29 January 2012

IP Australia and US Courts Grapple with Computer-Implementation

Sheng-Ping Fang [2011] APO 102 (20 December 2011)
Dealertrack, Inc. v. Huber No. 09-1566 (Fed. Cir. 20 January 2012).

Patentable subject matter – manner of manufacture – abstract ideas – whether computer-implementation sufficient to establish patent-eligibility

ENIACNo matter what jurisdiction you are looking at, there are some things which are just not patentable.  In some cases, such as the countries of the European Patent Convention, the subject matter that is not eligible for patentability is expressly codified.  In others – such as Australia, New Zealand, the US and Canada – there is no definitive legislative list of ineligible subject matter, although over time the courts have identified those things for which patents cannot be granted, regardless of novelty or inventiveness.

In all of the jurisdictions in which patent-eligibility has been left to the courts, there is broad agreement on the unpatentability of certain subject matter that is regarded as the pre-existing legacy belonging to all, which cannot be monopolised by any one entity. 

The US Supreme Court has established three exceptions to the broad principle that all machines, processes, manufactures and compositions of matter are patentable under 35 USC §101 – laws of nature, physical phenomena and abstract ideas. 

In Australia, it is settled law that the ‘manner of manufacture’ test for patent-eligibility excludes laws of nature, mere discoveries, ideas, scientific theories, schemes and plans.  Mathematical formulae and algorithms are also excluded, to the extent that claims are not meaningfully limited to their use as part of a patentable practical application.

It can therefore be seen that, while the precise terms used differ in the two countries, there is a broad similarity between the fields of excluded subject matter in Australia and the US.

However, decision-makers in courts and Patent Offices of both countries continue to grapple with the extent to which so-called ‘business methods’ should be patent-eligible.  In particular, two recent decisions – one from the Australian Patent Office, and another from the US Court of Appeals for the Federal Circuit (CAFC) – illustrate the lack of any clear direction on one particular question:

If a patent application discloses a method or process which, in itself, falls outside the scope of patent-eligible subject matter, can it be rendered patentable by virtue of machine-implementation, i.e. in the form of a programmed computer?


It does not seem unreasonable that the answer to the above question should be ‘no’, and this certainly seems to be the direction in which Australian and US decision-makers are leaning.  However, the legal basis for this conclusion is problematic in both countries.

On the face of it, a computer is a machine.  When programmed to do something new and useful it becomes a new and useful machine.  US patent law expressly provides for patenting of new and useful machines.  The concept of a ‘manner of manufacture’, from its very inception in the Statute of Monopolies of 1623, also encompasses new and useful machines.

On this logic, the answer to our question should be ‘yes’.  Once a claim is directed to a machine of some description, the eligibility test is resolved, and we should just move on to the substantive tests of novelty, inventive step, and so forth.

But if this were so, then any new and nonobvious algorithm would be patentable in computer-implemented form, independent of a practical application.  Many general-purpose algorithms would pass the ‘novel’ and ‘nonobvious’ tests at the time of their development, and their inventors would certainly be entitled to file patent claims directed to as many practical applications of the algorithms in fields of patentable subject matter as they are able to think of. 

But in many cases the most useful implementations of algorithms are in software, and a broad patent on a computer-implementation of the algorithm divorced from any application would, in practical effect, be equivalent to a patent on the algorithm itself.  When courts talk of claims that would ‘preempt’ all use of an idea, this is the kind of thing they are generally trying to get at.


The leading cases in Australia and the US on the patentability of business methods are the decision of the Full Court of the Federal Court of Australia in Grant v Commissioner of Patents [2006] FCAFC 120, and the decision of the Supreme Court of the United States in Bilski v Kappos (No. 08-964 28 June 2010).

Grant’s invention was a method of protecting an asset involving the use of a trust, while Bilski’s was a method of hedging in energy markets.  In both cases the inventors declined to claim the invention in an expressly computer-implemented form, forcing the court to make its decision with regard to the business process as such.

Also in both cases, the respective courts found the claimed methods to be unpatentable, but were at pains to point out that this was not because they were directed to business methods, and that in neither country is there an exclusion from patent-eligibility simply because something might be described as a ‘business method’.

In the US, Bilski’s claims were found to be directed to an abstract idea.  In Australia, Grant’s claims were found not to produce an ‘artificially created state of affairs’, as required for a manner of manufacture.

In neither case was the court required to offer any opinion as to whether it would have reached a contrary conclusion had it been presented with different claims, in which the invention was somehow limited to a machine-implementation.  Nor did either court do so.

Which leaves us where we are today, in both jurisdictions – with no clear guidance as to exactly what degree or nature of machine-implementation might suffice to render an invention patent-eligible.


The invention in Dealertrack was a computer aided method of managing a car loan credit application through an electronic clearinghouse.  Basically, the clearinghouse receives the usual information provided in a loan application, and sends it off to one or more prospective credit providers, either simultaneously, sequentially, or indirectly, in a effort to identify a funding source willing to provide the loan.

In particular, the court considered the patent-eligibility of the following claim:

1. A computer aided method of managing a credit application, the method comprising the steps of:

[A] receiving credit application data from a remote application entry and display device;

[B] selectively forwarding the credit application data to remote funding source terminal devices;

[C] forwarding funding decision data from at least one of the remote funding source terminal devices to the remote application entry and display device;

[D] wherein the selectively forwarding the credit application data step further comprises:

[D1] sending at least a portion of a credit application to more than one of said remote funding sources substantially at the same time;

[D2] sending at least a portion of a credit application to more than one of said remote funding sources sequentially until a funding source returns a positive funding decision;

[D3] sending at least a portion of a credit application to a first one of said remote funding sources, and then, after a predetermined time, sending to at least one other remote funding source, until one of the funding sources returns a positive funding decision or until all funding sources have been exhausted; or,

[D4] sending the credit application from a first remote funding source to a second remote funding source if the first funding source declines to approve the credit application.

On its face, this claim includes plenty of limiting hardware details.  The method is expressly stated to be ‘computer aided’.  Furthermore, there are display devices, data entry devices and terminal devices.

However, the court found the claim to be ineligible for patent protection, relying on the ‘abstract ideas’ exclusion (slip op. at 35):

Dealertrack’s claimed process in its simplest form includes three steps: receiving data from one source (step A), selectively forwarding the data (step B, performed according to step D), and forwarding reply data to the first source (step C). The claim “explain[s] the basic concept” of processing information through a clearing-house, just as claim 1 in Bilski II “explain[ed] the basic concept of hedging.” … The steps that constitute the method here do not “impose meaningful limits on the claim’s scope.” … Neither Dealertrack nor any other entity is entitled to wholly preempt the clearinghouse concept.

Astute readers will note that there is a gloss on the meaning of ‘abstract ideas’ here.  In particular, as the court expressly states (slip op. at 30): ‘any invention within the broad statutory categories of § 101 that is made by man, not directed to a law of nature or physical phenomenon, and not so manifestly abstract as to preempt a fundamental concept or idea is patent eligible.’ (Emphasis added.)

So ‘abstract idea’, according to the CAFC, means an idea which is ‘so manifestly abstract as to preempt a fundamental concept or idea’.  We are not sure that this definition leaves us much better off!

The court also found that Dealertrack’s patent suffered from a lack of detail in the description (slip op. at 35):

Although the district court construed “computer aided” as a limitation, the ’427 Patent “does not specify how the computer hardware and database are specially programmed to perform the steps claimed in the patent.” … The claims are silent as to how a computer aids the method, the extent to which a computer aids the method, or the significance of a computer to the performance of the method. The undefined phrase “computer aided” is no less abstract than the idea of a clearinghouse itself.


The applicant’s invention in Sheng-Ping Fang relates to a system for awarding prizes in a customer loyalty program.  Prior art programs (such as ‘frequent flyer’ or ‘air miles’ programs, or store points programs) award only one type of unit (e.g. ‘miles’ or ‘points’), which the customer must accumulate in order to obtain higher level (i.e. more valuable) rewards.  The invention provides for different ‘symbols’ to correspond with different prizes, thus separating prize value from accumulation of points.  The symbols may be ‘unique’, in that they correspond with only a single prize, or they may be ‘shared’, in that they may contribute towards redemption of a number of different prizes.

The patent application envisages that the reward system would be applied in an online environment, and that the symbols would therefore be collected by a user during use of an online system (such as an e-commerce web site), and displayed on the user’s screen.

The applicant’s claims were rejected by an examiner through a number of reports and amendments.  Not all of the amendments were considered allowable.  When the matter came up for a hearing, the Hearing Officer considered only the most recently filed set of claims which he found to be supported in the specification as filed.  By way of example, independent claim 1 as considered by the Hearing Officer is as follows:

1. A system for providing an award program on one or more servers and electronic devices, said system comprising:

a) a machine readable or executable instruction for providing a plurality of qualified events, wherein said qualified events provide a plurality of unique and, or shared award symbols to be collected or accumulated for completing a selection or plurality of complete award symbols with identical and different visual or physical characteristics and,

b) a machine readable or executable instruction for providing a selection or plurality of award symbols, wherein said award symbols consist of unique and, or shared award symbols that are selected from a collection of symbol pieces and, or symbol clones associated with a selection or plurality of said complete award symbols and,

c) a machine readable or executable instruction for providing a selection or plurality of prizes that are displayed alongside their corresponding or associated complete award symbols, said prizes are awarded on full or partial completion of said complete award symbols.

As in Dealertrack, this is a claim which, on its face, includes specific limitations tying it to a machine implementation.  However, the Hearing Officer made findings which – although based on a different statutory definition – are remarkably similar to those in the (more recent) US Dealertrack decision (at [59]-[61]):

From reading the specification as a whole, it is clear that the core of the invention lies in the use of the unique and shared symbols. This is also supported by the applicant’s submissions, both in writing and also orally at the hearing. These symbols enable users to win different prizes at different time periods, without the prizes being grouped and awarded hierarchically according to monetary value. Whilst I accept that the use of these symbols may provide advantages in relation to how prizes are won and thereby motivate users to engage in further activity beneficial to the online retailer, it does not alter the fact that what is claimed is a scheme for awarding prizes in a loyalty program. The unique and shared award symbols are merely information generated in the operation of the scheme. “Information even if represented in a physical way has never been considered sufficient for patentability save for some material advantage or mechanical effect in the arrangement of the information” (Iowa Lottery). Although the invention can clearly be said to be in a field of economic endeavour, namely retail trading, the use of different types of award symbols in a loyalty scheme is in my view an abstract idea, mere intellectual information involving new symbols, which had never been held to be patentable subject matter.

I accept that being an online award program the system has to be implemented using a computer system and in that sense a computer system is integral to the invention. However when I read the specification as a whole, there is very little description or details of the computer system. While there are brief references to computer programs, machine readable/executable instructions, machine readable devices, application and database servers and programming languages there is nothing in the specification to suggest that the use of these software programs or computer devices has brought about any substantial physical effect or transformation in the implementation of the award program.

Again we see here the basis for ineligibility is that the claimed invention is fundamentally an ‘abstract idea’, and that a general description and claiming of computer-implementation is insufficient to bring it within the realm of patentability.

We note that the claims were also found to lack an inventive step in view of prior ‘off-line’ product promotion schemes, including the Streets Paddle Pop ‘Lick-a-Prize’ competition which (although the specific internet materials relied upon date from 2008) has been running periodically in Australia for as long as we can remember!


Regular readers will be aware that we are generally not backward with our opinion when we think that IP Australia has got something wrong.

In this case we are not at all sure that the outcome – in either the US or Australia – is incorrect, although the reasoning applied in order to get to the result appears in both cases to be problematic.  We think that there is a substantive difference between these kinds of cases, in which computers are used as a convenient means of implementation, and cases such as Research Affiliates, in which new methods of data processing are applied to produce a ‘useful, concrete and tangible result’ – in the sense of the US State Street decision, cited with approval in Grant (see Why IP Australia is Clearly Wrong About Research Affiliates).

At least the US CAFC is trying to draw distinctions, having recently found claims directed to a practical application of the idea that advertising can be used as currency to be patent-eligible in Ultramercial v Hulu.

IP Australia, by rejecting a broader range of subject matter without discrimination, is only confusing the issue, making it almost impossible for applicants and their professional advisors to determine what is properly patentable (or at least acceptable to the Patent Office) and what is not.

As we have previously indicated, the Research Affiliates decision has been appealed to the Federal Court of Australia, where the claims are quite likely to be found patentable, considering their closely analogous nature with the State Street case.  This will not help us to determine whether claims such as those in Sheng-Ping Fang should be patentable, or to gain a definitive answer to the question of whether, and in what circumstances, an unpatentable process might be rendered patent-eligible by machine-implementation.

This is most unfortunate, because the public really is entitled to some clarity and certainty in this area, while IP Australia continues to serve up confusion and obfuscation.


dk said...

the process disclosed in sheng-p Fang is a patentable process and so is the computer implementation of that process. IP Australia has clearly gotten it wrong!

we can see how determined ip australia is/has been in rejecting computer related inventions by using obscure websites/publications that would not be considered as permissible evidence in court litigation.

