24 May 2020

Rokt’s Computerised ‘Marketing Scheme’ Fails Patent-Eligibility Test on Appeal

Web servicesFollowing an appeal by the Commissioner of Patents, a Full Bench of the Federal Court of Australia has overturned a decision of a single judge of the Court relating to the patent-eligibility of a computer-implemented invention.  The Full Court has ruled that a claimed method and system, developed by Singapore-based Rokt Pte Ltd, for providing ‘a dynamic, context-based advertising system, introducing a distinction between an engagement offer, without a direct advertising benefit, and an advertisement designed to lead directly to the sale of the product’ is not patent-eligible subject matter under the Australian ‘manner of manufacture’ test: Commissioner of Patents v Rokt Pte Ltd [2020] FCAFC 86.

Broadly speaking, the claimed invention provides a mechanism whereby a user has an opportunity to engage with an offer (characterised by the Full Court as ‘click bait’) while accessing a web site before being presented with targeted advertising.  In this way, advertisements are presented only to those users that are most likely to interact with them, and make a purchase.  Since advertisers must often pay for placement of their advertisements within web pages, the invention provides an improvement in that the costs associated with placing advertisements in front of consumers who do not interact with them may be reduced.

In an earlier Patent Office decision, a Hearing Officer had found that ‘the substance of the invention in this case amounts to business innovation’, which was not patentable, and therefore refused Rokt’s patent application: Rokt Pte Ltd [2017] APO 34.  (A previous decision, relating to an alternative set of claims proposed by Rokt, had reached a similar conclusion: Rokt Pte Ltd [2016] APO 66.)

Rokt appealed to the Federal Court, where the primary judge (Justice Alan Robertson) reversed the Patent Office decision, finding that ‘[t]he invention solved not only a business problem but also a technical problem’, ‘… there was a business problem of attracting the attention of the user and having the user choose to interact with the advertiser, but this problem was translated into the technical problem of how to utilise computer technology to address the business problem’, such that ‘use of computers was integral, rather than incidental, to the invention in the sense that there is an invention in the way in which the computer carries out the business scheme’: Rokt Pte Ltd v Commissioner of Patents [2018] FCA 1988 at [205]-[208].

The Commissioner, in turn, appealed to a Full Bench of the Federal Court (composed of Justices Rares, Nicholas and Burley), which has now allowed the appeal, finding that ‘the invention is a scheme or, more accurately, a marketing scheme’, that ‘nothing about the way that the specification describes the computer hardware or software indicates that either is any more than a vehicle for implementing the scheme, using computers for their ordinary purposes’, and that ‘the claim amounts to an instruction to carry out the marketing scheme’ expressed at a ‘level of abstraction [which] demonstrates that it does no more than provide a list of steps to be implemented using computer technology for its well-known and understood functions’: Commissioner of Patents v Rokt Pte Ltd [2020] FCAFC 86 at [108], [109], [115].  As such, the Full Court concluded that Rokt’s claimed invention is not a patent-eligible manner of manufacture, and that ‘the learned primary judge erred in finding otherwise.’

More Than ‘An Idea for a Computer Program’ is Needed

Among the findings of the primary judge was (at [205]) that Rokt’s computer system:

… provided a single platform in which user engagement data could be coupled with transactional data and user context data to provide a personalised ranking of engagement offers to the user. This technical problem of providing this single platform was solved by introducing the tracking database and the objects database and designing the ranking engine and the engagement engine which accessed and manipulated the data in the two databases to rank and select engagement offers. The ranking engine optimised the personalised output for the consumer. Critically, the ranking engine implemented a ranking algorithm which ranked the retrieved object by a combination of an engagement score and revenue score.

While this certainly sounds like a specific technical solution that might be found patent-eligible, the difficulty for Rokt identified by the Full Court is that the patent specification neither describes nor – critically – claims a particular technical implementation of these features.  For example, while claim 1 recites an ’engagement engine’ which ‘implements a ranking algorithm operable to dynamically rank the relevant engagement offers based on at least one of: (a) an engagement score …; [and] (b) a revenue score…’, the specification states that the provided examples should not be seen as limiting and that any suitable scoring regime could be implemented for the recorded metrics.

