Invalidation searches for software patents have had little success in identifying strong prior art due to the fact that most software innovations are not represented in patent databases when they first enter the marketplace. In general, prior art is rarely available from non-patent sources such as GitHub, Stack Overflow, academic papers, technical standards, archived product documentation, and records of open source projects. Invalidation searches often focus on only one industry, only on modern products, only English source documents, or only on language that is keyword matched to the terms of the patent claims. Since software patents are written at a level of abstraction, prior implementations that are described as architectures, workflows and design patterns are frequently not considered. Thus, the essential element of conducting a successful software patent invalidation search is to take a broad-based, cross-domain and not patent-centric approach.
I've come across this situation many times, where companies assume that an invalidation search is "thorough", simply because it has discovered patents and academic papers; however, most of the time, that is where the invalidation search ends. There are many examples of prior art that do not exist in any patent database. Examples include: GitHub repositories, Stack Overflow answers, RFCs, Open Source Release Notes, academic slide decks, etc. Additionally, I have seen multiple claims that were able to survive the initial review process simply because no one took the time to look at a 5-year-old README file that articulated the workflow. The searches teams conduct for product documentation, changelogs, and user guides on older SaaS tools tend to be very poor. Therefore, if a feature is known to the general public, is readily accessible to people and is online, it is considered to be prior art. Unfortunately, most of this material is buried in help center pages or archived blog posts, and is not indexed like patents. The searches teams conduct tend to be too generalized in nature. For example, many times prior art is found within vertical forums that cover specific industries (e.g., construction, finance, healthcare). For example, what is considered to be a "novel" workflow may have existed for years in a niche tool used by a specific industry. Another thing that tends to go unnoticed when examining public domain material is when features are released by software publishers. For example, sometimes a feature is released very quietly and is later given a new name. Therefore, if teams do not search through the version history (i.e., previous versions of the product), and release dates, they will not find this information. Although this process is simple, it is not a quick process. The best course of action is to search for prior art where practitioners have worked (as opposed to searching only where attorneys search). It is generally in the location where practitioners have worked where the greatest invalidation evidence can be found. Adam Scuglia works as an Account Executive with Cortex, assisting construction and project teams consolidate their drawing workflow, automate versioning and communicate confidently through AI-assisted drawing management.