I'm very interested in improving upon a patented software method (A) which, in it's turn, is an improvement over a relatively old existing method (B).

Originally, my understanding was that the old method (B) would fall as prior art, and thus, the patent application (A) be rejected. However, there was adequate disclosure and the original paper is referenced to in the patent grant.

According to 35 USC 101-103, 112, improvements to processes, machines, manufactures, or compositions of matter, are patentable but only if they are new and useful, and non-obvious (among other requirements like adequate disclosure etc.) reference

As most legal concepts, these all seem very abstract to me, and I couldn't find any trusted source that could give me an adequate answer of what constitutes new and useful or non-obvious improvements. Or at least examples of it.

My question is, what should be taken into account for something to be considered new and useful, and non-obvious?

Also, in order to not infringe the patent, should the new method not violate all claims, or is it an infringement to violate anyone claim? I ask this because there's a series of chained claims in the referred document.




3 Answers 3


Criteria for you to get a patent on an improvement

Almost everything can be seen as an improvement on something from the past.

The claims of previous work are not particularly relevant to a patentablity analysis. Something in a journal article of an unclaimed part of a patent or a published patent application might be relevant.

As you say it must be useful, novel and not obvious.

Useful - is easy unless the invention is claimed in a way that it can be accused of being abstract - like many software claims. Drafting a software claim that passes "abstract" muster is now difficult. Abstract = not useful.

Novel - For a claim to be novel no single prior art document must have an enabled description of something with all of the elements in your claim. You can add something relatively trivial to a prior art teaching to achieve novelty. That doesn't end up doing you any good if a potential infringer of your claim can just leave out your added element with minimal loss of performance.

Non-obvious - the hard criteria. Would a hypothetical someone of ordinary skill in the art who knew everything ever published or otherwise publicly disclosed ever, anywhere, that was relevant to the problem come up with your system by combining aspects of past systems without undue experimentation. Books have been written on the subject and it is, as most things in patent law, a moving target.

Questioning an existing patent

The AIA law created a lower-cost way to try to get an existing patent invalidated. However, "lower" means lower than a multi-million dollar patent law suit. The new IPR process is still a small number of hundreds of thousands of dollars. There is also, under the AIA, a free process for submitting references (3rd party pre-issuance submission) to the USPTO that the examiner is required to take into consideration. There is a window during the proceedings where this is in effect.

  • Thank you, George, I edited the title. For the case in question, the patent itself was created based upon extension of existing work. Could any patent like this (or at least some of the claims) be questioned on the grounds of prior art? May 17, 2020 at 18:17
  • 1
    @ERM Any patent can be questioned on the basis of prior art. The fact that it is or isn’t an improvement on a prior patent is irrelevant.
    – Eric S
    May 18, 2020 at 12:33

Although George White's answer is excellent, I wanted to explicitly answer your last question.

Also, in order to not infringe the patent, should the new method not violate all claims, or is it an infringement to violate any one claim?

To infringe on an existing patent, it is only required that you infringe on a single claim. Please understand that with improvement patents, it is very common that the new patent can not be used without a license to the original patent. For instance let's say patent A has a method with steps 1, 2, and 3. The new patent B has a method with the same steps 1, 2, and 3 but when adding step 4 provides an unexpected improvement in performance. The new patent B has real value, but can't be practiced without infringing on patent A.

  • thanks - I guess I didn't notice the last paragraph of the question.
    – George White
    May 18, 2020 at 18:59
  • @GeorgeWhite I didn’t either until recently. By the way, I didn’t realize the abstraction argument is a matter of usefulness. I learned something.
    – Eric S
    May 18, 2020 at 19:50

For your work to be considered new it shouldn't be disclosed in any prior art publications. As far as usefull it must considerably improve or achieve what prior art failed to, in simple terms the improvement must not be for asthetic puropose.

For your work to be non-obvoius it must have not been known from gerneral knowledge available to a POSA(PERSON SKILLED IN THE ART) i.e., a person who had experience in that pertinent field and could derived such improvement.

  • It is NOT correct that useful means a better result that prior art. Useful is simple useful in plain English and a very easy hurdle, unless it is software or anything else that can be characterized as "abstract".
    – George White
    May 17, 2020 at 5:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .