’s product such as advertising. Google’s entire goal was to break the hold Microsoft and Nokia had on the mobile platform operating system market. Google’s goal is in contrast to the Apache Harmony project, which was not for profit to begin with.
Google also claims that Android is transformative because it brought Sun’s Java platform into the mobile arena when no other company could. While the overall platform may be transformative, the issue is over the SSO of the packages. Google has not shown that they transformed the SSO into working in the mobile environment, merely that they transferred the SSO for their own commercial needs, getting a product to market as quickly as possible.
Furthermore, in order for a work to be transformative, the work needs to give the copyrighted work new expression, meaning, or message. In Dr. Seuss Enterprises, L.P. v. Penguin Books USA Inc. (109 F.3d 1394), the defendant copied the style of Dr. Seuss to retell the OJ Simpson trial. The court ruled that the work was not transformative because the work had no critical bearing on the substance or style of The Cat in the Hat. Since Google merely appropriated the SSO instead of commenting on it, Google’s use of the SSO is not transformative according to precedent.
The nature of the copyrighted work is artistic, not merely functional. Oracle argues that a great deal of time and effort was spent to construct the various API packages and the interrelationships between them. Google downplays their significance when they argue that the API is no more than a filing system. Google is trying to simplify the API to justify that its copying was a mere overlapping of ideas about arranging the packages in the platform. Oracle’s analogy of WestLaw’s KeyCite system better describes the complexity that went into construction of the API. The KeyCite for all of its arrangement deserves to be more than a functional work, especially considering the multiple times terms will link to other terms without much explanation.
The amount of copying is more than substantial. Google tries to argue that overall Android did not take much from Google. However, the issue relates to the SSO, not the Java API in general. Google already admits that the 37 Android API packages copy the SSO in corresponding Java API packages, so Google cannot draw away attention to the SSO for the purposes of a fair use determination. Google also claims that, as Jury Instruction 29 mentions, the “work as a whole” refers to all 166 Java API packages and the SSO in 37 of the packages is enough for a fair use determination. However, Google has not offered enough evidence that the SSO of the 37 packages are somehow not important compared to the entirety of the API SSO, and Google had the burden of proof.
Finally, the effect on the market value of the copyrighted work is little to none. Oracle argues that Google fragmented the Java platform and prevented Oracle and/or Sun from bringing Java to the mobile space. Google retorts that the Java platform was internally fragmented and that Oracle is seeing more revenue from Java licensing. While the potential market value lost should be considered, the copyrighted work is the SSO of the Java API, not the Java platform. Oracle did not provide evidence that the SSO lost value when Google copied it into Android. In fact, Google showed that it increased the market value of the SSO when Sun decided to run Java programs on the Android platform. Google may not have certified Java compatibility on Android phones, but Android still increased Java business in a way and did not decrease the potential market value of the SSO of the Java API.
Fair use determination requires the balance of four statutory factors. The effect on the market value of the copyrighted work is probably not negative, maybe even positive as Google claims. For the 37 packages in question, Google copied all of the SSO into Android packages, so the amount of copying is substantial. The SSO itself should be considered more than functional because of the complexities involved in its creation that Google unconvincingly tries to downplay. Finally, the Android platform and the SSO of its API are commercial products created to earn money, not a nonprofit or educational endeavor, and Google appropriated the SSO rather than transform it. Therefore, I conclude that Google’s copying is not a fair use of Oracle’s copyrighted work.
2. No, Oracle has not proved that Google infringed on the documentation for the Java API packages in question taken as a group. According to Jury Instruction 24, Oracle can prove copyright infringement of the documentation if the accused had access to the copyrighted work and that there is virtual identity between the copyrighted work and the accused work. Admittedly Google had access to the Java documentation because Sun made it available as part of the open source project. However, the virtual identity test is needed because the documentation describes narrow technical functions and expectedly uses certain terms and phrases to describe the material. Oracle’s expert witness had difficulty showing even substantial similarity between the documentations when Google cross-examined him. Since the burden of proof is on Oracle, it is more likely than not that Google merely used the same technical terms instead of copying the Java API documentation as Oracle alleges.
3. No, Oracle has not proven that Google’s conceded use of specific code was more than de minimis. According to Jury Instruction 28, copying is de minimis only if it is so meager and fragmentary that compared to the work as a whole the average audience would not recognize the appropriation. Oracle tries to stress the direct copying, but as Google argues, nine or so lines out of thousands of lines of code is the definition of de minimis. Furthermore, the person that put TimSort into Android before giving it to Java took the method out of Android and said it wasn’t a big deal. Oracle failed to prove that any direct copying was recognizable by a common viewer or more than a piece of the much larger platform.
4. No, Google has not proven that Sun and Oracle engaged in conduct the two companies knew or should have known would reasonably lead Google to believe that it would not need a license to use the SSO of the copyrighted compilable code. Google’s entire argument hinges on proving that the license was only necessary for the Java brand. Andy Rubin was fine releasing his work and not calling it Java if Sun did not support Android. Furthermore, Jonathan Schwartz while Sun’s CEO considered his blog to convey official Sun statements. While Scott McNealy, Schwartz’s predecessor, testified that the blog was not official Sun policy, Schwartz tweeted a link to an archive on what amounts to an official Sun corporate statement on his blog; unfortunately the jury cannot see that before reaching a verdict. Regardless of the nature of Schwartz’s blog, the post congratulating Google for releasing Android has no bearing on the legal implications of releasing a Java-based platform without Sun’s license. Schwartz admitted that he declined legal action against Google because it didn’t make business sense, not that Sun actually lacked a case on the merits. Eric Schmidt, Google CEO when Android was released, said he talked with Schwartz to make sure Google didn’t need Sun’s license to release Android, but Schmidt did not remember the specifics of his conversations with Schwartz, nor did he see an article mentioning that Sun was concerned Android would fragment the Java platform around the time Android was released.
The testimony and evidence overwhelmingly confirm that Sun and Oracle intended to protect their intellectual property through licensing the Java platform. Oracle counsel pointed to copyright notices in the Java source code and the Java language in general. Oracle also showed that Rubin, while CEO of Danger, took a license from Sun mentioning the specification even though his company did not touch any Java source code. Google kept arguing that the Java language and by extension the Java API were free for the public to use. However, as Oracle pointed out, after Apache’s conflict with Sun about the Apache Harmony project, Apache concluded that Java was not an open platform. Google knew about the Apache debacle and yet declined to take a license for its commercial product to satisfy at least some legal right Sun claimed even though Sun and Google could not agree to a partnership concerning Android development.
Throughout this trial, Google maintained that it has done no wrong. The trial has been very revealing concerning the programming community and its knowledge of legal issues. The copying is going to happen because programmers know the general structure of programming languages and will utilize them in order to convince others to adopt new languages and programs derived from them. Even though Google made a reasonable business decision and had enough backing from their competition to avoid trouble after negotiations for an Android partnership failed, Google did not demonstrate that it could avoid legal liability for copying the SSO from a language so popular and ubiquitous as Java. Google’s best hope is to contest the copyrightability of the SSO on appeal or to support legislation that would prevent a company from obtaining a copyright covering the SSO for the purpose of spreading the development of new and useful programming languages.