Patent Office Reboots CRI Guidelines Yet Again: Removes “novel hardware” Requirement

Computer performance is measured in terms of the number of floating point operations per second, or FLOPS – today’s supercomputers typically clock in at several petaflops, or quadrillions of FLOPS. The Indian Patent Office’s performance with reference to Computer Related Inventions (CRIs) can be described by a similar term – flipflops. Not to be left far behind by developments in computing, the IPO also reached the quadrillion mark with its latest flipflop – amending the CRI Guidelines published in February 2016 to remove the novel hardware requirement.

On Friday (30 June), the Controller General passed Office Order No. 36 of 2017, revising the examination guidelines for CRIs published earlier in 2016. The most notable amendment involved the deletion of the requirement that patents for software could only be claimed in conjunction with novel hardware. The erstwhile three-step test for patentability (whose third step specifically stated that computer programs in themselves were unpatentable, and could only be claimed in conjunction with novel hardware) has vanished into thin air, as have the numerous examples of patent-ineligible claims (most notably Yahoo’s search engine claims, which were shot down by the IPAB). In addition, there seems to be some ambiguity on the implications of the revision for integrated circuits and whether the IPO would now be open to considering them as part of a CRI application (please refer to Anubha’s piece linked below for more on this).

The revisions have also been carried out with great attention to detail, with even subtle or implicit references to hardware being removed. For example, paragraph 4.4.5 of the 2016 guidelines contained two references to the “implementation” of claimed inventions. These have now been replaced by the word “performance”. Implementation presupposes hardware, while performance does not. Like traitors in Ancient Rome or the intelligentsia in Stalinist Russia, the novel hardware requirement seems to have been killed and buried in an unmarked grave, purged completely from official memory.


Section 3(k) of the Patents Act excludes mathematical or business methods, algorithms and “computer programs per se” from the scope of patentable subject matter. The interpretation of the words “per se” has been the subject of much controversy. In 2016, the IPO significantly revised (read: flipflopped) the draft CRI guidelines that it had released the previous year, and inserted the novel hardware requirement – meaning that its interpretation of Section 3(k) was that “per se” precluded patentability for inventions where the sole technical contribution was in the computer program, which could be run on a generic/general purpose computer or processor.

This obviously didn’t sit well with most large players in the industry, since they felt that they had to cross an additional hurdle to secure patents for their already inventive software programs. Their preferred interpretation of Section 3(k) is that the “per se” limitation may be circumvented in ways other than the coupling of novel hardware. Their justification for this interpretation came from the 2001 Joint Parliamentary Committee Report which inserted the provision in the first place:

“In the new proposed clause (k) the words ”per se” have been inserted. This change has been proposed because sometimes the computer programme may include certain other things, ancillary thereto or developed thereon. The intention here is not to reject them for grant of patent if they are inventions. However, the computer programmes as such are not intended to be granted patent. This amendment has been proposed to clarify the purpose.” (emphasis mine)

The JPC was comfortably vague about what these “other things” could be, and the two schools of thought seem to conflict on whether or not they are merely limited to novel hardware.

We should ideally take a step back and ask ourselves why the JPC report matters in the first place. It appears to be an imitation of the EU approach, where judges and scholars took to reading the traveaux preparatoires of the European Patent Convention to interpret the words “as such” in Article 52 of the European Patent Convention. There’s one very important difference between interpreting the EPC and interpreting Section 3(k): treaties lend themselves to purposive interpretation a lot easier than statutes, primarily because of the volume of preparatory material that is available.

The challenge to purposive interpretations that seek to narrow the ambit of the words “per se’ is strengthened by the fact that parliament later explicitly rejected an amendment to this effect in 2005, as Amlan notes here.