This highlights a recurring problem with many patent applications relating to computer-implemented inventions: namely, a desire to claim the invention at a sufficient level of generality to ensure that a broad range of technical implementations and variations are protected.  However, where the field of application of the invention relates to subject matter that is not patent eligible – such as a marketing scheme – the risk is that a claim cast in abstract terms will, as the Full Court has found in Rokt, amount to no more than an instruction to carry out the unpatentable method using generic computer technology for its well-known and understood functions.

Referring to the decision of an expanded Full Bench of the Court in Encompass Corporation Pty Ltd v InfoTrack Pty Ltd [2019] FCAFC 161, the Rokt Full Court stated (at [114], emphasis in original, citations omitted):

In Encompass the Full Court found that where the claims in suit do not secure, as an essential feature of the invention, any particular software or programming that would carry out the method and the method is entirely left to those wishing to use the method to devise and implement a suitable program for that purpose and all the specification teaches is that the processing system may be “suitably programmed” then the method is really an idea for a computer program, it being left to the user to carry out the idea in a computer system. In making these observations, the Full Court did not purport to preclude a scheme implemented using computer software from patentability. It manifestly did not lay down principles of general application. It was considering whether invention lay in the implementation of computer technology based on the disclosure of the specification. By the claim and the specification leaving entirely to those wishing to use the method to devise and implement a suitable computer program for purpose, it was apparent that the invention did not rise above the level of being an instruction to use computer technology for its well-known and understood functions to implement the scheme.

What is the Role of Expert Evidence in Assessing Patent-Eligibility?

In writing about the decision of the primary judge, I expressed some concerns about the hefty role played by expert evidence in the Rokt case.  I noted in particular that:

…three-quarters of the 217 paragraphs making up the complete judgment in [the decision at first instance], are devoted to a discussion of the evidence, which included two affidavits of the expert engaged by Rokt (Professor Karin Verspoor), an affidavit of the expert engaged by the Commissioner (Mr Scott Ries), and an account of the cross-examination for Professor Verspoor at the hearing.  I find this vaguely disturbing.  It is a more extensive discussion of evidence than you would often see in decisions relating to traditionally fact-intensive inquiries, such as the assessment of inventive step.  The cost to the parties in adducing this evidence, and to the court in considering it, would have been substantial.

The Full Court in Rokt also expressed concerns about the role played by expert evidence in the case.  In fact, reliance on the experts was found to be a major factor that led the primary judge into error.  As the Full Court explained:
  1. whether an invention comprises patent-eligible subject matter is a question of law that ‘is to be answered according to a common law methodology under the rubric of “manner of manufacture” as developed through the cases’ (at [68]);
  2. to answer this question, the claimed invention must be characterised as ‘a matter of substance, not merely of claim form’, which ‘requires definition of the allegedly patentable invention, based on the construction of the claims’ (at [69]);
  3. it is fundamentally a matter for the Court, not for any witness (however expert), to construe the specification, including the claims (at [71]);
  4. in this, the role of expert evidence is limited, being ‘to place the Court in the position of the person acquainted with the surrounding circumstances as to the state of the art and manufacture as at the priority date’ such that, ‘[t]ypically, the Court will read the specification with the benefit of expert evidence as to the meaning of words that are terms of art, or with an explanation of technical concepts relevant to the understanding of the invention as described and claimed’ (at [73]).
Accordingly, the Full Court found that in treating patent-eligibility of the Rokt invention as a problem ‘in the realm of fact, and preferring the evidence of Professor Verspoor over that of Mr Ries in order to resolve it, [the primary judge] fell into error’ because ‘Professor Verspoor could not address and determinatively answer the legal question of the proper construction of the specification and characterisation of the claimed invention’ (at [72]).

