Will an Expanded Full Bench of the Federal Court Bring Sense to Australian Law on Computer-Implemented Inventions?

The law relating to patent-eligibility of computer-implemented inventions – and its practical application by the Australian Patent Office – is currently a mess.  As I have noted recently, over the past few years disproportionate resources have been devoted by both applicants and the Patent Office to disagreements over whether or not particular inventions are patentable according to the ‘manner of manufacture’ test required by the Australian law.  In just the past three years, there have been over 10 times as many hearings held, and decisions issued (42 in total), in relation to this particular aspect of the law than in the entire first decade following commencement of the current Patents Act 1990. 

Anybody who believes that this is not indicative of a serious problem is fooling themselves.  The Institute of Patent and Trade Mark Attorneys of Australia (IPTA) flagged its frustration with the current situation in a submission [PDF, 123kB] in response to the recent consultation on the exposure draft of the Intellectual Property Laws Amendment Bill (Productivity Commission Response Part 2 and Other Measures) Bill 2018, noting that:

IPTA is particularly concerned by the current approach being adopted by IP Australia, and lower courts, in relation to the assessment of patentability of inventions based on software. Many inventions which satisfy novelty and inventive step requirements are being rejected outright by IP Australia, even when such inventions are considered patentable in Europe. IPTA is so concerned about this current trend within IP Australia that it has recently set up a working group to study this area to see what can be done to head off this disturbing trend.

While this is a valuable initiative, there is a real prospect that the issue will be resolved in the courts over the coming months.  In the past year or so there have been four appeals filed with the Federal Court of Australia against Patent Office decisions refusing applications for computer-implemented inventions: Todd Martin v The Commissioner of Patents, QUD374/2017; Rokt Pte Limited v The Commissioner of Patents, NSD1292/2017; Repipe Pty Ltd v The Commissioner of Patents, WAD323/2018; and Aristocrat Technologies Australia Pty Ltd v The Commissioner of Patents, NSD1343/2018.  Of these, the Todd Martin case was discontinued following an (unsurprisingly) unsuccessful mediation.  The Rokt case was heard on 18-20 July 2018, and is currently awaiting judgment.  The Repipe and Aristocrat cases are currently in preliminary stages, and are yet to be set down for a hearing.

However, the case that will trump them all is one that does not involve the Commissioner of Patents.  Back in June I wrote about the decision of a single judge of the Federal Court of Australia in Encompass Corporation Pty Ltd v InfoTrack Pty Ltd [2018] FCA 421.  In that case, a pair of innovation patents owned by Encompass survived attacks based on alleged lack of novelty and innovative step, only to be found invalid on the basis that they did not relate to patent-eligible ‘manners of manufacture’.  In reaching this conclusion, the judge appeared to be strongly influenced by the state of the prior art, as represented by the evidence upon which InfoTrack relied for its novelty and innovative step cases.

Encompass has appealed that decision (Encompass Corporation Pty Ltd v InfoTrack Pty Ltd, NSD734/2018) to a Full Bench of the Federal Court of Australia.  Notably, the Court has empanelled an expanded bench of five judges – Chief Justice Allsop, and Justices Kenny, Besanko, Nicholas, and Yates.  This is a rare event, that usually only occurs when a case is of particular importance, and/or where there is some prospect that the Court may be called upon to overrule an earlier decision of a three-judge panel.  In this case, however, there is reason to suppose that the prospects of the Full Court overruling either of its two most recent decisions on patent-eligibility of computer-implemented inventions, Research Affiliates LLC v Commissioner of Patents [2014] FCAFC 150 and Commissioner of Patents v RPL Central Pty Ltd [2015] FCAFC 177, are fairly slim, given that the expanded panel includes two of the three judges – Justices Kenny and Nicholas – responsible for those decisions.  The third judge, Justice Annabelle Bennett (as she then was), retired from the Federal Court in 2016.

The Encompass appeal is scheduled to be heard on 8-9 November 2018, and a decision is therefore likely in the first half of 2019.  A five-judge ruling can override not only all of the ongoing appeals currently before single judges of the court, but also any or all of the existing first instance and Full Court decisions on patent-eligibility of computer-implemented inventions.  In my view, the Patent Office has good reason to be concerned that the Encompass decision, when it comes, will invalidate the approach that it is currently taking to the examination of computer-implemented technologies. 

