Oracle v. Google: May 2021 Update

Oracle America, Inc. v. Google LLC (Fed. Cir. 2021)

Earlier this year, the US Supreme Court decided this case in favor of Google’s fair use argument and rejecting the Federal Circuit’s opposite conclusion. On remand, the Federal Circuit has now issued a short order in the case sending it back down to the District Court.

In its order, the Federal Circuit recalled its mandate in the case “solely with respect to fair use.”  That means that its judgment favoring Oracle on the question of copyrightability still stands.  Finally, the “district court’s final judgment in favor of Google is affirmed.” Neither party had filed additional briefs in the appeal following the Supreme Court’s decision.

29 thoughts on “Oracle v. Google: May 2021 Update

  1. 5

    Bros, especially Max, I just figured I would bring you an update from the Engleland, home of the ebil Angles. Apparently the left wing overplayed its hand, and didn’t bother to do much in the way of proposing actual policies anyone wanted so the “conservative” “tory” party swept their national level elections. Now they’re going full anti-woke program.

    link to

  2. 3

    API copyrightable but fair use to copy it.

    So, now you have modules with all the names corresponding to the API.

    And it is only another half-step to copy all the support routines as they have names similar to the API that provide support services for the API modules.

    I think logically under this Oracle that you can just copy any software that has an API under fair use. Maybe just shuffle it a bit with an automated shuffler and rename some of the support modules, but I don’t think you can protect any software now.

    I wonder why Trump just doesn’t copy Twitter and FB and set up his own? He would get 100 million joining right away and so it would be viable.

    I wonder what is stopping him?

    1. 3.1

      Sadly, nether Twitter nor Facebook have published the implementation of their API’s. In the case of Facebook, it is 62,000,000 lines of code to run their entire operation.

      At, for example, 10$ per line of code, it would cost 620,000,000 dollars to recreate everything.

      Nor wanting to spend 620,000,000 dollars is stopping Trump from having TrumpFace. (not Trumpbook of course, because Facebook is a registered trademark!)

      1. 3.1.1

        Unless I missed something, the barriers to entry for a T or F competitor are not economic or technological, they’re mostly network effects and some first mover advantage thrown in for good measure.

        (I suppose any barrier to entry could be quantified in terms of dollars. You could say the network effects could be overcome by, e.g., paying every T or F user $50 to switch over. [Not realistic, I know.] But in practice I think it’s a different category from a pure economic barrier.)

      2. 3.1.2

        I doubt 62 million lines of code are necessary for a Facebook implementation.

        And maybe they haven’t published the code but that doesn’t mean it isn’t available or that the executable couldn’t be reverse engineered.

        I bet it wouldn’t be that hard.


          Basically, according to many on the CAFC, there is nothing protectable about Twitter or Facebook.

          And—please—I used to be a developer. A few of us could put together a rough Facebook in a month.


              Stevens said you just write down what you want the computer to do and hand it to a clerk to get the job done.


                I think that was not Stevens, but Breyer.

                I do not think that Stevens would even get to cogitating that far.


                Turns out, it appears to have been Justice Kennedy:

                link to,-CLS%20Bank%20International&text=inventions%20are%20patentable.-,One%20of%20the%20long%2Dstanding%20exceptions%20was%20that%20an,idea%20can%20never%20be%20patented.

    2. 3.2

      Yes, it’s a brilliantly conceived overlooked strategy: Just copy Twitter down to the last pixel and SAY you are Twitter. You won’t have an audience at first of course. And 50 different courts would be finding reasons to stop you, of course, and copying a whole platform down to that last pixel is OF COURSE fair use in this brave new IP regime….


      Naturally the Supreme Court got it backward: wholesale copying is not fair us, but API’s are functional, not expressive and should not be copyrightable. They should be patentable.

      1. 3.2.1

        Remember Martin that Trump has a read audience of about 100 million which is more than enough for a critical mass.

  3. 2


    APIs are copyrightable — but (like most all software copyright) have that right relatively easily distinguished by some “transformation” to which the item (lock, stock and barrel) can be taken and used for.

