After a lengthy appeal, the Oracle v. Google trial on various Java APIs is headed back to the district court for a new trial. The federal appeals court in this case sided with Oracle, agreeing that the structure, sequence, and organization of the 37 Java APIs in question constituted copyrightable material.

While I still disagree with this on a fundamental level (I'd argue Oracle is merely using copyright as a false shield - it really wants to protect functionality, not form, which copyright does not protect), the 9th Circuit's Court of Appeals for the Federal Circuit's decision falls in line with the court's reputation as being one of the strongest on intellectual property protection. I won't get into the law, because frankly, I'm not a copyright expert - I understand a good deal of what the circuit court explains in its decision, but analyzing it piece by piece is a daunting task unless you've been following the minutiae of this trial from day one. The short of it is this: the circuit court was presented with a novel question of law: are APIs copyrightable in their organization, structure, and sequence as well as the actual code itself? The court decided the answer to that question is yes. And thus, the district court judge's ruling is overturned, and a new trial on two issues must now take place.

Those two issues, though, are going to be real sticklers for Oracle: fair use and damages. The jury unanimously agreed that if Oracle's APIs were copyrightable as asserted (which is now law), Google did indeed infringe that copyright. What the jury did not agree upon, and in fact deadlocked in regard to, was fair use. If Google's usage of these APIs falls under the scope of fair use (which includes arguments like interoperability), then none of this really matters from a damages perspective: fair use is an absolute affirmative defense, and negates a finding of infringement.

Google's primary arguments for fair use will likely fall under two basic ideas: interoperability and economic value. Interoperability is a touchy subject legally in the US, and while there is established law, it's still kind of a murky concept with a lot of potential pitfalls when asserted. The gist is this: assuming the party copying the copyrighted material didn't violate a license agreement or divulge a trade secret in doing so, interoperability is available as a defense if the material copied was necessary to ensure the interoperability of one system with another. This is also known as reverse engineering. Reverse engineering in the scope of copyright is often confusing to most people, because reverse engineering is by definition about functional utility. Copyright is, by definition, not about function - it is about form and attribute. Copyright, by definition, does not protection function, only form. Interoperability is a defense because the copier is saying "I copied the material not because of its structure, organization, or sequence [its copyrightable attributes], but only because copying these things was necessary to ensure the interoperability of one system with another." That means the copying was not at all about the copyrightable aspects of the material, just about what that material does (function, not form) to allow one system to work with another.

In Google's case, this would be to argue that it copied the structure, organization, and sequence of those 37 APIs solely because it was necessary in order to ensure the interoperability of Java applications with the Android OS. I am honestly not enough of a software expert to say whether or not this defense is reasonable, it's very much a fact-intensive issue. We'll see what Google says in the new trial. Oracle's job will be to argue that it wasn't necessary, and that Google could have achieved the same functional end without copying Oracle's APIs.

The other factor to consider will be the economic value of the work copied: do the sequence, organization, and structure of Oracle's 37 Java APIs actually represent a valuable work? That's another tough question, because Google will likely attempt to show that it's not the copyrightable aspect of these APIs that makes them valuable - it's their function. I really think Oracle will struggle with this, even if it is able to convince a jury these APIs have some value in their sequence, organization, and structure, putting a number to that could prove very hard.

And then, even if Google does lose out on these fair use issues, there's a question of damages. If there is value to Oracle's works, and if Google's copying wasn't solely for interoperability, what kind of damage has Oracle actually suffered here? Another question that will likely be hard to answer.

Anyway, we'll probably hear about this again in 6 months, so be on the lookout! Or don't.

FOSS Patents, Decision

David Ruddock
David's phone is whatever is currently sitting on his desk. He is an avid writer, and enjoys playing devil's advocate in editorials, and reviewing the latest phones and gadgets. He also doesn't usually write such boring sentences.

  • Sir_Brizz

    Big shocker. The 9th circuit is full of morons who care more about intellectual property rights than good judgement.

    • http://www.androidpolice.com/ David Ruddock

      The 9th Circuit is widely considered to be the most oddball federal appellate court. Very liberal on social and human rights issues, but surprisingly conservative [pro-business] on intellectual property, employment law, and contract. But then they'll go and do something that turns that on its head, too.

      I agree this seems like a bad decision, SCOTUS or congress eventually need to chime in on software copyright and lay out some concrete rules, because right now things are a bit of a mess.

    • MJ

      The article corrected it. It's actually the Court of Appeals for the Federal Circuit, the same court that (mis)handles patent cases.

      • Sir_Brizz

        Yeah, I realized that later on. CAFC is a joke, and the text of this ruling makes it clear that they did not understand Alsup's ruling to the least degree. Big disappointment and failing of the judiciary...

