In Part I of this two-part post, I introduced the SCOTUS’ judgment, the concept of API and summarised the majority’s finding. In this Part II, I summarise the minority opinion and share my critique of the judgment.
The minority’s opinion
The minority agrees with the majority insofar as “fair use” being a mixed question of fact and law. The minority disagrees with everything else. The minority criticises the Courts approach of determining the application of an exception/defence without issuing a finding on the existence of the right, i.e., copyrightability. On this point, the minority endorses that the “declaratory code” is copyrightable subject matter since it fulfils the “minimal degree of creativity” test from the earlier ruling in Feist publications. Among other reasons, Google had conceded that Oracle could have created the lines of “declaring code” in many ways, i.e., the doctrine of merger did not apply in this case. The minority holds that the “declaring code” incorporates “implementing code” – both are inextricably bound, and one cannot function without the other. Therefore, if the “implementing code” is copyrightable (a point considered by Google), the minority held that the “declaring code” could not have a different conclusion (Sl. Op. at pages 4 – 7).
On the 1st factor of the “fair use” doctrine, the minority criticises the majority’s position to relegate “declaring code” to lower strata of copyright protected subject matter. The statute considers all computer programs as one group and does not differentiate in the threshold/level of protection. Further, the minority ventured, even if the actual value of “declaring code” is contingent on programmers/developers investing in learning it, familiarising and popularising the same, this cannot result in a lower threshold of protection. The minority alludes to examples of other types of copyrighted works where such logic would fail. E.g., a Broadway musical script contingent on the theatre, the actors and singers investing time and popularising it.
On the 4th factor, the minority notes the majority’s failure to account for the damage caused to Oracle. For instance, evidence suggested that as a direct result of Google’s act, Amazon renegotiated a license fee with Oracle with a 97.5% reduction in royalty rate. Similarly, evidence of Samsung reducing its royalties to Oracle from $ 40 million to $ 1 million was also recorded. Merely because Oracle, by itself, was facing difficulties in the smartphone market, did not preclude Oracle from gaining in that market through licensing arrangements. The alleged possible hold-up situation that Oracle could create was considered by the minority as speculative because other players like Apple and Microsoft created mobile operating systems without using Oracle’s declaring code. Per contra, the minority notes, Google had already been found to have used the Android operating system to violate anti-trust laws. Therefore, alleged potential abuses by Oracle carried little weight as compared to the actual abuse by Google.
On the 2nd factor, the minority criticises the majority for redefining what is understood by “transformative” in copyright law. The mere fact of making new physical products carrying the verbatim code is not “transformative”. “Transformative” is a term used to refer to adding creativity to the underlying work; not adapting a verbatim copy in another platform. The minority notes the undisputed finding of fact that Google’s use of Oracle’s code was not to reverse engineer a system to ensure compatibility or interoperability. Instead, Google’s copying was a simple case of verbatim copying of the “declaring code” and using it for the same purpose/function but in a competing platform.
On the 3rd factor, the minority observes that neither Google nor the majority disputed the Federal Circuits conclusion that Google copied the heart or focal points of Oracle’s work. According to the minority, in such a circumstance, especially where Google contribution was not “transformative” of the copied work and was intended to result in a product that competed with Oracle’s potential revenue, the copying was indeed qualitatively and quantitatively substantial.
Some thoughts
The SCOTUS was confronted with the issue for the first time, and it is fair for the Court, especially the highest court of a country, to consider the contemporaneous situation and technological evolution when laying down the law. But I feel that the majority’s opinion is intellectually dishonest.
The fact remains that Google took the easy way out – copying the “declaration code” from Oracle – whereas other competitors such as Apple and Microsoft did not. This copying was admittedly for a commercial purpose, albeit indirectly achieved – Android OS was made freely available, but Google gained tremendous revenue through ad sales. Much of the revenue volume from ad sales was thanks to Android’s popularity and attributable, in part, to the using of Oracle’s “declaring code” that attracted program developers to build for Android platforms.
This undisputed finding of fact settles this point:
“…While Google repeatedly cites to the district court’s finding that Google had to copy the packages so that an app written in Java could run on Android, it cites to no evidence in the record that any such app exists and points to no Java apps that either pre-dated or post-dated Android that could run on the Android platform. The compatibility Google sought to foster was not with Oracle’s Java platform or with the JVM central to that platform. Instead, Google wanted to capitalise on the fact that software developers were already trained and experienced in using the Java API packages at issue…” 750 F.3d 1339 at 1371.
