It surprised me to learn that algorithms can be patented. If that's the case, why aren't famous algorithms like Dijkstra's algorithm patented? More importantly, how would they be enforced - wouldn't you need to look at the source code of all suspicious programs?
-
An abstract algorithm -- such as Dijkstra's algorithm -- cannot be patented (in the US or most other jurisdictions). On the other hand, a specific application of an algorithm -- e.g., a machine that performs a particular chemical transformation on a product by means of a set of computer instructions -- is likely patentable (subject to other considerations, such as novelty and non-obviousness). See Diamond v. Diehr (1981) and more recently Bilski v. Kappos (2010) and Alice Corp. v. CLS Bank International (2014).– user4545Commented Mar 17, 2015 at 18:47
Add a comment
|
2 Answers
I think you've hit on a good point for those considering pursuing patents. Before investing large sums of money on patenting, it is important to figure out how you will identify infringement.
Often, product data sheets, technical specifications, user guides, administrator guides, white papers, case studies and/or other technical and marketing literature are of use, but other times the under-the-hood details are unavailable and reverse engineering and/or discovery is required to confirm infringement.
In general, I recommend considering at least the following factors in connection with performing a cost/benefit analysis before pursuing a particular patent:
- Likelihood of use by others
- Ease of detecting infringement (e.g., reverse engineering, marketing materials, product specifications)
- Novelty and breadth of coverage
- Likelihood the invention will be used in current/future product(s)/service(s) of the patentee
- Useful life of the invention (“stop gap” vs. foundational)
- Cost of patent application(s) - preparation and prosecution
- Extent to which the invention is in an emerging field of technology
- Importance of the particular product(s)/service(s) in which the invention is used
- Importance of the invention to the particular product(s)/service(s) in which it is used (e.g., does it affect the consumer’s buying decision?, are there reasonable non-infringing alternatives?)
- License or cross-license potential
- Availability of trade secret as an alternative (e.g., is the invention exposed to the outside world?)
- Importance of a patent in strengthening the overall patent portfolio
- Importance of a patent to fundraising activities (e.g., venture capital or other investors may require IP protection before considering investing)
Mathematical / Algorithmic / Abstract Concept patents, AND/OR methods for calculating / performing them are only available in a limited number of territories (but the list includes the: USA, JPN, KOR, AUS, SGP), most jurisdictions expressly forbid any attempt to patent or copyright a purely abstract concept or algorithm. Anyway numerous famous algorithms have obtained a US patent, or have outstanding applications covering their calculation eg.
- Lempel–Ziv–Welch
- RSA
- BMI (methods for calculating: sliderule, central processing unit...)