      • Philip Kahn

        Didn't the USSC *strongly* imply that the Federal Circuit Court of Appeals is waaay too copyright/patent friendly?



        In Octane Fitness v. ICON and a companion case, the high court held that previous decisions by the Court of Appeals for the Federal Circuit — the specialized court for patent appeals — had imposed standards that went far beyond what Congress intended when it added a “loser pays” provision to the patent code. The court also reduced the Federal Circuit’s ability to overturn fee awards by lower-court judges.

        And, of course, Google can always appeal to the Supreme Court ...

  • didibus

    This is so strange. How far do you take APIs as copyrightable? Is it only in the context of one programming language, or does it extend to the word themselves? Could you say C# copyrights infringed on Java for using some of the same keywords and syntax?

    If I make a method called string Read(Stream stream) and I had open sourced the code somewhere, and I was the first to have a public version of this form of method, am I eligible to a copyright to it?

    • http://www.androidpolice.com/ David Ruddock

      Open sourcing generally by definition gives up your copyright.

      To your first question, that's tricky, but generally, yes, a given language is the logical extent to which that copyrightability would probably apply unless there were extenuating circumstances. And no, specific keywords and syntax generally aren't going to be copyrightable. Think of it like this: the melody of a song is copyrightable, but the notes making up the melody aren't. You can't pick and choose individual pieces like that to form a copyright.

      • h4rr4r

        Open source does not give up copyrights.
        Open source licenses can only be enforced because of the very existence of copyrights. Copyright is what makes the GPL have any teeth as an example. You need to remove or edit your extremely misleading post.

        • Ricardo

          The GPL is not the only open source license. The MIT and the Apache licenses are considered open source licenses by the OSI and they do give up copyright. You are spot on about the GPL though.

          • MJ

            They don't give up copyright. The whole of open source (or Copyleft if you want) is the use of the framework of copyright to set conditions upon use.

            What the Apache and MIT licenses do is simply remove most every condition that one can reserve as a holder of copyright. If they removed copyright, they the penalties of not following the license couldn't be enforced.

          • Ricardo

            Ok granted, you don't really give up copyright if you distribute software under the Apache and the MIT licenses (other wise a third party claiming to have written the apache http server could sue the Apache foundation), but with these licenses there's not much else to enforce. What I failed to explain in my previous post is that not all open source licenses are copyleft licenses, which is a mistake you are also making. However h4rr4r was right in the fact that copyright laws allow the FSF to enforce the GPL.

          • MJ

            I honestly don't really like the work "copyleft". I simply used it because it does indicate somewhat that it's based on the idea of copyright. I'm not going to get into the whole FSF terms fight, it's not something particularly care to engage in.

            There's still stuff to enforce with these licenses. Check out section 4 (redistribution) of the APL 2.0 or the single term of the MIT license. Granted, the terms are pretty easy, but they are still terms that can be enforced.

            If the user of the product doesn't follow those terms, then the license doesn't apply to their use. Since they are using the work without a license at that point, as copyright holder you retain all rights and they get none. That means they are violating your copyright.

            There's a reason why the application of basically any license to a software work starts out with a "Copyright " line.

          • Ricardo

            'I honestly don't really like the work "copyleft" '
            So do I. The FSF should drop the word.

          • h4rr4r

            Why? It describes the situation pretty well. It is using copyright for another counter to normal purpose.

  • h4rr4r

    Copyright on an API?
    That is a bad news for the entire world of software.

    • RealMe

      Can't agree more. expect this becoming trend on coming years, more API copyright infringement.

  • selonmoi

    Can this decision be appealed? It seems really problematic for the whole industry.

  • PhilNelwyn


    • http://www.androidpolice.com/ David Ruddock

      Yeah, but Google's still got 4+ infringing years to pay for even when ART takes over. That is, if they lose.

      • Sir_Brizz

        ART doesn't end their infringement. ART still reimplements all of the 37 APIs.

      • ssj4Gogeta

        This lawsuit is about the Java standard APIs and not the VM (rt.jar, etc. vs Hotspot).

    • Ugo

      This lawsuit has nothing to do with the runtime. It's about the API. As long as the same APIs are used on Android, the case would remain the same, regardless of ART or Dalvik.

      • PhilNelwyn

        Oh... yeah. Right.

  • MJ

    It's not the 9th Circuit court. The ruling was made by the US Appeals Court for the Federal Circuit. They normally handle patent appeals, and since the case originally was primarily about patents until they got dropped from the case, the Federal Circuit got the appeal.

    • http://www.androidpolice.com/ David Ruddock


      • MJ

        Also, it might be nice to link to the actual decision an not just Florian's spin. It's over at http://cafc.uscourts.gov/images/stories/opinions-orders/13-1021.Opinion.5-7-2014.1.PDF

        • http://www.androidpolice.com/ David Ruddock

          I'm happy to link to it, but it is crazy long and way too dense for most people to be interested in. Florien's analysis and incessant boasting aside, he pulls very relevant quotes from the decision.

