Patent Manual of Practice and Procedure after Aristocrat: new words, same song

IP Australia's rewritten examination guidance for software patents repeats the headline propositions of the Aristocrat decisions, then rebuilds the tests Aristocrat rejected. This does examiners and applicants no favours. Applicants should expect objections based upon the Manual's posture but answer them from the law. While examination should be to the standard of the prevailing law, the Manual plays a pivotal role informing examination practice, and Patent Office hearing decisions as well.

David Perkins

Founder & Principal

Examination of Australian patent applications at IP Australia is guided by the Patent Manual of Practice and Procedure (PMPP). The PMPP covers amongst many topics the patent eligibility of computer-implemented inventions, specifically at section 5.6.8.6 of the PMPP. This section has been successively updated since Aristocrat, and the latest update for the foreseeable future was published 3 July 2026. The PMPP states the headline propositions of the Aristocrat decisions and then rebuilds the tests based upon earlier decisions Aristocrat displaced. Applicants with software and other computer-implemented claims can expect similar patent eligibility objections as before, and should answer them from the case law.

How we got here

The Full Federal Court refused certain software patent claims for more than a decade based upon reasoning that has now been set aside by Aristocrat. The line of decisions began with Grant in 2006. The next major decision was Research Affiliates. This decision drew on the US Supreme Court's decision in Alice Corporation Pty Ltd v CLS Bank International 573 US 208 (2014) for the idea that the computer itself must be improved (refer Encompass at [40], [104]–[106]). RPL Central hardened that idea in 2015 into a rule that the invention must lie in the computerisation. The implication is that software would really only be patentable if there was a demonstrable hardware innovation accompanying the software innovation. Generic hardware would not suffice. This represented a quite restrictive approach to patent eligibility for software that many believed unreasonably carved out software innovation from the patent system. The question the courts were grappling with was how do you distinguish an abstract idea from its expression as a patentable invention.

Table 1·The line of decisions, 2006 to 2026

Year	Decision	Result
2006	Grant v Commissioner of Patents [2006] FCAFC 120 Mr Grant appealed after losing at first instance ([2005] FCA 1100)	Ineligible. An asset-protection scheme with no computer involved
2014	Research Affiliates LLC v Commissioner of Patents [2014] FCAFC 150 The applicant appealed after losing at first instance ([2013] FCA 71)	Ineligible
2015	Commissioner of Patents v RPL Central Pty Ltd [2015] FCAFC 177 The Commissioner appealed after the applicant won at first instance ([2013] FCA 871)	Ineligible. High Court special leave refused ([2016] HCASL 84)
2019	Encompass Corporation Pty Ltd v InfoTrack Pty Ltd [2019] FCAFC 161 The patentee appealed after its patents were revoked in an infringement action ([2018] FCA 421). Five judges sat, and the Commissioner intervened to defend Research Affiliates and RPL Central	Ineligible
2020	Commissioner of Patents v Rokt Pte Ltd [2020] FCAFC 86 The Commissioner appealed after the applicant won at first instance ([2018] FCA 1988)	Ineligible. High Court special leave refused ([2020] HCASL 202)
2021	Repipe Pty Ltd v Commissioner of Patents [2021] FCAFC 223 The patentee sought leave to appeal after losing at first instance ([2019] FCA 1956)	Leave refused. Ineligible
2021	Commissioner of Patents v Aristocrat Technologies Australia Pty Ltd [2021] FCAFC 202 The Commissioner appealed after the applicant won at first instance ([2020] FCA 778)	Ineligible, applying an "advance in computer technology" test
2022	Aristocrat Technologies Australia Pty Ltd v Commissioner of Patents [2022] HCA 29 Aristocrat appealed by special leave	Split 3–3, so the 2021 decision stood. All six Justices criticised its test
2024	Aristocrat Technologies Australia Pty Limited v Commissioner of Patents (No 3) [2024] FCA 212 Remitted to the first-instance judge for the remaining claims	Ineligible. The judge was bound to apply the 2021 decision
2025	Aristocrat Technologies Australia Pty Ltd v Commissioner of Patents [2025] FCAFC 131 Aristocrat appealed	Eligible. The court adopted the allowing reasons
2026	Commissioner of Patents v Aristocrat Technologies Australia Pty Ltd [2026] HCADisp 15 The Commissioner sought special leave	Refused, with costs

The Commissioner was a party to every one of these cases except Encompass, a private infringement action, and appeared in that one too (Encompass at [65], [70]–[71]). Three times a single judge found for the applicant, in RPL Central, Rokt and Aristocrat. And three times the Commissioner appealed and won. In two of those cases the applicant then sought special leave from the High Court and was refused. The table begins with Grant in 2006, the business-method case on which the software decisions built. So there has been a history of 20 years in the making, and the law which developed around these cases is now displaced. Yet IP Australia's examination guidelines have stayed reverted to displaced reasoning.