It is precisely at this juncture of reasoning that the majority contradicts itself. The majority recognises that the creative element in the “declaring code” lies in making it intuitively appealing and easy-to-use so that it becomes popular with the programmers. The popularity with the programmers was the intended effect. However, the majority considers this effect of being an independent factor for which Oracle/the creator gets no credit. Instead, the Court uses this fact against Oracle!
If Google used entirely new “declaring code”, programmers would have had to invest time to learn it and familiarise themselves with it. There was every chance for this to have failed, and Android may not have been successful as successful. This tremendous commercial gain made by Google was established at trial to be at Oracle’s cost – the loss of revenue from licensing deals and thwarting Oracle’s effort to enter the market through 3rd parties.
Consider yet another undisputed fact:
“…[Google and Oracle] negotiated for months but were unable to reach an agreement. The point of contention between the parties was Google’s refusal to make the implementation of its programs compatible with the Java virtual machine or interoperable with other Java programs. Because Sun/Oracle found that position to be anathema to the “write once, run anywhere” philosophy, it did not grant Google a license to use the Java API packages” 750 F.3d 1339 at 1350.
Given this finding of fact, for the SCOTUS to have used a potential/speculative hold-up by Oracle as a factor against Oracle is a non-sequitur.
Finally, the minority’s criticism of the majority is an approach to redefine what is “transformative” hit the nail on its head. It is an excessive twist to the principle – the copied work was not transformed in any way – it was simply copied verbatim and applied to a new platform.
Relevance to India
The Supreme Court assumed copyrightability of the “declaring code” of Sun Java API. But it chose not to interfere with the Federal Circuit’s finding that this was, indeed, copyrightable. To that extent, this litigation journey in Oracle v. Google has significant persuasive value in India on the issue of copyrightability of APIs even under Indian law.
On the “fair use” point, most readers will now be aware that the “fair use” doctrine is distinct from the “fair dealing” doctrine under Section 52(1)(a) of India’s Copyright Act, 1957. This fair dealing exception in the Indian law would not be attracted to the facts in Oracle v. Google because the provision is expressly not applicable to computer programs. Section 52(1)(ab) provides a separate exception relating to interoperability, but given the finding of fact that the copying by Google was not for interoperability, even this exception will not apply. On facts, it does not appear that even sub-clause (aa) would apply. On the face of it, none of the other exceptions contained in Section 52(1) of the Indian Act seems to apply. To this extent, the SCOTUS’ judgment in Google v. Oracle has limited value to the Indian legal setting.
Nevertheless, it does show that the SCOTUS can (and often does) engage in some incredible sophistry to reach a particular conclusion, though it may veer far away from its precedents (or even common sense). The conclusion reached by the SCOTUS on the “equitable” doctrine of “fair use” was anything but equitable! Perhaps, someday in the future, this dissent may trigger an overturning. At the very least, for other jurisdictions that attach persuasive value to US jurisprudence, the dissent provides a reasonable alternative view to be considered.
Note: I represent/have represented clients on related and unrelated issues. Views expressed here are personal.
Really informative summary, thank you!
thank you.
So I am restricting my comments to the thoughts given by the author.
1) “google took the easy way out, when Apple and Microsoft did not”. Well, if you see the history of the case, sun’s API was open source when google had sought to use it, and 10 years later it was bought by Oracle, and google was being restricted, in spite of having used it for around 10 years, suddenly. Also, it is a norm in the software industry to allow for unlicensed usage of API for interoperability purposes, given the core use of API in making alternate platforms and tasks compatible. (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3137209 and https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2171451) – very exhaustive studies on this. )
2) most of the critique is focussed on the 4th-factor analysis, that is the effect on google’s market. But the important question here is by virtue of “use” in a “useful article” like software, the aspect of its functionality and a monopoly thereto cannot be conferred under the guise of copyright. What can be protected, if at all (given that the court assumed it – although I find that act of assumption problematic), is merely the literal publication of the code, and getting money out of that publication – NOT the use. Given monopolizing non-sensory functionality/ utility, goes in the patent territory (if at all). Very interesting extract from an article I read:
“In the aftermath of Alice and court decisions striking down software patents, it may be tempting for courts to interpret the scope of copyright protection expansively, as the CAFC did in Oracle, because the role of patents in protecting program innovations is receding. Under a kind of conservation of IP incentives theory, copyrights might seem to need to be broader to make up for the fact that patents are providing less protection for program innovations. Copyright does important work in protecting programs, but as the Second Circuit recognized in Altai, “fundamental tenets” of this law should not be distorted to fill a perceived gap in legal protection for programs. If software developers need some additional legal protection for industrial design elements of programs that neither copyright nor patent law can provide, they should take their case to fill that gap to Congress, not use the courts to fill the gap through expansive interpretations of copyright protections. Sui generis forms of protection for software have been proposed in the past, and perhaps this approach should be reconsidered. Industrial design laws typically provide a relatively short term of protection against market-destructive appropriations. At present, however, competition and innovation seem to be thriving in the software industry without additional legal protections.”