          • h4rr4r

            Florien could be posting the color of the sky, if you took his claims at face value you would have to be the dumbest SOB that ever lived.

          • MJ

            Florian is also being paid as a consultant by Oracle. They put it in a court filing on this exact case. I wouldn't consider him an unbiased source.

          • Sir_Brizz

            TechDirt and Ars both have decent comments on the ruling.

  • FatherPhoenix

    I'm going to take it all with a few tons of salt. You linked to FOSS Patents. Florian is a notable idiot who is biased and a noted shill. I pretty much distrust anything and everything he says.

  • Neill Smith

    Florian of FOSS Patents is a known Oracle shill, why link to him at all much less link to him on the subject of an Oracle case?

    • h4rr4r

      He is a shill for hire. In years past he was employed by MS.

  • br_hermon

    EXCELLENTLY written David. This is why I come to you and AP first when it comes to this legal Android coverage.

    • h4rr4r

      He is taking this from Florian, a known shill. That source could not be trusted to report the wetness of water.

      • Neill Smith

        In fairness he comes to the complete opposite conclusions Florian does.

    • MJ

      I suggest also reading the article from Vox (http://www.vox.com/2014/5/9/5699960/this-court-decision-is-a-disaster-for-the-software-industry) that takes a broad view of what this ruling means if it stands. It's a big change in how APIs have been treated for a very long time.

      I don't trust Florian, who wrote the article from which this story was sourced. Oracle confirmed in a court ruling that they pay him as a consultant.

  • http://mekakiwi.blogspot.com.br/ ED-Z が あらわれた!

    Oh man, I miss Groklaw so much...
    Please PJ, come back! Please!

  • paxmos

    I guess judges of these 2 courts didn't go to the same law school.

    • MJ

      Judges can disagree, it's not uncommon. The major problem is that the CAFC where this appeal was heard is very pro-IP (they're a specialty court that deals with patent law). It's not surprising that they would be biased towards giving more power to a company that makes an argument towards protecting more stuff.

  • Floss

    "In Google's case, this would be to argue that it copied the structure, organization, and sequence of those 37 APIs solely because it was necessary in order to ensure the interoperability of Java applications with the Android OS. I am honestly not enough of a software expert to say whether or not this defense is reasonable, it's very much a fact-intensive issue."
    That is a very weak argument for Google to try to make unfortunately. Because they did a clean room implementation of Java, it does not ensure interoperability. Compiled java apps are completely incompatible with Android. They might be able to argue to a jury that they can be recompiled to work on Android, but at that point a lawyer could easily sway a non-techy jury by arguing at a high level and letting them get confused by the details.

    • Sir_Brizz

      Compiled JAva apps are incompatible, but there is a huge number of Java libraries that are not incompatible because of their implementation.

    • Neill Smith

      I didn't realize interoperability was a binary scale.

    • David Sousa

      As the guy below points, interoperability is not defined just as all or nothing. The point of using Java with interoperability means you can use regular Java code in Android (those shared across with Android) and most Java utilities and code on the Internet, with the incompatibility being only in the part of code that strictly depends on Android functionality. The ability to use a JAR library, for example...

      Not sure, but aren't you confusing interoperability with software extension (a similar concept to the extends keyword)? Just because you're interoperable does not mean you need to refrain from *adding* incompatible software on top of it.

      But yes, I can clearly see why this concept can be lost in a jury!

  • didibus

    I think I finally understood this issue. And I'm afraid, it's a mess.

    Java the language (meaning the syntax, keywords and semantics that make it up) is open sourced by Oracle, and so, Google is indeed free to use it, granted it respects the terms of the license, which I think it did.

    Java the Virtual Machine is owned by Oracle, and while it distribute it freely in most cases, does not let anyone simply use it inside a new OS of them. So Google couldn't use it for free, and did not. Instead, they made their own Virtual Machine. This isn't an issue at all.

    But, the Java Library, which is a set of packages, classes and methods that Oracle makes available standard to people coding in Oracle Java is not open source. Oracle does not let you use it's library of data structures, file system access, etc. in any way you want. Google, recreated this library for use in Android, though they implemented it differently, they did copy exactly it's API. This is what Oracle claims is what they have copyright to which Google infringes.

    The truth is, an API can make a difference in terms of usability, one programmer might favor the way .Net let's you do things, over the way the Oracle Java Library does. One API might make you more productive, might have you code with less bugs, and be more efficient. Think of jQuery, the API really turned JavaScript on it's head and made it so much better and faster to work with, all because of the choice of API, not the implementation of it.