    Thus endith thou “but copyright is often protection“ mantra.

    1. 2.2

      It’s a great point. As happens on rare occasions, Thomas’s dissent had a kernel (no API pun intended) of insight. He noted that if API copying ends up always or almost always being fair use as a matter of law, then it’s tantamount to no copyrightability in the first instance.

      1. 2.2.1

        I think that such a dissenting comment is a minor point (not to diminish you for saying it).

        Rather, the larger point — from the holding of the case — is that copyrighting itself of ALL software is brought to a ‘tantamount’ point of NO real copyright protection based on the (very loose) face that ‘transformation’ was effected for the large scale, lock-stock-and-barrel BLATANT copying.

        This is a watershed moment in protecting software innovation through copyright.

        Read that as an extinction level watershed moment – based on the looseness of ‘transform’ and just how much that single aspect TURNED every one of the Factors.


          I think we actually agree on the larger point. The majority purported not to resolve the copyrightability issue, but the dissent highlighted that it was in effect deciding it with potentially wide-ranging consequences, as you note.


          That might be a bit strong…you’ll still need to do the sweat-of-the-brow work of actually implementing the API.


            You are kidding, right?

            Just drop by the coffee shop near the college and offer a pizza – anything you want done over the weekend.

    2. 2.3

      The statement that “APIs are copyrightable” is as uncontroversial as it is meaningless. The standard for originality in copyright is really low; a work simply has to possess a minimal degree of creativity to be protected. As a result, even in the early days of API copyright cases, copyrightable subject matter was not seriously contested.

      But obviously the degree of creativity and originality can infect the fair use and infringement analysis in profound ways, and I think this case is an illustration of that. What has always amazed me is that, throughout the case, Oracle pretty much went along with Google’s arguments that framed the Java API around the simplest example in the entire API — the “max(a,b)” method from the java.lang.math class, which returns “a” or “b” depending whichever has the higher value. Google held that one up as the representative Java API method throughout the entire case, and that’s the one the Supreme Court seized on in explaining the Java API. That example made Oracle’s claim seem silly, as it’s hard to see how Oracle should have a copyright monopoly over an API method showing so little originality, and where there are very few alternative ways of expressing the same functionality. But there are literally hundreds of other API method declarations copied by Google that were far more complex and had more than two parameters (and in which the order of the parameters was significant to the function of the method), which showed far more originality and in which there were infinite alternative ways of expressing the same functionality. Oracle made a half-hearted attempt to point this out in its Supreme Court brief but never developed the argument.

      By holding up math.max(a,b), Google was able to give off the impression that Oracle’s claim over the Java API was seeking to monopolize (through copyright) something that was almost entirely functional. Maybe it wouldn’t have mattered in the end, but it always seemed to me that Oracle did itself no favors by allowing the entire Java API to be represented by perhaps its simplest and least original example.

      1. 2.3.1

        While agreeing with your lead-in, I do NOT agree with your take-away of, “and I think this case is an illustration of that.”

        I think that this statement quite misses the boat on what the Supreme Court did in this case – the language that they used – and thus the direct result of eviscerating the ability to rely on copyright for ANY software protection.

        This was decidedly NOT a “well, APIs were going to fall anyway” type of takeaway from the actual legal decision provided by the Supreme Court.

      2. 2.3.2

        was seeking to monopolize (through copyright) something that was almost entirely functional.

        Also expressly incorrect.

        Merger — which is the legal doctrine that WOULD apply in your assertion — was expressly not a part of the holding here.

  4. 1

    The impact of the SCOTUS decision will be interesting to see in the Warhol case now up on appeal at CA2.

    1. 1.1

      Another interesting facet is the CAFC holding on copyrightability remains, but it’s only persuasive authority in the district courts (or other circuits, of course). Technically I believe that’s true even in a district court case with patent claims where an appeal lies to CAFC, because the court would be bound by regional circuit authority on copyright issues.


Comments are closed.