The High Court's refusal of special leave in February 2026 settled the question. IP Australia has since rewritten section 5.6.8.6 across two substantive revisions: on 16 March 2026, and again on 3 July 2026 following public consultation. It received 11 submissions. The published ones come from Spruson & Ferguson, FPA Patent Attorneys, AIPPI Australia, FICPI Australia and the Computer Implemented Inventions Task Force, along with this firms. The existing published guidelines from 3 July 2026 are not obviously informed by any of the published consultation submissions.

The High Court described the Full Court's decision as an application of "established principles". IP Australia's response to the consultation does not use that phrase. It says instead that "computer implemented inventions require unique consideration", and it declined to change the Manual on that point. The allowing reasons, which the Full Court adopted, say the opposite (at [152]–[153]).

What the courts decided

Aristocrat settles the law on patent eligibility for computer-implemented inventions. Aristocrat refers here to three of the interconnected decisions tabulated above: Aristocrat 2022, Aristocrat 2025, and Aristocrat 2026. In August 2022 the High Court split three to three in Aristocrat. In September 2025 the Full Federal Court adopted the reasons of Gordon, Edelman and Steward JJ, the "allowing reasons" (at [125]).

In February 2026 seven Justices refused the Commissioner special leave in Aristcrat 2026. This decision is extraordinarily brief, but significant: the High Court Justices endorsed the Full Court as having "applied established principles concerning the assessment of manner of manufacture" and reaching "a unanimous and clear conclusion as to characterisation".

Read together, the Aristocrat decisions stand for six propositions:

One question. "There is only one question": is there a manner of manufacture? (allowing reasons at [151]).
Characterisation first. The claim is characterised as a whole, inventive and non-inventive integers together (Full Court at [126]–[127]).
No improvement needed. The result "may be an improvement in computer technology, but it need not be" (allowing reasons at [122]).
No inventive implementation needed. "The ingenuity may lie only in the idea" (allowing reasons at [123]).
Conventional hardware can suffice. It is "too rigid and narrow" to say that conventional computer technology, used for its well-known functions, cannot produce a manner of manufacture (Full Court at [131]).
No separate test for computers. Eligibility is not to be broken into a distinct enquiry about computer implementation (allowing reasons at [152]–[153]; Full Court at [81]).

What the Manual gets right

The updated section repeats several of these propositions in its own words:

A claim "is not unpatentable merely because it is possible to characterise the invention as involving the implementation of an idea in a computer using conventional computer technology for its well-known and well-understood functions".
"No integers should be disregarded."
Examiners "should not place undue weight on the inventive aspects of a claim over non-inventive aspects".
None of the listed factors is "a requirement that must be satisfied for there to be patentable subject matter".

All these statements are consistent with the current law of Aristocrat. Examination practice would have to change if the section took these premise to their conclusion. The Manual does not however stop there.

Where the Manual leans back

Instead of reflecting the "one question" nature of the enquiry on patent eligibility, the PMPP sets forth a "decision-making framework" built from the reasoning in decisions Aristocrat explicity set aside. Five points of divergence matter most.

Table 2·The law against section 5.6.8.6

The law after Aristocrat	PMPP Section 5.6.8.6
Two-limb formulation catgeories the earlier displaced authorities (allowing reasons at [122] and footnote 190; Full Court at [131]). The Full Court used them once, to describe a result (at [141]).	Two-limb formulation is how every claim "needs to be assessed" to determine which limb applies, and is promoted as an essential test.
Common general knowledge helps construe the claim (Full Court at [126]). The gaming machine's hardware was conventional and the claims were still eligible (allowing reasons at [145]–[147]).	Asks first whether the hardware is a "new (not CGK) physical product or system". If not, the claim goes to further factors.
No improvement in computer technology is needed (allowing reasons at [122]–[123]).	Asks whether the product works "differently or better as a product generally", and not merely by "executing a new specific method".
Foreign authorities are "at best, of limited assistance and, at worst, dangerously misleading" (allowing reasons at [125]–[127]). The Full Court decided the case without reference to technical effect.	Foreign authoriies imported by referencing need for "relevant technical effect" outside the computer, or a "technical problem" solved. No authorities cited for either.
Dismisses reasoning of displaced decisions but indicates that the results are "plainly correct" without further explanation (allowing reasons at [122]; Full Court at [130]).	Lists "generic or general-purpose computers" as a reason for refusal, and reverts to reasoning from displaced decisions, indicating the computer "must be more than an intermediary (see RPL Central)".

