3 June 2024

G1/19 (Pedestrian Simulation): three years on

On 10 March 2021, the Enlarged Board of Appeal (EBA) issued its decision in G1/19 (Pedestrian Simulation). It is the most authoritative case on the assessment of computer-implemented inventions (CIIs) at the European Patent Office (EPO).  It affects all European patent applications having claims containing a mixture of technical and non-technical features (so-called “mixed-type” claims) and is highly relevant to the burgeoning areas of AI, machine learning, data analytics and material informatics.

The judgement set out a coherent way of assessing mixed-type claims. It stated that all CIIs should be examined in the same way and approved the approach to examining mixed-type claims set out in T 641/00 (Two Identities/COMVIK). It relied on the principle already applied in the chemical field that claims should be technical over the whole scope of the claims.

Number of citations

In the three years after it was handed down, G 1/19 has been cited in about 100 appeal decisions. COMVIK, however, is still routinely mentioned; in fact, it has been cited more often than G 1/19 over the same period. Nonetheless, G 1/19 is becoming one of the most cited EBA decisions; for comparison, G 3/08 (Programs for computers), G 2/10 (Disclaimer) and G 1/15 (Partial Priority) were only cited in about 30 to 50 times in the three years after they were issued; some EBA decisions are rarely cited at all.

Interestingly, the popularity of G 1/19 seems to be mainly down to a handful of Technical Boards of Appeal. Of the nine TBAs that hear appeals relating to physics and electricity, three are responsible for nearly 60% of citations of G 1/19 and just one Board, 3.5.07 (the one which referred the questions to the EBA), is behind about half of those.

How Technical Boards of Appeal (TBAs) have applied G 1/19

What can we learn from these TBA decisions about how G 1/19 has been applied in practice?

First, Boards have been looking at whether a claim is technical over its whole scope (point 84 of G 1/19).

For instance, in T 2681/16, a claim directed to measuring blood glucose variability was only found to be allowable once it had been narrowed to recite a minimum number of data points so that its technical effect (that of providing a measure of glucose variability and a better prediction of a glycaemic event) was achieved over whole claimed subject matter.

The same Board also applied the principle in T 1892/17 regarding a patent relating to optimizing consumption of electric power over time, and maintained the patent as granted.

In T 1370/18, the Board found that encoding a symbol for transmission achieved the technical effect of reducing the amount of data to be transmitted over substantially the whole scope of the claim and remitted the case to the examining division with an order to grant a patent.

Thus, applicants need to ensure the requirement that a claim should be technical over the whole scope is met. This means that features contributing to the technical character of the claim must do so for all specific embodiments of the claimed invention.

Secondly, Boards are examining whether there is a credible technical effect. If there is no technical effect, then there can be no inventive step (point 49 of G 1/19).

In T 3208/19, the Board re-iterated that only those technical effects that are at least implied in the claims should be considered in the assessment of inventive step, and if the invention does not solve a technical problem, it has no distinguishing features which could contribute to inventive step.

No credible technical effects were found in T 0297/20, which related to enabling an operator to obtain a better overview of a presentation of electrical power transmission control stations connected to power lines in a power transmission or distribution system, in T 2014/21 which concerned a plant simulation device, and in T 1560/19 pertaining to a data processing device for characterizing behaviour properties of equipment under observation.

Thirdly, when looking at potential technical effects, the Boards are looking closely at “downstream” effects (point 96 of G 1/19). Boards are also considering whether decisions are made for business reasons (point 123) and how numerical values are used (point 124).  For example, where a claim relates to the generation of a parameter, a Board considers carefully the technical purpose or subsequent use of the parameter.

For instance, T 1313/17 concerned a method for optimizing quality, productivity and/or profitability of a manufacturing process in a textile plant. Despite the industrial context, the Board found that the potential use of data stored in a database, as well as the parameter determined by comparison of the data stored in the database, was not limited to technical purposes.

T 1035/18 related to a computer-implemented method for predicting electrical energy production of a photovoltaic system included in an aircraft. The applicant argued that the predicted amount of electrical energy translated into estimated fuel savings which provided a technical effect. The Board, however, argued that the technical effect was not a direct consequence of the estimated fuel saving and, moreover, that the estimated fuel savings can also be used for (non-technical) business decisions.

T 1371/16 concerned apparatus for aiding the design of a wire harness which outputs numerical data about a wiring path design. The Board noted that the claim did not specify any further use of this data, further properties or specific data format that could limit the possible uses of the data and took the view that the data could be used for non-technical purposes, for example informational, study or training purposes.

Therefore, it is important for applications to convey clearly the purpose of any parameters or data generated, how they are used and, if appropriate, their structure or form.

Drafting applications with G 1/19 in mind

In G 1/19, the EBA stated that it did not see a need to require a direct link with (external) physical reality in every case for there to be technical effect (point 87).

That may be, but it sure helps.

A couple of examples show how platforms used for material informatics can be linked to physical reality to produce a technical effect.

EP 3 809 118 B1 is directed to a computer-implemented method for predicting a property value of interest of a material. The claim recites providing a population of infrared spectra of samples as an input for predicting the property value.  Thus, the direct link with external physical reality is provided by the real-world measurements.

In the area of applied AI, this requirement is often met since these cases involve training or inference using real-world data, such as images, sound, or transducer signals.

EP 4 038 620 B1 is directed to a method for generating a thermoplastic composition. The composition is generated by a computer system which has access to a database of known compositions, and the computer system is connected to a facility for producing and testing.  In this case, the direct link with external physical reality is provided by testing the composition in the real world.

These provide examples of how computer simulations and machine learning can be patented by linking them to the real world, in one case, through input data and, in the other, through its output.

Is the EPO being consistent?

A brief review of some of the European patents granted since G 1/19 suggests that the EPO may not be entirely consistent in the application of the principles in G 1/19.

Take, for example, EP 3 779 619 B1 granted after G 1/19.

Claim 1 is directed to a computer-implemented method for analysing the dependability of a technical system. The claim includes the steps of receiving a configuration of a technical system and operating risks of sub-systems and determining emergent risks. The determined emergent risks are immediately passed back to the design process by installing additional protective measures in the installation.

The technical effect is improved risk analysis of operational safety of a technical system. The results, however, are passed back to the design system and installation of the additional protective measures does not seem to be inevitable.  It can be argued though that the emergent risks could be used for non-technical purposes, such as informational purposes. Moreover, it can also be argued that the decision of whether or not to install protective measures involves human decision making.

Final remarks

Although G 1/19 has not caused a step change in how mixed-type claims are examined at the EPO, the way it is being applied and the questions that are being asked (particularly regarding technical effect) seem to suggest that TBAs find the decision helpful, providing a better framework for examining claims.

Whilst G1/19 provides a highly effective approach, it remains rooted in COMVIK and consequently seems to be too rigid. The requirement of further technical effect and what constitutes a technical effect seems to exclude protection for many forms of computer-based tools, such as simulators, and the use of digital twins.

The EBA states that the notion of technicality needs to remain open, but it is difficult see how this can be done given the constraints of COMVIK. The COMVIK approach was established over 20 years ago and is well embedded in EPO examination practice: a radical change seems unlikely.