Copyrights territory is limited to the text and its usage/ copying as text. And verbatim copying, if done qua usefulness, rather than riding out of copying the text directly, cannot be copyrights domain. For example, if I copy a part of art, that is visible in my secondary work, I am riding on your art, as the viewer can sensorily view what you showcased in my work. In the case of a computer software copying like this one here, googles copying was solely qua the functionality (to achieve the function which API allows to), and not at all concerned with riding out of the “expression” of oracle’s work. Also merger doctrine has to have relevance here, to that extent given the usefulness of the work dictates the expression.
3) You say- creativity lies in the “easy-to-use” nature of the work. That itself ensures that its copying – to use – it, rather than to ride on it being intuitively appealing and its textual expression, is what the concern is. That cannot be copyright’s domain. Copyright is limited to the code, its SSO, (i.e. its textual SSO and not its function). Verbatim copying to achieve a function, if prohibited, monopolises the performance of that function, through copyright – which copyright is not supposed to do – a basic analogy to the bikram yoga case. If the movements (however aesthetically creative, beautiful it might be) dictate a functional performance (in bikram yoga being spiritual and health related, and here being utility qua performance of a function on the computer), copyright cannot prohibit use for that functional purpose. Ofc this does not mean it is not copyrightable- it may be, only qua the text being copied for publication.
This decision in fact makes it clear that under copyrights domain, specifically in things like API, copyright (assuming that it exists) only extends to disallow the secondary user from copying the code verbatim, and publishing it- showing the world that hey I wrote this. but if I just use it- copyright cannot prohibit it. Its not its domain. And to that extent- the decision is as sane and aligned with copyrights foundational principles as it can be (we cannot forget the useful articles doctrine)
4) On the aspect of transformative- there are two kinds of transformative uses- content transformativeness and purpose transformativeness- according to precedents like connectix and sega. So here, it is clear contextual transformativeness as google has copied the text and put in the context of different codes, to achieve certain functions, as against the context in which JAVA had put. its like me sampling your song. the purpose is the same- to give that rhythm but the context is different so the resultant expression is ultimately different, when considered as a whole. Further, even the function being performed by this copying is a transformative function (so if you are looking at function inspite of it not being copyrights domain, the transformativeness of the function performed by googles usage also has to be considered- although I wouldnt make this argument, as I fundamentally think that it is not copyrights domain)
Also just for once, imagine the repercussions if API copyrights are permitted, qua functionality – disastrous for the tech world. Not something policy should promote- and in my opinion, copyright can anyway, given it is merely concerned with the expression, and nothing beyond it.
Thanks, Akshat, for joining this discussion and sparking a very interesting discussion.
I haven’t had a chance to read through the articles you have referred to. But overall, I feel, perhaps, the concerns you are highlighting are best addressed by denying copyrightability to all parts of APIs. That’s one approach and in my view, for the type concerns being taken up, that would have been the right approach. But to not interfere with copyrightability issue (and even assume it) but try to wiggle out through the “Fair use” point, I felt is not appropriate.
I won’t say most of my critique (or the dissent’s critique) is limited to 4th factor. I think there are points covering all 4 factors. Perhaps, they are truncated and/or I simply adopt the dissent’s view. In any event, in Harper SCOTUS said that the 4th factor is the most important. Later in Campbell, SCOTUS did not make such a statement, but it affirmed Harper – did not overrule it. Some dist. court orders I have read before have suggested that there is a divergence in approach between Harper and Campbell on the importance of the 4th factor, but I side with the school of thought that the Harper’s emphasis on the 4th factor is still good law.
The interoperability point concerns me – but in my reading of the lower Court orders, despite several contentions on this point, the finding of fact was that Google was not able to establish how and why the interoperability was being affected in this case. I had reproduced one of those extracts. This is precisely why the dissent highlights the finding of fact that interoperability is not the concern here against Oracle, suggesting this interoperability as a false bogie. May be its a larger issue for APIs but not in the facts of this case. That is my reading of the various judgments in this case.