Two of these have an identifiable source, and neither is the law. The hardware question follows the three Justices who would have dismissed Aristocrat's appeal (at [73] and [77]–[78] of their reasons). The technical questions follow Nicholas J's 2021 minority decision (at [117]). The Full Court adopted neither.

The framework's shape is also familiar. It asks whether the claim runs on a known computer and, if so, whether the computer is improved. That is close to the two-step approach that all six High Court Justices criticised in 2022 (Full Court at [50], [116]).

Two-limb formulation

This is the main error in the PMPP. The two-limb formulation categories the displaced authorities in distinct categories: (i) an abstract idea manipulated on a computer (unpatentable); and {ii) an abstract idea implemented on a computer to produce an artificial state of affairs and a useful result (patentable). The catgories are somewhat self-defining as there is no inherent difference between manipulated vs implemented.

The problem is PMPP promotes the two-limb formulation to an exact formula for deciding upon patentability. A more careful reading of Aristocrat indicates that it was never framed as a test. Rather, the two-limb formulation is offered as a framing that reinterprets the conclusions of the displaced authorities. The two-limb formulation is "a better way of expressing the point in such cases" (allowing reasons at [122]; Full Court at [131]). The cases were the earlier ones whose results were "plainly correct" and whose reasoning was not. The formulation replaced the phrase "improvement in computer technology", which the Full Court traced through Rokt back to RPL Central at [96] (at [130]).

The only support the Full Court gave it is that it fits the results of those decided cases (at [132]–[133]).

When the Full Court decided Aristocrat's own claims, it did so by characterisation and the language of NRDC. It would have found claim 1 eligible "for the reasons given in the allowing reasons" (at [135]). It held the remaining claims eligible because each was an altered gaming machine producing "an artificial state of affairs and a useful result" (at [136]). Only after that did it add that the claims were "as described in alternative (ii)", and it explained the label in NRDC's own words (at [141]). Its stated grounds were analogy with claim 1 and characterisation of the claims (at [142]).

That matters because NRDC rejected attempts to place on manner of manufacture "the fetters of an exact verbal formula" (Full Court at [128]). Read as a test, the formulation adds nothing to NRDC. "Manipulated" and "implemented" name the two possible results without saying how to reach either.

Why examination will not change much

An examiner who follows the PMPP section 5.6.8.6 step-by-step will inevitably reject many cases that are allowable under Aristocrat. The saving grace is that it is likely that same cases may be allowed despite examination guidelines by virtue of appeal to the law of Aristocrat. Almost every software claim runs on standard hardware, and the section sends each of those claims to its factors. The factors ask the questions of the past decade: is the computer improved, is there a technical effect, is a technical problem solved.

The worked cases reinforce the habit. Each of the six refused examples ends with the same line: "There were no factors supporting the existence of patentable subject matter". Factors described as optional become decisive once they are the only route left.

The first Patent Office decision to apply the new approach shows the pattern. In CQMS Pty Ltd v Joy Global Surface Mining Inc [2026] APO 19, decided on 18 August 2026, the delegate worked through the Manual's sequence (at [294]). A claim using a generic processor was refused. A near-identical claim was allowed because its wear sensor was not shown to be common general knowledge (at [319], [322]).

That is an outcome the Manual explains and Aristocrat does not. The gaming machine in Aristocrat had no new hardware at all.

What applicants should do

Expect the objection the Manual produces, and answer it from the judgments.

Plan for the objection. Reports will say the hardware is standard and that no factor is met. Budget for a substantive response, and for a hearing on claims that matter.
Start with characterisation. Set out the claim as a whole, including how its software and physical integers depend on each other (Full Court at [127]; allowing reasons at [149]).
Refuse the split. An objection that separates "the scheme" from "the computer" repeats the error the allowing reasons called filleting (allowing reasons at [97]).
Cite the law with pinpoints. The Full Court at [131], and the allowing reasons at [122]–[123] and [151]–[153], answer most objections built on improvement, conventional hardware or technical effect.
Use the Manual's own words. No factor is "a requirement", and conventional technology does not make a claim unpatentable. An examiner who treats the factors as decisive is departing from the section as well as the law.
Draft for it. Claim the interacting elements, and describe in the specification how they depend on each other. The Full Court relied on exactly that description (at [127]).

None of this makes every software claim eligible. While passing comment, Aristocrat expressed the (unsubstantiated) view that that former authorities concerned applications in which the conclusions were "plainly correct" in the words of the allowing reasons (at [121]–[122], echoed [132] of Aristocrat 2025). The argument is about the route an examiner must take, and on many claims the route decides the result.

The Manual is guidance for examiners. It is not the law. Examiners, delegates and courts apply the Patents Act as the courts have construed it, and where the Manual and the law differ, examination should follow the law.