    So I can understand why an API might require protection under copyright, but how do you define how much similarity is similar and how much isn't. A method that Saves something can not be named that many different ways. A Save() versus a save() and a saVe(), are these infringing, or different ideas? It would be annoying if this became a case by case thing, where if you get sued, it would be up to the judge to decide how similar your API is from another, and if it counts as infringing. I think we'd need stricter criterias, so that a programmer could know as he is coding his own API, if he is going to infringe or not. This might not be possible, so it could be that the best solution is to say API are too difficult to make copyrightable, and simply make them not copyrightable. Or maybe a clear percentage of similarity to a complete API should be decided. Compare one API with another, if A is a 51% or more set of B, then it infringes. How would you make the comparison though? Word to word? Method to Method? Class to Class?, Letters to Letters? Line to Line?

    All in all, it is quite a mess!

    • Matthew Fry

      The structure and naming of an API can give some real clues to how the underlying system is designed. In that case, a duplicated API might suggest a very similarly designed system but not necessarily functionally equivalent. So they could argue that a duplication of API suggests a duplication, in general, of the VM.

      If they truly duplicated the API for the sole purpose of making it easier for developers, this truly is a case of the interoperability/reverse engineering defense (as I understand it from David's description).

  • Cesar

    "Anyway, we'll probably hear about this again in 6 months, so be on the lookout! Or don't."

    Sound advice. I'll probably do the latter.

  • Matthew Fry

    "what kind of damage has Oracle actually suffered here?"

    That's a very, very good question especially since the Dalvik VM is open source. It's kind of hard to say "You made money off of this," when it's free. Obviously, money was made and the use of a Java VM had a hand in it but that is one deep dark hole. Open source Java VM inside of an open source OS they don't make money off of, inside of a device they don't make money off of, the use of which generates money in the form of ads and app store transactions. Now what is the average (VM% of OS)/(profit/user)?

    • David Sousa

      They do NOT use the copyrighted Java VM. Dalvik is NOT Hotspot.

      • Matthew Fry

        I never said it was. I was saying that the dalvik VM being an open source but functionally similar and duplicated API makes it difficult to assess damages.

        • David Sousa

          Ah, OK. Sorry, I misunderstood your point.

  • Tim Cook

    Oracle is right in this,I agree

    • Neill Smith

      Oracle would see the development community burnt to ashes around them if they thought it would make them a few extra bucks next quarter.

  • black

    Hmm... so is it a copyright violation to function an inter-operable structure? I certainly hope not!

  • Mark

    Oracle should pay Google. Java would have died years ago, had it not been for Android keeping alive.

    • mrjayviper

      umm no. still used in enterprise

  • alsocurious

    I understand similarly - that copying a typical method declaration like this

    copy ( from, to, length )

    is now a violation of copyright. Fun times a head.
    Perhaps coding will move out of the US?

  • lynx

    Java is a dying language without Android supporting it. Less and less folks are using Java for anything involving computers and indeed even Microsoft has run away from it as best they could in Win8. This is very much a case of biting the hand that feeds you. Mind you, I dislike Java as a language. Of all 7 or so of them that I know, Java is my least favorite to write in.

  • Ken Tsang

    I visit groklaw once a day and hope one day it will reborn...

  • iKrontologist

    Google will no doubt either appeal this to the full appellate jurists (judges) or take it straight to Supreme Court. Meanwhile Congress will have to decide if they are going to weaken these software copyright laws or do away with them all together on API's. Basically a list of names and attributes of the code, they gave to Apache without suing them. Why? Because Apache doesn't have deep pockets. It's why Oracle bought Sun in the first place. Oracle's GREEDY MONEY GRUBBING BRAINLESS LEADER!!!! ....you know the guy who still can't figure out that the Cloud isn't a Cloud afterall!!! lol....

  • Prahlad Yeri

    Here is a real world analogy:

    In a small town, there lived a big Rail-road company called Oracle that used to run hi-speed trains on its metre-gauge railway tracks. It owned full copyrights to the manufacturing of this railway network it called "Java", along with all the equipments involved.

    Few years later, another innovative company called Google came to the town which had better expertise in building railway networks. It performed its own clean-room implementation of a new railway network in a different market and called it "Android". However, it still used the same one-metre tracks (aka APIs) to keep the new network compatible with the existing "Java" network (and primarily, to leverage the skills of existing Java Engineers (aka Programmers) to their new Android system).

    Now, the company called Oracle suddenly gets pissed off that someone else gets to steal their idea of a "railway network" and hence it sues Google. However, the trouble is that Google's implementation of the trains and machinery are that of their own (clean room), hence they go after the only thing that Google had to directly copy for staying compatible i.e. the copyright to creating a one-metre track (APIs)!

    The judges/lawmakers being blissfully unaware of auto engineering concepts, hold that the copyright to the one-metre track rest with Oracle and not copyable by anyone else!