Patents are available in Australia to protect a wide range of advances in computer technology. However, in light of increased scrutiny being applied to such applications, particularly in regards to the requirements for patentable subject matter, close attention must be paid, at the time of both drafting and prosecution, to emphasise the patentable aspects of the invention.
Up until 2011, the question of whether algorithms are patentable subject matter in Australia was reasonably settled. In a nutshell, the position was that algorithms that direct computers to operate in a way that produces a tangible benefit were considered to be proper subject matter for a patent. On the other hand, algorithms that merely embodied non-tangible phenomenon, such as abstract ideas, financial schemes, mathematical formulae or scientific theories, were not.
However, commencing with a 2011 decision of the Australian Patent Office in the opposition proceeding of Myall v RPL Central1 uncertainty has crept into the area. This continued throughout RPL’s journey through the appeals process, which culminated in a decision of the Full Federal Court of Australia in 20152 ("RPL Central"). The Full Federal Court also delivered an important decision of Research Affiliates LLC v Commissioner of Patents3 ("Research Affiliates") in late 2014 concerning the patentability of computerised business methods.
In the aftermath of RPL Central and Research Affiliates, there have been numerous published decisions of the Australian Patent Office rejecting applications, particularly those relating to electronic gaming technologies, for failing to define patentable subject matter. While one could take a high level view of these events as representing a change in law and practice in Australia when it comes to patenting algorithms, closer analysis reveals this is a major oversimplification.
The fact is that algorithms that were patentable in the pre-2011 legal landscape continue to be so today. This is reflected in the practice of the Australian Patent Office to grant patents covering algorithms in a wide range of software technologies, in some cases in increasing numbers.
However, there is no doubt that applications in this technical field are being more carefully scrutinised for subject matter compliance than was previously the case. In light of this increased attention by the Australian Patent Office, this article outlines the current Patent Office practice and how we arrived at this point, as well as setting out some tips for how applicants can improve their position in Australia.
Patents are available in Australia and are being granted, for developments in computer technology, including novel algorithms that direct computers to operate in a way that produces a tangible benefit. Examples of patentable algorithms include those that arise in the fields of natural language processing, artificial intelligence, encryption and security, machine learning and pattern recognition, software development environments, data analytics, data structures and user interfaces.
Patents are also available, at least in theory, for inventions that lie in the computerisation of non-tangible phenomena, but only to the extent that the computerisation is conducted in a ‘non-conventional’ way. However, in light of the difficulty in defining exactly what is or is not a ‘conventional’ computer implementation, steering such applications through the Australian Patent Office is frequently a much more involved process.
It is plainly apparent that the Australia Patent Office is routinely raising and maintaining subject matter objections to applications involving software and computer technology and sometimes applying the patentability principles in an unpredictable way. Applicants must bear this in mind, both at the time of drafting the specification and during prosecution when a subject matter objection is almost inevitably raised. Successful outcomes are more readily obtained in Australia, if the time is spent correctly drafting the specification to emphasise the patentable aspects of the invention.
The seminal Australian decision involving subject matter eligibility of algorithms is IBM v Commissioner of Patents4 ("IBM"), which was decided in 1991. That decision, which concerned an algorithm for depicting smoother and more accurate curves in computer graphics displays, confirmed that the application of a mathematical formula to computers was potentially patentable in Australia. At the time of the decision there were concerns both in Australia and the United States that granting patents over computer-applied mathematical formulae was tantamount to patenting the non-tangible formula itself. However, the Federal Court of Australia was comfortable that the claims, rather than covering the formula in isolation, were limited to a computer environment. It was also noted that the production of an improved curved image was a commercially useful effect in computer graphics and the method was thus found to constitute patentable subject matter.
Incidentally, the concern with granting patents over mathematical formulae has been recently echoed in the discussion of the ‘pre-emption’ of an abstract concept in the leading United States Supreme Court decisions of Bilski5 and Alice Corporation v CLS Bank International6 ("Alice Corporation").
IBM was a liberating decision in that it confirmed that algorithms, described by the Court as a ‘procedure for solving a given type of mathematical problem’, are potentially patentable even if the underlying mathematics that implement the algorithm are already known. In this regard, the Court likened an algorithm that applies known mathematical techniques to improve the functioning of a computer to the new use of a known chemical substance that takes advantage of one of the properties of the substance that were previously unknown or unsuspected.