It has long been recognised that the opinions of technical experts regarding the ultimate legal questions requiring decision by the courts in patent cases are likely to be of limited, if any, value.  For example, with regard to inventive step the Full Court in Minnesota Mining & Manufacturing Company v Tyco Electronics Pty Ltd [2002] FCAFC 315 stated, at [50]:

Questions of obviousness and inventive step are ultimately for the Court to determine, irrespective of the opinion expressed by any number of experts. Evidence by an expert as to whether a claimed invention was obvious or did not involve an inventive step will of necessity be essentially argumentative. Even if admissible, such evidence is likely to result in a waste of time by reason of the analysis of and cross-examination of the evidence that will be required.

Despite this established understanding, it appears that the experts in the Rokt case – and Rokt’s expert, Professor Verspoor, in particular – were asked to provide opinions on the ‘substance of the invention’ even though, as noted above, this is ultimately an issue of construction that is fundamentally a matter for the court.  The Full Court found that the primary judge fell into error in preferring the opinion of Professor Verspoor over that of Mr Ries, and failing to give separate attention to the issue of construing the specification to determine the ‘substance of the invention’ for himself.

Henceforth, then, it should be clear that the role of expert evidence in ‘manner of manufacture’ cases – if required at all – is limited primarily to such matters as informing the court of the meaning of terms of art and explaining technical concepts relevant to the understanding of the invention.  This can only be a good thing, and hopefully Rokt will prove to be the high water mark in the deployment of experts in such cases!

What is the Role of Prior Art in Assessing Patent-Eligibility?

A further vexed question that is addressed to some degree by the Full Court in Rokt is the extent, if any, to which prior art – including common general knowledge – is relevant to determining whether or not a claimed invention is a patent-eligible ‘manner of manufacture’.  In recent years, IP Australia has increasingly adopted an approach to patent eligibility in which prior art may be taken into account in assessing the ‘contribution’ to the art made by a claimed invention.

For example, in the Patent Office decision of McCormick & Company, Incorporated [2017] APO 62, the Hearing Officer determined (at [54]) that he was ‘satisfied that it is legitimate to consider the prior art in order to assess the actual contribution to the art, and thus to determine the substance of the invention.’  The same reasoning was adopted in Google LLC [2018] APO 13 (at [26]).  IP Australia’s Patent Manual of Practice & Procedure (a.k.a. the Examiners’ Manual), states in Section that:

…while examiners can take into account what the specification asserts is the contribution made by the invention, and can even make an assessment by way of consideration of the common general knowledge, the substance is to be determined on an objective rather than subjective basis. In order to do this, consideration should be given to the claimed invention and the contribution that it makes to the art.  Logically this consideration may take into account the prior art at the priority date rather than merely common general knowledge.

A useful way of approaching this issue in the context of computer related inventions, particularly when dealing with claims involving physical system architecture, is to consider which features of the claim confer novelty over the prior art.  For example, where the system architecture is known apart from the particular program that is operating, the substance of the claim will generally be the program and not the system.  On the other hand, where the claim is directed to a novel architecture it is sometimes helpful to ask whether the individual integers are known in the art, and/or whether the combination of the integers gives rise to a new physical computer system/device that does more than merely automate a process.

In a series of seminars held in late 2018, on its approach to the examination of computer-implemented inventions (a recording of the webinar version is available on YouTube), IP Australia justified its approach by reference to the High Court’s decision (relating to claimed gene sequences) in D'Arcy v Myriad Genetics Inc [2015] HCA 35.  It suggested that this decision (and in particular the concurring, but minority, judgment of Gageler and Nettle JJ) implies that ‘it is necessary to compare the invention as claimed to the state of the art to determine what has been added to the sum of human knowledge by the work of the inventor.’

I believe that this position is becoming increasingly untenable.  The Full Court in Encompass largely evaded the question of whether the primary judge in that case had erred in conflating the ground of ‘manner of manufacture’ with those of novelty and inventive step, on the basis that he arrived at the correct conclusion.  Nonetheless, the Encompass Full Court did appear to acknowledge that his reasoning could be interpreted as allowing considerations properly associated with novelty and/or inventive step to have influenced his conclusions on patent-eligibility; and that, to the extent that this were so, it would have been an incorrect approach (see my earlier article on the case for a more detailed analysis).