On the ‘transformative’ issue – i think your comment highlights the grey area/debate in US law. I read the cases are suggesting two schools of thought and frankly, a lot of it semantics – how later courts interpret the words used in prior judgments. My thoughts on this are quite lengthy and not suited as a reply to the comment. Perhaps, I’ll get a chance to put out another blog entry on this issue of what is “transformative”.
*If* copyrightability is attached to the portion of the API as argued in this case, I continue to believe that the judgement is not justified.
Thank you for taking the time and sharing your thoughts – the best way to improve discourse.
Thank you for continuing the conversation Adarsh. I will just try and be a little more clear. So what I think is this:
1) Why the court did not address the copyrightability concern: If you see US SC’s Star Athletica judgment- on useful articles doctrine and conceptual separability- the court had held that in the case of industrial designs, the aesthetic features of “useful articles” can still be copyrightable if they are separable from the industrial aspect of it. J. Breyer had dissented saying that no- any part of industrial designs being copyrighted is wrong- as it monopolizes the functionality under the garb of the expression- something which copyright cannot do. Now, if we see the definition of computer software in the US Act- its called- series of “instructions” to be used. and then s. 102 (b) prohibits protection of methods of operation/ processes. What are instructions anyway? arent they processes? So to that extent- due to that conflict in opinion qua star atheletica- I think the court did not enter the realm of copyrightability of software – and reasonably so.
But that cannot and should never mean that copyright permits monopolizing functionality, under the garb of protection of the expression, i.e the code. even assuming copyrightability, which the court did- it obviously can only be limited to the code as a text- and not as a useful article – so for example- if tomorrow I go on to write a book on the codes I have written in my life- and copy some code you had written and publish it as mine- you can prohibit me from doing so- given you have a copyright on that “code” as a “literary work”. But if I copy the code- to USE it, and you get to prohibit me, doesn’t the monopoly qua the code- extend to its usage as instructions/ or employment of the process embedded in the code? Can copyright do that? I don’t think so.
2) Therefore, any “usage” of the code- which is not at all related to publication of it, or related to claim ingenuity on it by a non-author- has to be transformative- as its “use” which is being done- rather than it being appropriation/ copying to claim as ones own, as a “literary work”.
3) Respectfully, IMO even if the market factor is the most important one- here oracle’s market – WHICH COPYRIGHT protects- that is its market qua the literary features of the code- is not being affected, but rather it’s market qua the “use” of the code – in terms of the use of the instructions embedded within the code- that is being affected. That market is not copyright’s concern- and would be requiring a higher threshold for a monopoly- or a perhaps sui generis regime- but copyright? I don’t think so. That’s not the relevant market here.
CONC – So imo, even if copyrightability is attached (and rightly so as codes may be written with signatures etc., to make it aesthetically creative (considering the experiencer itself being a coder) and within the domain of copyright, under the rationale of Star Athletica), the protection has to only be limited to the copying of the code and claiming it to be one’s own, or using it, literally, but cannot extend to copying it to “use the instructions that are embedded within it”. Its basically claiming a monopoly on the healing practices which the yoga steps allow, through a monopoly on the yoga steps – or claiming a monopoly on the function of the particular innovative bicycle design, by claiming copyright on its aesthetic part (Re Brompton Bicycle), or in simple terms- claiming a monopoly on the “utility” of the code, by a copyright on its literal structure, as a literary work. I don’t think that’s copyrights domain at all.
Also- looking forward to the post on transformative works- thanks for the nice discourse. V glad. 🙂
Thanks, Akshat. I agree that the majority in Oracle is a reflection of the dissent from Star Athletica. This supports my overall view that the majority in Oracle used “fair use” doctrine to achieve indirectly what it could not directly – i believe the underlying tone of the majority was to deny copyrightability but they did not want to (or could not do so – I will leave it you and readers to determine which is more appropriate) and so they impermissibly twisted “fair use” to reach the same practical end.
Your view seems to be that there is a monopoly qua the literary aspects and monopoly qua the functional aspects. You further add that the copying in this case was purely functional and that copyright was not intended to prevent this.
I see three legal questions in a case like this:
1. whether copyright exists?
2. whether infringement is proved?
3. whether infringement can be defended under any exceptions, such as “fair use”?
