TPN Europe 2019: Keep software patents simple, says IBM lawyer
According to Kevin Fournier, IP lawyer at IBM in the UK, when drafting a software patent application, the best means of assessing an idea or an invention was to show it to a software developer.
Speaking at Technology Patent Network Europe 2019, hosted by WIPR in London last Thursday, June 20, Fournier said that “if software developers are impressed by the invention, then you’re really on a winner”.
Outlining his approach to patent claim construction, Fournier said “the worst thing that you can possibly do is have a method claim that requires the action of three or four different entities”.
The IBM lawyer advocated for the single method patent claim as a useful tool for patent attorneys.
The more complex a claim and the more actors are involved, then an applicant has “more to prove; [it’s] another argument that you have to make,” Fournier said.
When drafting a patent claim, the best way to avoid litigation, Fournier said, was to “use words in the claim that are very clear”.
In his experience, litigation tends to arise when the interpretation of a claim can be disputed. Accordingly, attorneys should “use basic words that people know … reduce the chances of an argument”.
Touching on the difficulties of operating in different jurisdictions with respect to software patents, Fournier commented that the US Supreme Court’s Alice/Mayo test for determining patent eligibility was a challenge for all attorneys.
“We’re all trained technically to try and follow logic, but it is very hard to follow the logic of that decision,” he said.
He agreed with the assessment of session moderator Richard Kennedy, partner at Venner Shipley in London, that the test was “inherently ambiguous”.
The UK jurisdiction also posed a challenge with respect to software patents, he said, particularly in proving an inventive step or technical contribution.
Commenting on the UK Intellectual Property Office’s approach, Fournier said that, in his experience, “the technical contribution seems to be interpreted more narrowly”.
“With the EPO, it’s a lot clearer [and] a lot easier, as long as you can show that the features that are critical to the invention are contributing to the technical effect,” he added.
Fournier also questioned Kennedy on his observations on emerging principles in AI patent drafting.
According to Kennedy, machine learning inventions tend to involve a two-step stage of development; the first being training a model based on training data; and secondly using that model to make predictions.
“One of the first decisions you’ll have to make is whereabouts in the pipeline are you going to target your claim,” Kennedy said.
In his perspective, it is the second stage that may be more fruitful in terms of patent protection.
“Looking at a trained model and trying to determine what methodology was used to train it is probably going to be very difficult,” he said.
With a commercially available AI-model however, there is much more scope to attempt to reverse engineer the product and obtain the technology behind it, Kennedy added.
Did you enjoy reading this story? Sign up to our free daily newsletters and get stories like this sent straight to your inbox.
Today's top stories:
TPN Europe 2019: Emerging tech rights owners must hedge their bets
Fed Circuit hands victory to ‘World of Warcraft’ game maker
