14 February 2013

Research Affiliates Loses Appeal Against Business Method Rejection

Research Affiliates, LLC v Commissioner of Patents [2013] FCA 71 (13 February 2013)

Computerised InvestmentsA judge of the Federal Court of Australia has dismissed the appeal by Research Affiliates, LLC (‘RA’) against two decisions of the Australian Patent Office rejecting two patent applications relating to the construction and use of passive portfolios and indexes for securities trading.  The rejections were made on the basis that the claims of the applications were not directed to a ‘manner of manufacture’ – the test under the Australian law for whether a claimed invention comprises patent-eligible subject matter.

The RA method generates an index representing the relative value of a notional portfolio of stocks, similar to (for example) the Australian ‘ASX 100’, the US ‘Dow Jones’ and ‘S&P 500’, the UK ‘FTSE 100’, the Japanese ‘Nikkei 225’ and the Hong Kong ‘Hang Seng’.  Such an index is not merely the final numeric value which is normally published in financial reports.  It comprises a specific listing of stocks, along with a weighting value of each stock.  The index value is normally recomputed by adding all of the current stock values multiplied by their respective weighting values.

Conventional indexes are based on weighting the investments according to one of market capitalization weighting (i.e. in proportion to total company value – most of the well-known indexes are of this type), equal weighting (self-explanatory – examples are the Value Line index and the S&P 500 Equal Weighted Stock Index) or share price weighting (i.e. in proportion to the price of individual shares – for example the Dow Jones Industrial Average).  RA’s claimed contribution is to use measures of company size other than market capitalisation, or the other common weighting methods, to determine the proportions of shares to hold in a notional portfolio made up of a selected set of stocks.

There seems to be no doubt about the economic value of RA’s contribution.  The Federal Court decision reveals that RA has created an index of Australian stocks using its method, called the ‘Colonial eRAFI Large Index’, which it licenses for a ‘significant’ fee to fund managers, including Colonial First State Investments Limited.

Index Generation Method Not a ‘Manner of Manufacture’

The court’s primary reasoning for its decision to dismiss the RA appeal, finding the RA claims unpatentable and not a ‘manner of manufacture’, is to be found in paragraph’s [67] to [71].

The court says firstly that the RA index is ‘nothing more than a set of data’, ‘simply
information’, that ‘it is a set of numbers’ which is ‘no more a manner of manufacture than a bank balance’.  It then proceeds to complain that while the specification ‘appears to be intended to create the impression of detailed computer implementation,’ it ‘says almost nothing about how that is to be done’, going on to say (at [68]):

The reliance placed on the Colonial Index embodiment is a good example of what is not in the Specification. The discussion in the Specification provides no substantive detail regarding the implementation of the claimed method. The upshot of the discussion is merely that the method is implemented by a computer, but there is no disclosure of how that is to be done.

The court thus concludes (at [71]) that:

The case propounded by Research Affiliates depends upon the proposition that
information of economic significance, once entered into or produced by means of a computer, becomes an economically valuable artificially created state of affairs, and thus patentable. That proposition must be rejected

I disagreed with the Patent Office decisions (and I still do).

The reasoning of the Federal Court is somewhat different from that of the Patent Office, despite arriving at generally the same conclusion.  I have a couple of issues with the Federal Court’s decision, also.

What is the Potentially Patentable ‘Product’ of a Method?

Firstly, the court’s focus on whether or not the RA index is a manner of manufacture appears to be misplaced.  As the court itself states, in paragraph [15] of the judgment:

One must not take a narrow view of what constitutes the product of a method. If a method is purely an idea, that method’s product may be mere information, such that the method itself is then not patentable and is not a manner of manufacture. If, however, the method or idea results in a new machine or process, or an old machine giving a new and improved result, that new process or result should be regarded as the product of the method and the method is patentable. (Emphasis added.)