The Rise and Rise of Hearings on Patent-Eligibility

In 2018 to date, 17 hearing decisions have been issued by the Patent Office resulting from examination objections that applications for computer-implemented inventions were not directed to patent-eligible subject matter (including four in September that post-date the data I recently published).  With three months still to go in the year, this now exceeds the 15 subject-matter-related hearing decisions issued in 2017, and 10 such decisions issued in 2016.

As I have said before, the reason for the unprecedented numbers of hearing requests in recent years in respect of subject-matter objections is not because applicants are seeking protection for unpatentable inventions in greater numbers, nor is it because patent examiners are not doing their best to follow the law or the guidance set out in the Patent Examiners Manual.  The fundamental problem is uncertainty in the application of the law, and inconsistency in Patent Office decision-making. 

It is not just that patent applicants, and their attorneys, do not agree with the Patent Office about what is, and is not, patentable – although this is certainly true, and a source of increasing frustration.  The guidance is so open to interpretation that the Patent Office is not even able to agree with itself!  Inventions found to be patent-eligible in examination have been rejected on the same ground in opposition (see, e.g. Myall Australia Pty Ltd v RPL Central Pty Ltd [2011] APO 48, Sportsbet Pty Ltd v Diogenes Limited [2017] APO 60).  Attorneys, examiners, hearing officers and judges seem to be finding it all but impossible to extract coherent principles from the relevant judicial decisions, or to reach any consistent position regarding patent-eligibility of computer-implemented technology.

The Dubious Role of ‘Prior Art’

Perhaps most disturbing of all is a recent trend to bring prior art into the assessment of patent-eligibility.  In McCormick & Company, Incorporated [2017] APO 62, the Patent Office Hearing Officer listed a range of considerations for patent-eligibility of computer-implemented inventions, including whether the invention can be ‘broadly described as an improvement in computer technology’, whether there is ‘ingenuity in the way in which the computer is utilised’, and/or whether the invention involves ‘steps that are foreign to the normal use of computers’, and concluded that ‘[i]t is difficult, if not impossible, to see how these considerations can be addressed without an understanding of the art, including relevant prior art.’  He went onto find further justification for this approach in the High Court’s decision in D'Arcy v Myriad Genetics Inc [2015] HCA 35 – a case which, notably, related to the patentability of isolated gene sequences, i.e. something found in nature, rather than being a product, like computer technology, of prior human ingenuity.

As a general proposition, however, the higher courts have been very clear that it is not appropriate to bring prior art into the consideration of whether or not particular subject matter is eligible for patenting.  I have written about this previously, calling it the ‘ball-point pen principle’, after the observation by a unanimous Full Bench of the Federal Court of Australia in CCOM Pty Ltd  v Jiejing Pty Ltd [1994] FCA 1168 that ‘manner of manufacture, novelty, inventiveness and utility are stated as distinct requirements of a patentable invention.  Thus, whilst a claim for the ball point pen now would fail for anticipation and obviousness, it would still be a claim for a manner of manufacture.’ 

This really could not be any more clear, and it is absolutely binding on the Patent Office.  And, contrary to the approach that the Patent Office seems to have taken in some recent decisions, CCOM is not ‘overruled’ by Myriad, which does not in any way rely upon human-created prior art as being relevant to patent-eligibility in its ratio decidendi.  Furthermore, in other decisions (e.g. Lockwood Security v Doric Products [2004] HCA 58 at [73], Advanced Building Systems Pty Limited v Ramset Fasteners (Aust) Pty Limited [1998] HCA 19 at [40], and NV Philips Gloeilampenfabrieken v Mirabella International Pty Ltd [1995] HCA 15 at [9]) the High Court has itself been very clear that the grounds of novelty, inventive step and manner of manufacture are non-overlapping and conceptually distinct, and that patent-eligibility is to be assessed ‘on the face of the specification’.

Anomalous Outcomes

Bringing prior art into the consideration of patent-eligibility also leads to anomalous results.  According to the ‘ball-point pen principle’, whether or not something is suitable subject matter for a patent is timeless – ball-point pens have always been, and forever will be, patent-eligible ‘manners of manufacture’, even after they ceased to be novel and inventive.  However, the ‘prior art’ approach breaks this basic principle, since it implies that as technology advances, and matters that were once ‘foreign to the normal use of computers’ become widespread, subject matter that was once patent-eligible can become ineligible (as opposed to simply lacking in novelty or becoming obvious).

