31 August 2022

High Court’s Failure Exposes the Festering Eligibility Sore in Australia’s Patent Laws

DisappointmentIn a keenly-awaited – and thus hugely disappointing – ‘decision’, the High Court of Australia has failed to satisfactorily resolve the question of whether patent claims directed to electronic gaming machine (EGM) technology developed by Aristocrat Technologies Australia Pty Ltd are directed to patent-eligible subject matter: Aristocrat Technologies Australia Pty Ltd v Commissioner of Patents [2022] HCA 29.  With Justice Gleeson calling in sick on the crucial days, arguments in Aristocrat’s appeal from the adverse decision of the Full Federal Court of Australia were heard by just six judges of the High Court, which astute readers will notice is an even number.  The diminished bench split 3:3, with Kiefel CJ, Gageler and Keane JJ finding Aristocrat’s claims unpatentable, while Gordon, Edelman and Steward JJ would have allowed the appeal, concluding that the claimed invention was patent-eligible.  In this situation, section 23(2)(a) of the Judiciary Act 1903 provides that ‘the decision appealed from shall be affirmed’.  It does not, however, provide that the reasons of the court appealed from are affirmed.  On the contrary, one thing on which all six High Court judges appear to be in agreement is that the two-step approach taken by the Full Federal Court majority was not correct.  We were hoping for some clarity on the patent-eligibility of computer-implemented inventions.  Instead, we have been delivered two opposing, but equally authoritative, outcomes in the High Court, along with a Full Federal Court decision that stands affirmed, but which has been disapproved.

To say that this is an unsatisfactory situation would be one of the great understatements in the history of Australian patent law.  Oh, how the mighty have fallen!  In its famous and rightly-lauded 1959 decision in the NRDC case ([1959] HCA 67) the High Court delivered a judgment on patent-eligibility of such wisdom and prescience that the Australian government chose to retain the archaic language of ‘manner of manufacture’ as the touchstone of eligibility in the Patents Act 1990, placing its trust in the courts to continue development of the law to flexibly embrace the emergence of new and unpredictable technologies.  Yet here we are in 2022, and the High Court has tripped over the now relatively mature technology of computer software and tumbled flat on its face.

What follows is, I concede, a lengthy article, even by my usual standards.  But I am attempting to take up some important issues, and to suggest a different way forward on the treatment of computer-implemented inventions in Australia.  Following a relatively brief review of what actually happened at the High Court, I attempt to determine what, exactly, the split decision means for the lower courts, and other decision-makers such as patent office examiners and hearing officers that are bound by the authority of the courts.  I then look at how we arrived at the unsatisfactory position in which we now find ourselves, and argue that division among judges is not a new feature in this area of the law, and has in fact been evident in the Federal Court for some years. 

This leads me to the view that judicial development of the ‘manner of manufacture’ test has failed us in relation to computer-implemented inventions.  I think that a new approach is required to bring some certainty and clarity to the law and its application.  Firstly, I suggest that it would be productive for IP Australia to consult more widely with stakeholders in developing its examination practice in the wake of the Aristocrat case.  Beyond this, I think the Australian government needs to clarify its policy in relation to the protection of digital innovation, and to reach a principled position on the appropriate balance between competition and the patent incentive in the case of computer-implemented inventions.  And if the current law is not working for us, then the parliament may need to consider a legislative solution.

What Happened at the High Court?

For convenience, and to avoid any ambiguity, I shall refer to Kiefel CJ, Gageler and Keane JJ as the anti-Aristocrat judges and to Gordon, Edelman and Steward JJ as the pro-Aristocrat judges.

Both groups of judges agreed, following the majority decision in N V Philips Gloeilampenfabrieken v Mirabella International Pty Ltd [1995] HCA 15, that section 18 of the Patents Act imposes a threshold requirement that there be an (alleged) invention disclosed on the face of the specification, as it would be read and understood by the skilled person in light of their common general knowledge in the relevant field of endeavour.  (See anti-Aristocrat judges at [66]-[72], and pro-Aristocrat judges at [106]-[111]; but note that the pro-Aristocrat judges recognised, at [110]-[111], that the Philips decision has been attended by some controversy.)  It was not suggested by either group of judges that Aristocrat’s claims did not satisfy the low threshold requirement, and each group proceeded to address the question of whether the claimed invention comprised patent-eligible subject matter, i.e. was ‘a manner of manufacture within the meaning of section 6 of the Statute of Monopolies’ as required by sections 18(1)(a) and 18(1A)(a) of the Act.