Foreign to the Normal Use of Computers
However, one less positive aspect of the decision was a reference to the relevant mathematical techniques as being ‘foreign to the normal use of computers’. Interpreting the word ‘foreign’ literally could have the result of denying patent protection to all algorithms apart from those that utilise numerical and logical operations that were thought could not be performed on (ie are ‘foreign’ to ) a computer processor. As all processors are capable of performing mathematical and logical operations, including the techniques in issue in IBM, we do not believe that the word was used in that sense. Instead, we understand the term ‘foreign’ was used to describe techniques that had, up until that stage, not been performed on computer processors, but not necessarily due to inherent properties of the techniques making them unsuitable for execution by computers.
This question of the normal functioning of a computer and what is a ‘foreign’ or a non-generic use thereof would become relevant, both in the United States in the Alice Corporation decision and in Australia in Research Affiliates and RPL Central.
In the months following the United States Supreme Court decision of Alice Corporation, the Full Federal Court of Australia considered the issue of the patentability of computer-implemented inventions in the two important decisions of RPL Central and Research Affiliates.
Research Affiliates was concerned with a patent application for an invention involving the creation of an index of securities (for application to passive investment strategies) using a particular valuation methodology. In a nutshell, the Court decided that while the technique could well have been an ‘inventive’ idea, it was nevertheless an un-patentable 'abstract idea' or 'mere intellectual information'. The use of the term 'abstract idea' strongly echoed the United States Supreme Court in Alice Corporation, a case also concerned with the patentability of a financial scheme.
From Research Affiliates, the Australian position on the patentability of software is, in essence, that an otherwise un-patentable abstract idea or business scheme does not become so merely by implementing the idea using standard computer technology. Although computer implementation necessarily involves the creation of an artificial physical effect inside the computer, such effect alone is not sufficient to transform the idea into a patentable invention.
Although the Full Court decisively found Research Affiliates' techniques for business analysis and securities selection were not the sort of thing for which patents are granted (even if implemented in software), the Court stressed that many advances in computer technology, including developments in software, do indeed constitute patentable subject matter.
For example, the Court noted that patent protection is available for the category of innovations that are 'inventive uses of a computer that utilise previously unknown abilities of software and hardware'. Also of relevance is the Full Court's approval of a statement from the United States Court of Appeals for the Federal Circuit’s decision Bancorp Services7 , that many of the capabilities of modern computer technology embody 'significant advances that reside firmly in the category of patent-eligible subject matter'.
The Full Court also noted that the question of whether a patent application is directed to patentable subject matter can only be answered by gaining an understanding of what the invention actually is as a matter of substance, and not merely as a matter of form. To gain this understanding, it appears to be permissible to look beyond the claims to the entire patent specification.
The other critical decision, RPL Central was concerned with a method for educational competency recognition. Unlike the application in issue in Research Affiliates, the RPL Central patent application was accepted by the Australian Patent Office, but then successfully opposed on the ground of patentable subject matter. Interestingly, the ground of patentable subject matter was not raised as a ground of opposition by the opponent, but by the Patent Office Delegate.
RPL Central appealed the opposition to the Federal Court, which at first instance, accepted RPL's submission that each step of the competency-recognition method involved a physical effect, due to the fact that the method operated over interconnected computers exchanging information with each other. This transfer of information resulted in a physical manipulation of the data stored on the computer, and, as such, resulted in a physical effect. In addition, the primary judge found that the automatic generation of questions, and the presentation of those questions to the user, created an artificial state of affairs in the user's computer that was itself a physical phenomenon in which the effect of the invention could be observed.
The Australian Patent Office then appealed against the decision to the Full Court2, which took a different view of the method of the invention. Noting the established law that business methods and 'mere schemes' are not patentable per se, the Court focused on the fact that the method did not actually involve any computer processing steps. Instead, the computer merely presented questions to the user through a website and received the answers. Being devoid of technical computer processing, the method constituted a non-patentable business innovation. However, in reaching this conclusion, the Court suggested that potentially patentable inventions might have arisen if the claims specified that the computer was programmed to function as an adviser or artificial intelligence entity actually engaged in the evaluation of the user's input.