Zoobab said...

"On the face of it, a computer is a machine. When programmed to do something new and useful it becomes a new and useful machine."

No, it is not a new machine. It is just the same machine, but following different instructions.

If a machine can have 10 different states, programming it to be on a particular state does not make it a new machine.

Patentology (Mark Summerfield) said...

I would argue that your example is too trivial to be meaningful.  Attempted reductio ad absurdam is no basis for law or policy making!

Certainly, placing a mchine with only 10 states into one of those entirely  predesignated states does not involve invention.  If I have, say, a (known) 4-bit register, I cannot claim as my own 'the 4-bit register, when in state 0x0A'.

But that is not what a computer-implemented invention generally does.  A computer program firstly places the machine into an initial state, and then causes it to proceed through a sequence of subsequent states which themselves will depend upon inputs from outside the machine.

The initial state (i.e. the program and its static data) is uniquely the product of the programmer's creative effort.  Of all the states that a typical computer may have, only a vanishingly tiny fraction of them do anything useful.  You might just as well have the proverbial infinite monkeys with typewriters trying to generate the works of Shakespeare.

Furthermore, the functionality of the program, i.e. the unique sequence of states through which the machine passes in response to unique inputs, is no more or less the product of the programmer's intellect than the functionality of a mechanical device, such as a motor, is the result of the engineer's intellect.

Our intellectual property laws are based upon the notion that there is social and economic utility in providing incentives for people to be creative, and to be able to secure fair returns on that creativity.  You can agree or disagree with this as a general proposition, but either way there are no effective and justifiable criteria for drawing a bright line between 'protectable' dedicated-purpose machines, and 'unprotectable' programmed machines.  All this does is to arbitrarily exclude an entire class of innovators from access to the system.

Zoobab said...

"All this does is to arbitrarily exclude an entire class of innovators from access to the system"
All you do with the patent system is the same:

"All this does is to arbitrarily exclude an entire class of programmers from access to the market"

Furthermore, it is insane to think that programmers might be restricted to by the state to give instructions to programmable machines.

Patentology (Mark Summerfield) said...

You are absolutely right.  Patents are state-sanctioned monopolies.  This is not news.  You might just as well say that a patent on a drug excludes an entire class of competing pharmaceutical companies from access to the market for that drug.  Because that is exactly what it does.  And a drug is nothing more or less than a particular configuration of pre-existing atoms into a compound that happens to produce a useful effect in the human body.  Physics, chemistry and biology at work!

My point is not that you are 'wrong'.  It is that if you are opposed to patents on computer-implemented inventions on grounds that they restrict what others can do, then you are simply opposed to patents in general.  If so, then there are plenty who share your views, and debate on the topic is perfectly valid.  But you need to stop asserting that there is something special about software, unless you can articulate more clearly exactly what that is.

A person who designs an improved electric motor does so subject essentially to Newton's laws and Maxwell's equations.  Nothing more modern is required to describe the physics of such an invention.  And those few equations are far simpler than any modern computer system.  According to your logic, all the inventor of the new motor does is to 'program' the state of the materials making up the machine into a new configuration, the validity of which was entirely predetermined by the laws of nature.  Physics does the rest.

I have no problem with people being philosophically opposed to the patent system.  But I do wonder why they would be reading this blog, which is primarily about practical issues relating to the state of the world, the economy, commerce and the law as they actually are, not as some idealistically-inclined people would like them to be.

dk said...

  "Furthermore, it is insane to think that programmers might be restricted
to by the state to give instructions to programmable machines."

This statement is flawed. The state obviously does NOT restrict
programmers from implementing ALL computer-related instructions. You
can't patent something that has been around many years before you filed
your patent application.

dk said...

Its the utility and functional changes to a machine that makes the machine 'new' in the sense that the machine has become economically and functionally useful when running the software.

Zoobab said...

"My point is not that you are 'wrong'.  It is that if you are opposed to patents on computer-implemented inventions on grounds that they restrict what others can do, then you are simply opposed to patents in general.  If so, then there are plenty who share your views, and debate on the topic is perfectly valid.  But you need to stop asserting that there is something special about software, unless you can articulate more clearly exactly what that is."

My point is that the state interferes with freedom of expression by asking computer programmers like me to respect patents before writing code, and they should not be allowed to make such requests.

@dk"The state obviously does NOT restrict programmers from implementing ALL computer-related instructions."

So what software patents are supposed to do then?

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.