It was here that the pro- and anti-Aristocrat judges parted ways, being divided on the proper characterisation of the claimed invention.

The anti-Aristocrat judges took the view that ‘[t]he claimed invention takes its character, as an invention, from those elements of the claim which are not common general knowledge’, thereby concluding that ‘there is no reason to characterise the claimed invention as other than a claim for a new system or method of gaming: it is only in relation to the feature game that the invention is claimed to subsist’ (at [73]).  They considered this approach to be consistent with Philips and with the Court’s decision in D'Arcy v Myriad Genetics Inc [2015] HCA 35 (on the patent-eligibility of isolated gene sequences) which, they said, requires Aristocrat’s invention to be characterised ‘by reference to the terms of the specification having regard to the substance of the claim and in light of the common general knowledge’.  The anti-Aristocrat judges thus concluded that the claimed invention was ‘no more than an unpatentable game operated by a wholly conventional computer, using technology which has not been adapted in any way to accommodate the exigencies of the game or in any other way’ (at [76]).

The pro-Aristocrat judges accepted that ‘[i]t is well established that a “mere scheme or plan”, or any other merely intellectual endeavour, cannot be the subject matter of a patent’ and that this exclusion encompasses the rules for playing a game (at [116]).  However, they found that game rules ‘may be … proper subject matter when combined with physical materials used for playing the game’ (at [118]).  While they considered that merely using a machine to manipulate abstract ideas would not be sufficient to confer patentability upon an otherwise unpatentable scheme (at [120]-[121]), this was to be contrasted with ‘an idea implemented on a computer to produce an artificial state of affairs and a useful result’ which may be patentable (at [124]).  Following their approach, the pro-Aristocrat judges concluded that ‘the best characterisation of Claim 1 is: an EGM incorporating an interdependent player interface and a game controller which includes feature games and configurable symbols’ (at [149]).  They went on to find that ‘the operation of the game controller cannot be severed from the interdependent player interface in the EGM’ with the result that the claimed invention ‘is an altered EGM involving an artificial state of affairs and a useful result amounting to a manner of manufacture’ (at [154]).

The approach of the anti-Aristocrat judges might be regarded as an ‘inventive contribution’ analysis.  In assessing whether there is patent-eligible subject matter, they stripped the claim of all previously-known elements (i.e. the EGM hardware and pre-existing functionality), characterising the claimed invention by reference to only those elements that, on the face of the specification read in light of the common general knowledge, are fairly alleged to be new.  Where this contribution falls solely within excluded subject matter (being, in Aristocrat’s case, merely the rules of a game) then the entire claim is ineligible.

Conversely, the pro-Aristocrat judges’ approach might be regarded as a ‘whole of claim’ analysis.  Having established that the claim satisfies the low threshold requirement of being for an (alleged) invention, they directed their attention not to the game rules in isolation, but rather to the interdependence between the game rules and the other elements of the claim.  Significantly, the pro-Aristocrat judges did not require ingenuity to reside in patent-eligible elements of the claim (e.g. in an improvement or variation to the hardware).  On the contrary, they observed that ‘the method of carrying out the idea in the computer, and the artificial state of affairs and useful result, need not be inventive or ingenious’ and that ‘[t]he ingenuity may lie only in the idea but, when the idea is applied to produce an artificial state of affairs and a useful result, there will be a manner of manufacture’ (at [123]).

Authority of the High Court Judgments

As I have already noted, the result of the High Court’s 3:3 deadlock is that the Full Court judgment, from which Aristocrat appealed, is technically affirmed.  But this does not mean that the reasons of the majority in that judgment are affirmed.  On the contrary, the anti-Aristocrat judges expressly disapproved the Full Court majority’s two-step test, which asks whether the invention is a computer-implemented invention and, if so, whether it can broadly be described as an advance in computer technology.  They stated that this approach ‘unnecessarily complicates the analysis of the crucial issue’ and observed that ‘[i]t is not apparent in the present case that asking whether the claimed invention is an advance in computer technology as opposed to gaming technology, or indeed is any advance in technology at all, is either necessary or helpful’ (at [77]).  The pro-Aristocrat judges can be assumed also to disagree with the Full Court majority, given that they reached a different conclusion by different reasoning.  Additionally, and more explicitly, they observed that ‘[t]he question of whether a claim, as properly characterised, is the proper subject matter of a patent should not be deconstructed to require, separately from the general principles of patentability, consideration of whether the subject matter is “computer-implemented”’ (at [152]).