If it were the case that the actual product of performing a method must be a ‘manner of manufacture’ it would not be possible, for example, to patent a new and improved method of printing.  The product of a printing method is a document, which is merely a presentation of information that has never been held to be patentable.  In a modern printing device, it is likely that the printed page would be the result of digital processing of data representing the page layout, which is entered into a computing device controlling the printer.  Yet, according to the above reasoning, even if the printer hardware is not itself new, i.e. the method is implemented by providing a new and improved program for an existing printer, then the ‘old machine’ gives ‘a new and improved result’ which must be regarded as patentable.

Furthermore, the Full Federal Court in Grant v Commissioner of Patents [2006] FCAFC 120 cited the US case of State Street Bank & Trust Co. v Signature Financial Group 149 F. 3d 1368 with clear approval (at [21]-[24] and [32]).  That was a case in which the result of the invention was ‘a final share price momentarily fixed for recording and reporting purposes and even accepted and relied upon by regulatory authorities and in subsequent trades’.  Of course, a share price is an item of information that is not, in itself, patentable.  The Full Court plainly did not regard this output as the patentable product of the method.  It was the method itself – or, more precisely in the State Street case, a computer programmed to perform the method – that was regarded as patentable.

If there is something about the RA index that makes it relevantly different from the example of a printed page, or a share price, it is difficult to discern this from the judgment.

Confused Grounds of Invalidity?

Secondly, the court’s subsequent reasoning regarding the failure of the specification to disclose ‘substantive detail regarding the implementation of the claimed method’ appears to conflate issues of patent-eligibility with issues of full description, covered under section 40(2)(a) of the Patents Act 1990, and/or fair basis under section 40(3).

If it is the case that the specification lacks substantive detail to support the claims made, then those claims may well be invalid, under one or more of the section 40 grounds.  However, that is an entirely separate matter to whether or not the claims are directed to a patent-eligible manner of manufacture.  To give another hypothetical example, an application could be filed with claims plainly directed to a mechanical device, such as a motor, comprising a number of interconnected parts.  There would be no dispute that such claims are directed to a manner of manufacture.  However, if the specification failed to describe adequately how to put the machine into practice – or, indeed, if there were no description filed at all, the claims would be invalid on section 40 grounds.

In Lockwood Security v Doric Products [2004] HCA 58, the High Court could not have been any more clear regarding the necessity to consider each available ground of invalidity of a patent claim separately from the others, e.g. at [48]:

A specification can comply with s 40 even though what it claims has been invented is not a patentable invention because it is not novel or it is obvious. Each of the grounds of invalidity referred to in ss 18(1)(a), 18(1)(b)(i), 18(1)(b)(ii) and 18(1)(c) is distinct from the others. Thus there is a “logically precise” and “fundamental” difference between the objection for want of novelty and the objection for want of an inventive step. The lack of inventive step ground of invalidity is distinct from all the others, including fair basing. A patent can be successfully challenged on the ground that the claims are not fairly based even though every other possible ground of challenge fails. (Citations omitted.)

The RA method claims comprise steps of accessing data, processing the data to identify assets for inclusion in an index, accessing and applying a weighting function, and thereby generating the index.  Objections that these steps are not described in adequate detail within the specification do not properly address the ground of lack of manner of manufacture.

Surely the court cannot be saying that, had the description been more forthcoming with details of the accessing and processing steps, exactly the same claims could have been found patent-eligible?  If so, then that is, in my view, a completely illogical result.


In the earlier Grant case, the claims were considered, and rejected, at four different levels – examination, Patent Office hearing, a single judge of the Federal Court, and a Full bench of the Federal Court.  Mr Grant also sought leave to appeal to the High Court, but was denied.  At every level, the Grant claims were rejected, though on the basis of different reasoning in each case.

The final decision by the Full Court in Grant was, in my opinion, a good one, although it has proven less than useful in identifying the boundary between unpatentable schemes, such as Mr Grant’s, and patentable computer-implemented inventions.