Lets forget step 1 because that is assumed in favour of Oracle. I don’t see how you make this distinction when assessing step 2 of infringement. If you “reproduce”, you infringe – step 2 ends there. ‘How’ or ‘Why’ you reproduce is irrelevant in step 2. The statute does not care and a Court cannot exceed the statute.
‘How’ and ‘Why’ you reproduce can only be relevant in Step 3 depending on the statutory exception employed.
The ‘why’ is certainly part of the 2nd factor of “fair use” – whether it is transformative or not. I believe you are also suggesting that it is part of 4th factor as well – market qua the “literary features” of the code and the market qua the “use” of the code. I disagree that one can make such a distinction in the 4th factor. But forget my opinion for a moment – the majority itself concedes in several places that the separating code from its functionality in this case is not possible – if so, I don’t see one can split markets like this hypothetically.
Hi Adarsh. Thank you. Continuing the conversation. Okay so I think step 1 is important. Copyright exists- but on what does copyright exist? The court has many times in the judgment referred to functional aspects not being copyrightable- to some extent denoting that even if copyright is permissible, it is “thin”, given the “inextricable” link between the literary features of the code and its functionality.
Now step 2- I agree there is definitely copying but I would be careful to call that infringement, given a determination of “infringement” is subject to s. 107 of the US Act. I don’t think there was any infringement here, given the copying which was done was not the exploitation of the programs commercial value “as a copyrighted work” (PDF page 39). The point is we cannot use copyright to allow to monopolize functional elements, and I hope we consent to this point, given how it is not copyright territory to do this.
Getting to step 3, “why” is a part of “purpose of use” agreed – and the court here smartly says that okay- if we are going to assume copyrightability of functionality, then use for a transformative functionality (or use for a purpose which ultimately results in a different function, one which sun intended to get into but was not successful) is transformative. The court took this reason only because the copyrightability of an inherently functional “work” was in contention.
Also, if we do not segregate the market qua literary features and utilitarian features, doesn’t it directly result in a market monopoly, through copyright, in utilitarian features? Isn’t the standard of patentability high qua such utilitarian features? Doesn’t a conclusion of a combined “use” + “literary feature” market being monopolized through copyright, completely render the merger doctrine/ scenes a faire doctrine, and methods and processes not copyrightable doctrine (standard old doctrinal principles of copyright coming from baker v. selden, mazer v. stein etc. otiose?
I think the copying for “publication” and copying for “use” distinction is important. The primary is in copyrights domain, the latter can never be. For eg. copying recipes and publishing them is wrong, but using that recipe cannot be protected by copyright. It may feel wrong- maybe I deserve a monopoly on authorizing use of the recipe I wrote- given how new and interesting it is- but sorry copyright is not the way. the standard for such monopoly has to be higher- given utility being involved rather than mere experience/ aesthetic sensations.
And no- imo this does not bulge into the copyrightability territory. The copyrightability question, which the court ignored was different – whether the literary features of the code (which are dictated by functional considerations) should be allowed to be copyrightable or not? not whether the functionality is within the domain of copyright or not – of course, it is not. But sometimes, literary features of codes can be creative/ original and so the court didn’t want to draw this boundary. But interpreting this to mean that copyright can protect functionality, is a slippery slope- which will end up in overlaps with patent, industrial designs, and many other completely “utilitarian” doctrines. The fair use invocation by the court was fruitful to the extent- the court drew a boundary that even in software protection through copyright- we can never go on to monopolize functionality or use- given use will almost always result in a transformative function – or something different from what sun/ oracle intended to achieve out of it- as direct output.
I completely agree with you on this. While I find that one may disagree with Justice Thomas that decleyratory codes are not ideas, but expression of given ideas, it is hard to ignore his points on fair use. The majority’s opinion, I felt, is clouded too much by the fact that Google’s actions promoted a number of software developers and created an ecosystem for program development. In particular, the opinion on market effect does not make much sense wherein the majority chose to focus on hypothetical scenarios of what would’ve happened than what Google actually did. I fail to understand as to how Google’s work can be said to be transformative in any sense of word when it is what APIs are meant to do.
Overall, I agree with your analysis Mr. Adarsh.
Thanks Apoorv. On Thomas J.’s view on copyrightability, you should consider (perhaps, you already have) the Fed. Cir. opinion on it. They reach the same conclusion but they sing different tunes in my reading.
Hello sir!
You have mistakenly interchanged the “1st factor” and the “2nd factor”.
A brilliant article otherwise, always look forward to your pieces!