This is not mere speculation on my part.  There is already at least one example among the Patent Office decisions of such an anomalous outcome.  The decision in Aristocrat Technologies Australia Pty Limited [2016] APO 49 related to an improvement in a multi-game electronic gaming machine involving the presentation of game information in a way that allowed a particular game and bet denomination to be selected in a single action.  The Hearing Officer found (at [48]) that the invention required ‘only generic computer implementation’, but that it did ‘not appear that it was normal (at the priority date) to configure a gaming machine interface in this particular way.’  The priority date in this case was 6 August 2003.

In contrast, the Hearing Officer in Aristocrat Technologies Australia Pty Limited [2017] APO 1 arrived at a different conclusion in relation to a broadly analogous invention involving a touch-screen gaming machine providing multiple playable games with several bet denominations associated with each game, and having a specific arrangement of playable games of a particular type and the corresponding associated sets of bet denominations.  The priority date in that case was 9 January 2007 (i.e. three and a half years after the earlier patent), and the Hearing Officer found the invention to be ineligible on the basis (at [74]) that ‘[t]he computer implementation is generic without any ingenuity or steps foreign to the normal use of the computer or the player interface.’

Despite the lengthy reasoning in each of these two decisions, looking purely at the claims in each case it is very difficult to see why there should have been any difference in the outcomes, other than the passage of time between the respective priority dates.

Encompass at First Instance

This brings us to Encompass, a decision in which a single judge of the Federal Court allowed the content of the prior art to influence his decision on patent-eligibility.  In determining that Encompass’ claims were not directed to a patentable ‘manner of manufacture’, the court found, at [195]-[196], that:

In this case, the method disclosed in the Patents (and the apparatus) result in the computer being used to do something it has not been used to do before. But it is not clear to me that in doing so they have improved the functionality of the machine. This is because the method (and apparatus) merely involve a concatenation of three other methods, none of which is new: the use of a network representation, the querying of remote data sources and the use of a purchasing step.

It is true that I have accepted that the addition of the purchasing step has the consequence of providing an innovative step. But it does not follow that the method and apparatus involve any improvement in the computer. None of the three elements of the claims is, in itself, an improvement in the functionality of a computer. I would accept the combination of the three elements in the innovative way disclosed in the Patents certainly provides an enhanced experience for the user of the computer. But it is not self-evident that there is a necessary connexion between an enhanced user experience and an improvement in the computer. In this case, so it seems to me, the enhanced user experience results from the combination of well-known computing mechanisms. But unless that combination can itself be said to have brought about an improvement in the computer this will be beside the point.

So, to summarise, the computer is doing something that computers have not done before.  But that thing is ‘just’ a combination of things that computers have been known to do before.  And, despite the combination being ‘innovative’ and providing ‘an enhanced experience’, this does not result in an ‘improvement in the computer’.  So the invention is not patentable.

Predictions for the Encompass Appeal

I anticipate that the court’s reasoning will be overturned on appeal.  (This is not the same thing as saying that the outcome will be reversed, however I think that is also quite likely.)  There are fundamentally two problems with the approach that has been taken here.  Firstly, the decision implicitly relies upon the state of the prior art in order to assess subject-matter eligibility.  It is implied in the above-quoted passage that, had any one or more of ‘the use of a network representation, the querying of remote data sources and the use of a purchasing step’ been new, then a different conclusion might have been reached.  As I have explained above, such a prior-art-based approach to ‘manner of manufacture’ is wrong in law, according to High Court and Full Federal Court authority which was binding upon the single judge in this case.

Secondly, the court’s interpretation of the ‘requirement’ for a patent-eligible invention to bring about an ‘improvement in the computer’ sets the bar too high.  The reference to an ‘improvement in the computer’ comes originally from the decision of the Full Court in Research Affiliates LLC v Commissioner of Patents [2014] FCAFC 150, where it was in turn drawn from the opinion of US Supreme Court Justice Thomas in Alice Corporation Pty Ltd v CLS Bank International 134 S.Ct. 2347 (2014).  I am fairly sure, however, that the Full Court did not intend this formulation as a barrier to patentability for computer-implemented inventions that actually provide new and useful functionality, merely because some or all of the ‘component parts’ comprising the invention are known. 