As explained above, there appear to be clearly appealable aspects of the Research Affiliates decision, and RA seems to have substantial financial interests in Australia.  It has the right to appeal to the Full Federal Court within 21 days of the current decision.  I remain concerned that we still do not have clear guidance on what separates patentable from unpatentable computer-implemented or business process inventions.  I am therefore hopeful that this case, like Grant, will continue to progress up through the courts until we obtain a decision, one way or the other, that everyone can at least agree is consistent with established principles.

Of course, the RPL Central decision – in which I have an even greater interest – is yet to be issued.  Whether Justice Middleton, in that case, will be in any way influenced by Research Affiliates remains to be seen.

Image Copyright (c) 123RF Stock Photos


jo said...

"I remain concerned that we still do not have clear guidance on what
separates patentable from unpatentable computer-implemented or business
process inventions."

how do we make sense of these illogical decisions? if the judges are not either incompetent or corrupt, what could be the reason for issuing decisions that are confusing and illogical to experienced attorneys?

Robert said...

I think it's a good result, even if the reasons are confusing in parts.

The essence seems to be that if you have an idea that involves:

1. a process that produces some useful result, but that isn't patentable because it involves no manner of manufacture, and
2. an ordinary computer,

you can't insert a manner of manufacture into the process just by having some of the steps in the process performed through the ordinary use of the computer.

It's distinguishable from cases like the printer or the curve display because in both of them the computer is a necessary part of the process. You can't have a computerised printer or a computer display without a computer being involved. But here the judge found that the involvement of the computer was "no more
than the modern equivalent of writing down the index on pieces of paper" (para 72).

I'm not sure I entirely agree that the comments about State Street in Grant make State Street binding authority. But in any case, State Street can be distinguished in the same way; as the court in State Street said, "Given the complexity of the calculations, a computer or equivalent device is a virtual necessity to perform the task". Speed and automation were also essential. Emmett J's judgement in RA is based on finding this not to be the case.

I also don't think Emmett is conflating issues of eligibility with issues of full description. Rather, he's relating the lack of detail in the specification to the finding that the claims don't involve any specific effect being generated by the computer, and the later finding that the "aspect of computer implementation is nothing
more than the use of a computer for a purpose for which it is suitable". A more detailed specification wouldn't help, because the fundamental problem is that the involvement of a computer in the claimed method is essentially artificial. If anything, a more detailed specification seems likely to just highlight the completely generic nature of the computer's involvement in the process.

If this patent were eligible, you'd have to ask whether the same process would be patentable if instead of a computer it involved using a pen and paper. The result - a table of data - would be the same, and if anything the physical changes involved would be much more substantial! But nobody would claim that simply incorporating the ordinary use of a pen into an otherwise unpatentable process gives rise to a manner of manufacture allowing the process to be patentable (at least I hope not). Perhaps the courts are starting to see computers as the new pens.

(This is pretty much what Emmett J says in the last few paragraphs of his judgement, anyway.)

That all said, though, I'm not an IP lawyer and today was the first I heard about this case. I could be wrong. Any thoughts?

Mark Summerfield said...

I think the answer to your question can be found in the exchange between Robert and I here in the comments. Incompetence and corruption are not the only explanations for legitmate differences of opinion, and convoluted lines of reasoning, in complex and controversial areas of law.

A. Non said...

Therefore the upshot of this decision is that the patent office will continue to use this decision to reject business method claims.

Time for the attorneys to give it a rest and advice their clients that money is better used to define traditional claim sets?

Of course it may all change on appeal (if there is one).

Mark Summerfield said...

Yes, the Patent Office will do as it is obliged to do, and follow this decision of the Federal Court. I fear, however, that the convoluted nature of some of the reasoning will give the Office considerable residual discretion in how it applies the law.

I can assure you that this attorney, at least, has been advising clients for some time of the difficulties in obtaining patents in Australia for anything resembling a 'business method'. However, so long as inconsistency and uncertainty remain, some applicants will continue to make the choice to take their chances!

Post a Comment