Furthermore, the ‘inventive contribution’ analysis of the anti-Aristocrat judges is fundamentally incompatible with the ‘whole of claim’ analysis of the pro-Aristocrat judges.  Clearly there will be many cases in which, as here, these two approaches will produce opposing outcomes.  In such cases, it would be impossible for an inferior court, or other decision-maker, to follow them both.  And while the outcome before the Full Court has been affirmed, it is clear that the approach of the majority in that case has been disapproved by the High Court and should not be followed.  So what should a decision-maker do when placed in such a position?  And, importantly, what should the Patent Office do, given that it is faced with making such decisions on a regular basis?

As far as I have been able to ascertain, there has been very little judicial consideration of the authority of divided decisions of the High Court.  In Re Wakim [1999] HCA 27 the High Court was faced with the question of whether to reopen an earlier judgment in which the Court was, as here, equally divided in opinion, and which had therefore also been decided under s 23(2)(a) of the Judiciary Act.  In that case, Gummow and Hayne JJ stated that while the earlier decision ‘disposed finally of the appeal and bound other courts in Australia to arrive at a like result on the issues it dealt with, it established no principle or precedent having authority in this Court’ (at [100]).  The question of the Court reopening or reconsidering its earlier decision therefore did not arise.

In Harvard Nominees Pty Ltd v Tiller [2020] FCAFC 229 the Full Court of the Federal Court of Australia was presented with the question of whether it should follow ‘seriously considered dicta’ of the High Court in a case where, although the Court reached a 5:1 majority in the ultimate disposition of the appeal at hand, only three of six judges joined in the opinion containing the dicta in question.  In its unanimous opinion, the Full Court found that High Court authority did ‘not mandate that the seriously considered dicta of three members of a six judge bench of the Court is to be followed’ because ‘[t]hat is simply not “the majority of the High Court”’ (at [55]).

Arguably, then, lower courts and other decision-makers are bound to follow the High Court in Aristocrat only to the extent that the opinions of both the anti- and pro-Aristocrat judges commonly establish or confirm relevant principles.  I would tentatively suggest that the following such principles can be identified:

  1. that there is a (low) threshold requirement that an (alleged) invention be disclosed on the face of the specification;
  2. that Aristocrat’s claims satisfied the threshold requirement, and that the question to be decided was whether what is thereby alleged to be an invention is a patent-eligible manner of manufacture under s 18(1)(a) or s 18(1A)(a);
  3. that the two-step test proposed by the majority of the Full Court is not a correct approach to assessing patent-eligibility of a computer-implemented invention; and
  4. in particular, that the general principles of patentability established through cases such as NRDC and Myriad apply irrespective of whether or not the (alleged) invention can be said to be ‘computer-implemented’.

Beyond these principles, the two groups of judges are in agreement that the key to assessing whether or not a claim is directed to a patent-eligible manner of manufacture is a proper characterisation of the claimed invention.  But here they are hopelessly divided on the correct approach.  We will therefore only have any certainty in cases in which both approaches produce the same outcome, or in which the claim is so closely analogous to that in Aristocrat that we might consider ourselves bound ‘to arrive at a like result’.  Otherwise, we now find ourselves back in the position of waiting, once again, for the courts to clarify the law and, ultimately, for another suitable case to make its way up to the High Court (hopefully constituted by an odd number of judges).

I regard this as a tragic failure of the ‘manner of manufacture’ test in relation to computer-implemented inventions.

The Road to Failure

So how did we get here?  While there is probably not a simple narrative that everybody would agree on, it is perhaps possible to outline a few key steps along the path to where we now find ourselves.

Let’s start at IP Australia, which in about 2008 shifted its practice on computer-implemented inventions, probably with perfectly good intentions of blocking patents on such ludicrous ‘technologies’ as a barely-computerised scheme for commercialising inventions, but which has escalated its approach over subsequent years to the point where ever-more-convoluted arguments are being deployed to reject ever-more-sophisticated applications of computer technology.  With an increasing number of disputes between applicants and the Patent Office over the patent-eligibility of various computer-implemented inventions, it was inevitable that some cases would eventually be taken on appeal.