There’s also the question of the manner in which these guidelines operate in the first place. In an ideal world, some sort of litigation must have settled the point to a reasonable extent, allowing a court to interpret Section 3(k) one way or the other. In the absence of such litigation, the reign of the IPO’s guidelines appears to be the worst case scenario, especially given their proclivity to shuttle between radically opposing positions. Finally, the latest amendment seems to have been on the cards for quite a while now (DIPP stakeholder meetings from December 2016 seem to mention it, at the very least), but we haven’t been provided with the report of the “Expert Committee” that conducted the stakeholder consultation leading to this decision, so we have no idea what the Controller General’s reasoning for the change in policy is.

But wait, there’s more: the IPO’s official position on this seems to be that the new guidelines merely “clarify” the patent office’s earlier stance, and do not represent a policy change at all. An unnamed government source was quoted by the Press Trust of India as saying:

“The language of the guidelines issued in February 2016 was somehow giving the understanding that ‘novel hardware’ clause is mandatory to seek patents for CRIs, which was not the case. But the Indian Patent Office has revised those guidelines and clarified that this clause is not mandatory.”

I find the official’s use of the word “somehow” amusing, particularly given that the CGPDTM had to lop off half the document (17 of the 35 pages in the 2016 Guidelines have now been trashed) to bring about this “clarification”.

Over at the Centre for Internet and Society, our good friend Anubha has tabulated the highlights of the Guidelines’ revision. We will be bringing further posts later this week on the CRI guidelines, particularly with respect to the haphazard manner in which they are implemented by examiners. Until then, stay tuned!

Tags: , , ,

About The Author

4 thoughts on “Patent Office Reboots CRI Guidelines Yet Again: Removes “novel hardware” Requirement”

  1. Nice. Finally some common sense prevailed. I remember there was extended discussion on Spicy IP (post from Rajiv) when the previous guidelines were published in 2016. Lot of ppl made comments as being experts from industry and trying to justify this weird terminology “Novel Hardware”. If hardware is novel, why would someone bother about software functionality, they would rather claim the hardware itself. But then the novel functionality or solution to a problem would not be considered as patentable matter. I am glad IPO has saved themselves from lot of flak in future.

  2. Novel hardware was there for a reason and it is still there if you read the guidelines.
    There are only two possible artefacts, algorithms and SW. Algorithms are non-patentable. SW to exist needs implementation. No Indian patent granted so far has implementation details ! So in effect any patent to be granted can hinge only on the novelty of HW. The whole point is that innovative SW is not meant to be rewarded or protected. You have to live with that since that is the law ! Please also realize that the new section on excluding t eh exclusions i also against teh la. The parliament clearly when throwing out the 2004/5 amendment threw the technical effect language out. So as per law, algorithms are unpatentable, period, no exceptions can be given.

    CRI still says
    “Where no structural features of those means are disclosed in the
    specification and specification supports implementation of the invention
    solely by the software then in that case means in the “means plus
    function” claims are nothing but software.”
    Read this carefully. Structural features of a piece of SW is only algorithm, so this can only imply HW features. If it is general purpose HW, this cannot be provided since it
    by definition is GP.
    Show me one patent granted in India (cite a no) that satisfies this criteria.

    By the way I teach technology law, advise the Govt on technology legal issues and teach computer science at a premier institute and design supercomputers for a living. So hopefully I am an expert !

    1. I would also like to point out what Parliament said in this matter. From SpicyIP’s own analysis

      As is abundantly clear while teh per se exclusion exists, it use is severely constrained.
      Kind of like the teh rarest or rare cases in death penalty law. If someone in this forum thinks exclusions are easily possible, let us pick one patent and debate it. Let us see how any patent granted in India can actually pass teh algorithm test.

  3. I would also point to the wonderful construct of providing exclusions to exclude categories like mathematical methods. If mathematical methods are not patentable how can teh guideline make an exception for some methods ? It is patently illegal ! And curiously the exceptions relate to current litigation. Hmmm ….. To say that the guidelines are poorly drafted probably qualifies as the understatement of the century !

Leave a Comment

Scroll to Top