The Oxford English dictionary defines an algorithm as “a set of rules that must be followed when solving a particular problem.” In other words, it is a series of steps, usually mathematical, through which a desired result may be reached. In computing, it is a series of commands through which a technological action is done. The series of steps thus consist the algorithm.

The question of how to patent an algorithm has a broad scope. The law covers inventions in the form of machines, composition of matter, and new processes. An algorithm by itself is not patentable as it is an abstract idea. It must be shown to be embodied or applied within a particular system process and not a standalone product by itself. The algorithm must improve the functioning of a computer or provide a technical advantage in a related field. It must also be able to articulate tangible results and technical benefits.

While an algorithm’s abstract flavor sets off alarms, its practical application is not so abstract that it would be ineligible for patenting.

When you’re patenting an algorithm, you’ll need to include the following information:

  • The algorithm’s name
  • The computer program that implements it
  • A description of how the program works
  • The purposes for which it was designed (the problem it solves)
  • Consideration for first-hand knowledge of the project or algorithm.

As long as you have first-hand knowledge of the project, POSITA, or practical application, you should be able to secure your patent.

Table of contents

  1. Practical applications of an algorithm may be patentable.
  2. Non-obviousness of the method.
  3. Consideration for first-hand knowledge of the project or algorithm.
  4. POSITA must be able to practice your patent.

Practical applications of algorithms that may be patentable

Algorithms can be patented, but only when their practical applications are technical. Algorithms and software can be patented. In fact, they are now the most frequently patented subject matter. Computer programs, mathematical methods, and business methods are often excluded from patentability. However, there are a few practical applications of algorithms that may qualify as patentable:

  • Algorithms that solve difficult mathematical or technical problems in an innovative way. These algorithms often involve the use of a new technique or method to solve an existing problem.
  • Computer-related methods, systems and articles of manufacture that provide new ways to identify, classify, organize and present information on the internet (e.g., search engines).
  • Software programs that perform functions such as word processing, database management and spreadsheet applications.

One example is an algorithm used in navigation. It can identify spoken words around it and respond accordingly. Unlike traditional voice assistants, it does not require a wakeword to respond. However, it can recognize spoken words even when you are not near it.

Guidelines on what makes an algorithm patentable

Patents can be applied for on algorithms. However, it’s important to note that not all algorithms are patentable. The US Patent and Trademark Office (USPTO) has published a set of guidelines on what makes an algorithm patentable. These include:

  • It’s novel – There must be no prior art available in the public domain that covers the same algorithm. This means that you cannot apply for a patent if someone else has already developed the same algorithm independently.
  • Non-obviousness

Novelty of the algorithm

An invention is novel if it is not something that has been publicly known or used anywhere in the world before the date of filing. An invention is inventive if it does not consist solely of commonplace activities and ideas. If a computer program can be considered as “purely mathematical”, then it would not be patentable. However, if the computer program has a practical application (for example, a method for carrying out a medical test), then it may be patentable even though it uses mathematical concepts.

Non-obviousness of the method

A question that often arises when applying for patents is “Is the algorithm obvious?” The standard for obviousness is the distance between a specific invention and the prior art. The invention must be an “adequate distance” above the prior art. This requires judgment. As a software engineer, you may wonder if a new algorithm is truly “obvious.”

The non-obviousness of an algorithm criterion is tricky to define. While computer scientists understand the term, software engineers do not. In some cases, an algorithm may be patented if it is based on algorithms that are known to other scientists in the field. In these cases, the algorithm may be easier to license or assign to other parties. Nevertheless, there is no universally accepted non-obviousness criterion for software and algorithms.

The objective of evaluating a patent’s non-obviousness is to determine the extent of an invention’s relation to the prior art. The court will evaluate the claims based on the extent of the prior art and whether a person of ordinary skill would have anticipated the invention. This approach may not be based on novelty or utility, but rather on its ability to prevent infringement.

A patent application that contains an algorithm will almost certainly be rejected unless it is sufficiently inventive. While it may seem logical, the invention should be so novel that the existing industry practice cannot be reverse-engineered. The AI must be able to identify the POSITA involved in the invention. In addition, it should be capable of identifying the POSITA in the application, ensuring it is used only in an official capacity. Likewise, the Human owner will not be able to reverse-engineer the AI process.

Consideration for first-hand knowledge of the project or algorithm

When you are ready to file a patent application on an algorithm, you should consider whether you have first-hand knowledge of the project or algorithm. If so, it is more likely that your application will be accepted by the USPTO.

If you do not have first-hand knowledge of the project or algorithm, there are several ways in which you can obtain a patent:

  • Obtain a written opinion from an expert who has first-hand knowledge of the project or algorithm. The expert’s opinion must be limited to explaining why he/she believes the claims are novel and nonobvious. This opinion should be included with your patent application as part of your disclosure to show how it is novel and nonobvious over the prior art.
  • Obtain an affidavit from someone who has first-hand knowledge of the project or algorithm. The affidavit must explain how it is novel and nonobvious over the prior art and should include copies of documents related to development of the invention (e.g., emails, notes). This affidavit should also be included with your patent application as part of your disclosure to show how it is novel and nonobvious over the prior art.

A Canadian team recently trained a machine-learning algorithm to identify the early stages of Alzheimer’s disease. The early stages of this condition are subtle enough for most specialists to miss, but trained algorithms are capable of identifying brain losses in seconds. Such AI developments are unlikely to be patented, but they could revolutionize research into the disease.

But before you start filing for a patent, consider the following:

  • Novelty- This means your invention cannot have been made public, not even by you, before the filing date.
  • Step ingenious- Your product or process should be innovative. It should not be obvious.
  • Industrial applicability-This means that the invention must be feasible to manufacture. You can also apply for a patent to a new type of playing card that is more comfortable than the existing ones. However, a patent is not possible for an idea for a card game.
  • Be a patentable subject matter.

POSITA must be able to practice your patent

Whether a POSITA (person of ordinary skills in the art) is required to practice your patent depends on your invention and case. In general, however, POSITAs must have an advanced degree and experience in a particular disease state. The US Supreme Court has noted that POSITAs must be innovative and fit multiple patents together. Furthermore, in some cases, courts have found that a POSITA must have a background in the field of drug discovery.

The invention must be non-obvious to a person with ordinary skill in the art (POSITA). This means that someone who is skilled in the relevant field of technology would not know how to create or improve on your invention without thinking about it first. It also means that your invention is not just an obvious improvement over what has been done before but does something different that makes it new and useful.

All about POSITAs

Patent practitioners and patent examiners often ignore the importance of a “person of ordinary skills in the art” (“POSITA”) in discussing important patent law issues like obviousness rejections, claim terms in litigation, and other crucial issues. The “person of ordinary skills in the art” is a hypothetical individual around which most of the patent law revolves. Despite the importance and power of the POSITA argument, they are often weak and conclusory. It is difficult to imagine yourself as this hypothetical person (as he or she doesn’t exist), but it is a crucial first step in patent prosecution.

In practice, this means that a POSITA can know everything (and all) written in any prior art document. This is the foundational definition that allows a POSITA to begin using objectively its knowledge to show the POSITA’s knowledge. The POSITA becomes more than a hypothetical result of a careful review of the prior art. A patent attorney can then craft persuasive arguments based on the POSITA.