At its heart, the RPL invention involved the conversion of personalised information to questions and requests for documentation. The Commissioner of Patents successfully argued that this is simply a trivial programmed conversion of information into question format by prepending relevant words. The court concluded that the computer was in effect operating simply as an intermediary, and was not required to evaluate the user's input to provide an answer.
Although the reframing of criteria into questions may indeed be beyond the generic use of a computer, it was considered that the methodology was no more than an idea, and merely implementing the idea could not give rise to an invention. Indeed, the decision pointed to the fact that no programming instructions were provided in the specification.
The current Australian Patent Examiner’s Manual describes the practice applying to computer-related inventions (including those for novel algorithms) as not involving any ‘specific exclusions for software or methods that are implemented as computer software or a related product’.
However, consistently with the decisions of Research Affiliates and RPL Central (and indeed the position before then), the Manual goes on to provide that computer-related inventions are only patentable if what is claimed ‘as a matter of substance’ meets the requirements for a manner of manufacture and in particular is not a mere scheme, abstract idea or mere information.
The principles that are to be applied when assessing computer implemented subject matter are:
Reviewing decisions of the Australian Patent Office (made in the context of examiner objections to applications and contested oppositions) that cite either Research Affiliates or RPL Central) could give reason to be sceptical about the availability of patents for algorithms in Australia. More specifically, as shown in the following table of 25 cases for computer-related subject matter decided between 1 January 2016 and June 2018, an applicant was successful on only four occasions (British Broadcasting Corporation8, Aristocrat Technologies9, Cody Live Pty Ltd 10 and Bio-Rad Laboratories, Inc11) in definitely establishing that the application was directed to patentable subject matter.
On four occasions, Applicants were provided an opportunity to propose amendments to address the negative findings reached in the decision. Two of these applicants successfully achieved acceptance (Bio-Rad Laboratories, Inc12 and R F Products Inc13) after proposing suitable amendments.
The application was refused for not disclosing patentable subject matter in the 18 other cases reviewed.
|British Broadcasting Corporation
||Downloadable audio/visual content
||Opposition unsuccessful. Application to proceeded to grant.
|Aqua Index Ltd
|Cody Live Pty Ltd
||Computer-enhanced real estate sign
||Opposition unsuccessful. Application proceeded to grant
|Konami Gaming, Inc.
||Gaming machine with touch sensitive electronic display displaying a plurality of separate image elements
||Touch Screen Gaming Machine
||Claims not directed to patentable subject matter. Opportunity to amend provided. Innovation patent not subsequently pursed by applicant.
|Accenture Global Services Limited
||Customer database analysis
|Swiss Reinsurance Company Limited
||Computer-based method for establishing a safety assessment of a process industry plant comprising multiple organizational units
|Trading Technologies International, Inc
||Spread Trading Tool
|Bally Technologies ANZ Pty Ltd
||Electronic Gaming Machine
||Application originally accepted but later refused during re-examination.
||Lottery Gaming Machine
|R F Products, Inc
||RF distribution system
||Claims not directed to patentable subject matter. Opportunity to amend provided. Application proceeded to grant after amendment
||System for tracking usage of athletic equipment of an athlete
||Innovation patent revoked
|Rokt Pte Ltd
||Digital advertising system and method
||Application refused. Appeal to the Federal Court of Australia pending.
||Advertisement Price Discounting
|Bio-Rad Laboratories, Inc
||System and method for determining an optimum QC strategy for immediate release results
|Sportsbet Pty Ltd
|Patent Investment & Licensing Company
||Electronic gaming machine
|Eris Innovations, LLC
|Discovery Life Limited
||Computer implemented method for of managing a life insurance policy with a related medical scheme
|Bio-Rad Laboratories, Inc.
||System and method for producing statistically valid assay means and ranges for quality control materials
||Claims not directed to patentable subject matter. Opportunity to amend provided. Application eventually accepted after amendments.
||Online state lottery game
||Patent revoked after re-examination
|BGC Partners, Inc.
||Method for facilitating electronic trading
|BGC Partners, Inc.
||System and Method for Providing an Operator Interface for Displaying Market Data, Trader Options, and Trader Input
However, at the same time, a deeper dive makes clear that the above Patent Office decisions do not paint a complete picture of how computer-implemented inventions are currently treated in Australia, and particularly those involving algorithms. To present a fuller picture, it is necessary to look at the grant statistics in relevant subject matter areas.