In Rokt the Full Court has been far less coy.  While acknowledging that the authorities (including D’Arcy) require that a patent specification be construed in the light of common general knowledge and the prior art (see, e.g., at [68] and [71]), the Full Court is also clear that the use of prior art (being, in this case, only the common general knowledge) is limited.  The Court stated, at [85], with reference to a passage in Commissioner of Patents v RPL Central [2015] FCAFC 177 (at [96]) endorsed in Encompass (at [95]), that:

The reference to using the computer for its “well-known and understood functions” involved consideration of computers having regard to their basic and well-known functions. This did not require, and should not be taken to encourage, a review of the common general knowledge beyond the use of the common general knowledge, to the extent necessary, to construe the specification.

Further, the Full Court stated, at [91] (emphasis added):

It is apparent that where the cases refer to “generic software” or to the use of computers for their “well-known” purpose, it is not a finding as to common general knowledge. Rather, it is a reference to computer technology that is utilised for its basic, typical or well-known functions. The means of determining that this is so is primarily by a careful review of the specification in order to ascertain, by construing that document, whether the invention described and claimed is in substance any more than a scheme that utilises computers in such a way. This is a question of characterising the invention as set out in the specification.

This, in my view, situates ‘manner of manufacture in its proper, and conventional place, as established by the High Court in National Research Development Corporation v Commissioner of Patents [1959] HCA 67 (NRDC), as a matter to be determined on the face of the specification, properly construed.

Conclusion – A New Hope for Greater Certainty and Consistency?

Overall, I think the Rokt decision is a good one, with the potential to bring greater clarity to the law relating to ‘manner of manufacture’.  Its succinct summary, at [114], of the central holding in Encompass, is helpful.  Its analysis, at [106]-[115] – while largely specific to the case at hand – provides a useful example of how to evaluate a claim directed to a computer-implemented scheme to determine if it amounts to something more than the underlying, unpatentable, scheme, e.g. whether there is, in the implementation, an improvement in computer technology, or invention in the computerisation.  In particular, the Full Court clearly explains how it arrives at the conclusion that Rokt’s claim 1 amounts to no more than instructions to carry out an unpatentable marketing scheme, despite the length and level of apparent technical detail of that claim.

Hopefully, the Court’s discussion of the limited role of expert opinion in the eligibility analysis will reverse the trend of increasing reliance on expert testimony in these cases.  Rokt unequivocally confirms that questions of construing the specification and claims, determining the ‘substance of the invention’, and determining whether or not the claims recite a patent-eligible ‘manner of manufacture’ are matters of law for the court to decide, and not questions of fact for an expert.

Related to this, the Full Court has confirmed that knowledge of prior art, including common general knowledge, is relevant only insofar as it informs a proper construction of the specification and claims.  This is, I believe, the correct interpretation of High Court authorities such as D’Arcy and Kimberly-Clark Australia Pty Ltd v Arico Trading International Pty Ltd [2001] HCA 8 (at [24]).  The more expansive approach that has been developed by the Patent Office, under which it is allegedly ‘legitimate to consider the prior art in order to assess the actual contribution to the art, and thus to determine the substance of the invention’ is clearly wrong in view of Rokt, and should be abandoned.

There will be disappointment in some circles at the Full Court’s reversal of the ‘positive’ outcome of the first instance decision in Rokt.  I fully understand how the applicant and its representatives will be feeling, having had the same experience in the RPL Central case.  In my view, however, Rokt is the clearest and most coherent of the recent Full Court decisions relating to patent-eligibility of computer-implemented inventions.  It is my hope, therefore, that it will help bring some much-needed certainty and consistency to decision-making in this contentious area of Australian patent law.


Post a Comment

Copyright © 2014
Creative Commons License
The Patentology Blog by Dr Mark A Summerfield is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Australia License.