Which brings us to the Federal Court of Australia.  Since 2014 there have been four cases originating from patent office rejections of computer-implemented inventions that have made their way up to a Full Bench of the Court, and shaped the law: Research Affiliates LLC v Commissioner of Patents [2014] FCAFC 150, Commissioner of Patents v RPL Central Pty Ltd [2015] FCAFC 177, Commissioner of Patents v Rokt Pte Ltd [2020] FCAFC 86, and Commissioner of Patents v Aristocrat Technologies Australia Pty Ltd [2021] FCAFC 202.  It is notable that in three of these four cases (RPL Central, Rokt and Aristocrat) the original Patent Office decision was overturned at first instance, before being reinstated on further appeal to the Full Court.  In Aristocrat the three judges on the Full Federal Court panel also split, with Justice Nicholas concluding that he would have remitted the case back to the primary judge to determine whether Aristocrat’s claims involve any relevant improvements in EGM technology.  In a fifth case, Repipe Pty Ltd v Commissioner of Patents [2021] FCAFC 223, the patentee was denied leave to appeal to the Full Court.

A further case, Encompass Corporation Pty Ltd v InfoTrack Pty Ltd [2019] FCAFC 161, involved two innovation patents that had been examined and certified in 2014, but which the Full Court found to be invalid for lack of patent-eligible subject matter.  Here the Patent Office’s decisions finding Encompass’ claims patentable were challenged by the respondent, InfoTrack, as a defence in infringement proceedings.  In a stark illustration of the shifting state of the law in relation to computer-implemented inventions, the Commissioner of Patents intervened in Encompass, in support of InfoTrack, arguing that the very claims that the Office had previously certified were not, in fact, patent-eligible.  There was, in this, an echo of the RPL Central case, in which claims that had initially been examined and certified in 2009 were subsequently found to be patent-ineligible in 2011 after this ground was raised, sua sponte, by a Patent Office hearing officer in the course of an opposition filed on the sole ground of lack of novelty (which did not succeed).

Throughout these changes in Patent Office practice, and cases that flip-flopped between single judges and the Full Court of the Federal Court, the High Court sat back and declined to weigh in.  RPL Central, Rokt and Repipe all sought, and were denied, special leave to appeal, before the High Court finally allowed Aristocrat’s application for special leave.  Having done so, however, the Court has utterly failed to clarify the law, arguably leaving us worse off than before.

Given that every one of the eligibility rejections that has been taken on appeal to date has ultimately been upheld – although not always for the same reasons given in the original Patent Office decisions – IP Australia might consider itself vindicated in all of this.  But it should not.  As observed above, in each of RPL Central and Rokt one out of four experienced judges disagreed, while in Encompass the Commissioner of Patents put herself in the unusual position of intervening to argue against her own earlier decisions!  Worst of all, in Aristocrat just two Federal Court judges decided that the claims at issue were clearly unpatentable – using reasoning that has now been disapproved by the High Court – while the judge at first instance found the claims clearly patentable, and the third judge in the Full Court would have given Aristocrat a second chance with the primary judge.  With the split decision in the High Court, that means that of 10 judges who have looked at the case, five found against Aristocrat, while four were in favour, and one hedged his bets.

Anybody – including the powers-that-be within IP Australia – who believes that the current state of the law provides any basis for a stable patent system on which applicants and the wider public can rely to produce predictable results that can be used to inform commercial decision-making has got industrial-sized rocks in their head.  It is no wonder that applicants are requesting hearings to challenge examination decisions in historically unprecedented numbers, and with the High Court’s failure in Aristocrat I expect this pattern to continue.

‘Manner of Manufacture’ is Failing Software-Based Innovation

It is clear that judges in Australia’s Federal and High Courts are deeply divided on the correct approach to assessing the eligibility of patent claims directed to computer-implemented inventions.  As I have said, I would go so far as to suggest that the ‘manner of manufacture’ test that the legislature retained in the 1990 Act as a ‘future-proof’ approach to evaluating the patentability of unforeseen technologies has failed us in relation to software-based innovations.  The fundamental problem, as I see it, is that the courts have proved themselves unable to reach a consistent position on whether, and in what circumstances, the realisation of an innovative idea in software is ‘sufficient’ to result in a patent-eligible invention, given that an idea in and of itself has never been considered a patentable ‘manner of manufacture’.