In this regard, a representative IPC class is G06F 17/30, which although is defined as covering ‘database structures for information retrieval’, includes applications from a wide range of software and algorithm technologies such as search queries, natural language processing, hashing functions and content synchronization.
The wide-ranging subject matter covered by the class is also reflected in the owners of the Australian patents, which include Apple, eBay , Dropbox, Google, Accenture Global Solutions, Facebook, LexisNexis, Uber, Microsoft, Amazon and Fuji Xerox.
The following is a graph of the numbers of standard Australian patents classified in G06F 17/30 granted over the last 8 years, with the grant number for 2018 being a year-to-date figure as of October 2018.
Despite the possibility of 2018 finishing with a lower grant number than the previous few years, the longer-term trend gives reason to be cautiously optimistic about obtaining protection for software-related inventions in Australia.
In this regard, while there was a noticeable dip in patents granted during 2013, the trend was upwards for the succeeding four years. Interestingly, the dip in grants for 2013 cannot be attributed to the Full Federal Court’s decisions in Research Affiliates and RPL Central, as those decisions were not respectively delivered until 2014 and 2015.
It is also significant that the number of patents granted (at least in the G06F 17/30 classification) did not decrease in the years that the two decisions were delivered, nor in the following two years.
A further positive insight that can be gleaned from the 2017 grant number is the fact that it occurred against a backdrop of the negative decisions issued by the Australian Patent Office during that same year. This suggests that the examination section of the Australian Patent Office is continuing to issue patents covering core software technologies, independently of the activities of the Hearing division.
Finally, in regard to the lower year-to date grant number for 2018, it should be noted that it is in comparison to a six-year high grant number from 2017, and that the grant number has already surpassed the 2013 number.
As observed above, the critical insight to be drawn from the current case law and examination practice is that computer-related inventions are examined for ‘substance’. This necessarily involves a consideration of the entire patent specification and knowing that the entire patent specification will be scrutinised means that it is essential to include extensive technical disclosure for software inventions.
It is also strongly advisable to expressly describe the technical advantages that the invention provides and exactly how the claim features realise those advantages. In the case of algorithms, it is often the case that the invention resides in devising a solution that makes computer systems run ‘smarter’, rather than merely faster. In other words, novel algorithms do more than merely take advantage of increased processor speeds and storage resources. If this is the case, any such technical advantages, and the underlying logic that realises the advantages, should be expressly and extensively detailed.
On the other hand, if the invention is more about computerising a business method or abstract idea (particularly something that could be implemented manually), it is critical that the entirety of the implementation detail is described, including a discussion of the algorithms and software modules involved in the computerisation. Finally, the claims of such an application must include the essential ‘non-generic’ components involved in the computerisation.
1 Myall Australia v RPL Central Pty Ltd  APO 48
2 Commissioner of Patents v RPL Central Pty Ltd  FCAFC 177
3 Research Affiliates LLC v Commissioner of Patents  FCAFC 150
4 IBM v Commissioner of Patents  FCA 625
5 Bilski v Kappos 561 U.S 593 (2010)
6 Alice Corporation v CLS Bank International 134 S.Ct. 2347 (2014)
7 Bancorp Services LLC v Sun Life Assurance Co of Canada (US) 687 F. 3d 1266 (2012)
8 FOXTEL Management Pty Ltd v British Broadcasting Corporation  APO 19
9 Aristocrat Technologies Australia Pty Limited  APO 49
10 ABC Photosigns Pty Ltd v Cody Live Pty Limited  APO 58
11 Bio-Rad Laboratories, Inc.  APO 24
12 Bio-Rad Laboratories, Inc Bio-Rad Laboratories, Inc.  APO 38
13 Hagenuk Marinekommunikation GmbH v R F Products, Inc.  APO 58