It has long been established that a combination of known elements having a new working interrelationship may be patented (see, e.g., Advanced Building Systems Pty Limited v Ramset Fasteners (Aust) Pty Limited [1998] HCA 19 at [12]).  It seems most unlikely that the Full Federal Court in Research Affiliates intended to create a special exception to this established position in the case of computer-implemented inventions.  My reading of Research Affiliates (and RPL Central) is that the Court was merely drawing a distinction between the straightforward implementation of an otherwise unpatentable method on a computer, where the ‘improvement’ is nothing more than the speed or convenience for which computers are well-known, on the one hand, and methods in which the computer-implementation is actually integral to the invention, on the other

In any event, if the Full Court was imprudent in its citation of Alice, then that is something that the enlarged five-judge panel will be in a position to correct.  Indeed, Encompass made submissions to this effect at first instance, although a single judge is no position to ignore what must be regarded as binding precedent of a superior court.  Doubtless Encompass has advertised its intention to make similar submissions on appeal, which would have informed the decision to appoint an expanded panel.

Conclusion – Why IP Australia Should Be Concerned

If my predictions in relation to the Encompass appeal turn out to be correct, the Patent Office will have a problem on its hands.  Since at least the McCormick decision in December 2017, it has been increasingly relying on prior-art-based reasoning to reject computer-implemented inventions as patent-ineligible.  Both the Repipe and Aristocrat decisions, already on appeal to the Federal Court, expressly refer to prior art in their assessment of the ‘contributions’ made by the respective inventions.  By the time the Full Court issues a decision in the Encompass case, both of these appeals will have advanced, at substantial expense to the applicants and IP Australia.  A number of other subsequent decisions in which patents have been refused have also relied on prior-art-based reasoning, including at least two (Google LLC [2018] APO 13 and BGC Partners, Inc. [2018] APO 27) in which McCormick has been expressly cited.

This does not account for the many other applications that have doubtless been rejected in examination on the basis of a similar approach, where the applicant has lacked the will, or financial means, to proceed to a hearing.

All of these represent cases in which applicants’ prospective patent rights may have been lost or compromised as a result of Patent Office practices that may turn out to be wrong in law.

I have made the point about the value of consultation before, but I shall make it again anyway.  It was notable that when the High Court issued its decision on ‘gene patents’, in D’Arcy v Myriad, overturning two decades of established Patent Office practice that treated claims to isolated genetic sequences as patent-eligible, IP Australia suspended examination of affected applications while it prepared, published, and consulted on proposed new examination guidelines in light of the High Court’s ruling.  As a result, it had the benefit of stakeholder input to clarify its interpretation and implementation of the decision, while applicants were able, through the transparency of the process, to understand what would, and would not, be patent-eligible going forward.

However, there has never been any similar consultation process in relation to computer-implemented inventions.  There is evidence in Patent Office data that a change in practice, more hostile to computer-implemented inventions, occurred in late 2008.  If so, then this change took place without consultation, any form of official notice, or any changes to the Patent Examiners Manual.  It resulted in a series of Patent Office decisions, beginning with Invention Pathways Pty Ltd [2010] APO 10, which eventually led to the Federal Court and Full Court appeals in Research Affiliates and RPL Central.  Even then, in contrast to the gene patents example, there was no associated process of consultation – not even when the High Court declined to grant Special Leave for RPL Central to appeal on the basis that the Full Court’s decision was ‘plainly correct’.

Whatever the policy is behind the current practice in relation to computer-implemented inventions – which has been, unquestionably, driven from within government – it has never been clearly articulated.  Everyone knows that something changed, but nobody knows why.  The Patent Office is continuing to develop new approaches, building on Myriad, RPL Central, and Research Affiliates to justify reference to prior art, and a ‘technical contribution’ approach based upon the four-step UK test developed in Aerotel/Macrossan [2006] EWCA Civ 1371.  The requirements for patent-eligibility in this area remain a moving target, with applicants and their patent attorneys becoming increasing frustrated with their inability to predict what will, and will not, be considered patentable.

Yet, despite the obvious problem, and the ongoing expenditure of applicant and Patent Office resources on disputed cases, there is still no consultation.

Perhaps IP Australia believes that a consultation process is only appropriate following a High Court decision.  Or perhaps only when the courts overturn, rather than support, its decisions.  If so, then it is wrong.  Maybe a decision of five judges of the Federal Court of Australia will finally change its mind!

Tags: Appeal, Australia, Computer programs, Patent Office, Patentable subject matter