Underlying this problem is that fact that nobody – including the decision-makers at IP Australia who seemingly set us on the current path sometime around 2008 – has ever articulated any clear policy basis for drawing a dividing line between patentable and unpatentable computer-implemented inventions.  With the initial Patent Office decisions issued from 2010, the targets appeared to be traditionally unpatentable ideas and schemes that had been ‘dressed up’ to make routine use of computers for recording and displaying information, or for automating trivial hand computations.  Since then, however, increasingly sophisticated computer-implemented inventions that have, in many cases, been acknowledged by the courts to be novel, inventive, useful and commercially valuable, have nonetheless been rejected as ineligible subject matter.  With applicants clearly investing significant resources in the development and commercialisation of these innovations – as well as in pursuing patent rights through the Patent Office and the Courts – we might reasonably ask: why are these investments being denied the protections that are routinely available in other fields of technology?

At the very least, innovators who are heavily relying upon software to implement their inventions, across all fields of commerce and technology, should be entitled to certainty around what is, and is not, protectable through the patent system.  But when experienced judges are liable to reach different conclusions about exactly the same subject matter, and this continues to occur after the better part of a decade of judicial development, any reasonable degree of certainty is by now a pipe dream.

Conclusion – Time for a Reset?

I believe that it is time to take action on healing the festering sore that the law on patent-eligibility of computer-implemented inventions in Australia has become.  IP Australia has stated that it ‘will carefully review the [High Court’s] decision and consider its implications.’  But for too long now, IP Australia has been keeping its own counsel on this issue, unilaterally developing the examination practice that has brought us to where we find ourselves today – a 3:3 split decision of the High Court that brings us no closer to stability or certainty in the law.  What are we to do now?  Wait a further decade for another suitable case to make its way up and be granted special leave by the High Court?  With no guarantee that, even then, we will get any significantly greater clarity?  I hope not!

As a first step, I would call on IP Australia to consult more widely on the implications of the High Court’s decision in Aristocrat, as it did when the Court handed down its Myriad decision on the patent-eligibility of isolated gene sequences in 2015.  It certainly could not hurt to take input from stakeholders – both ‘for’ and ‘against’ the patenting of computer-implemented inventions – to see whether some principles can be derived from the two High Court opinions on which all interested parties might agree.  At the very least, an examination practice that has been honed through a process of consultation might be more widely accepted, better understood, and less frequently challenged by applicants.

I doubt, however, that this will be enough.  There will, inevitably, continue to be cases in which applicants who disagree with Patent Office decisions will appeal to the Federal Court, which is unlikely to achieve greater clarity in the future than it has to date, especially given the uncertain state of High Court authority on the issue.  Furthermore, it is far from apparent that the current law – whatever it may be – is aligned with any principled policy on digital innovation in Australia.  I therefore suggest that the time is right for a broader review of IP protections for computer-implemented innovations in Australia.  We need to consider what protections we are obliged to provide under international treaties such as TRIPS.  We need to look at the benefits – and costs – of providing patent protection to software-based innovations across all industries including those, such as commerce and finance (e.g. ‘fintech’), that are increasingly important to our economy but which have not traditionally been regarded as ‘patent-eligible’ fields of endeavour.

Above all, I believe we need to take a step back and consider what it is that we want from the patent system when it comes to computer-implemented inventions.  There is no question that companies like Research Affiliates, RPL Central, Encompass, Rokt, Repipe and Aristocrat are investing considerable sums to turn innovative ideas into valuable products and services, while being denied patent rights that would effectively protect those investments against copycat competitors.  This might be good for consumers, if it leads to greater competition and lower prices.  On the other hand, it might be bad if it ultimately results in fewer innovations being developed and successfully commercialised by a more diverse range of companies – because we know that, especially in the software space, there is a small number of multinational tech behemoths that can, if they see an opportunity, out-develop, out-market and out-distribute any smaller business that comes up with a good idea. 

But this balance, between competition and the patent incentive, is exactly the same one that exists in all fields of patentable technology.  We need to ask whether we have the balance right in the case of computer-implemented inventions and, if not, then what do we need to do about it?  It may well be that this will require parliament to step in with a legislative solution because, as I observed above, ‘manner of manufacture’ does not seem to be working for us here.


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.