In the trembling dawn of the production floor, where the rhythmic chime of machines was the only music allowed, the inventor toiled in silence. For years he had watched as human frailty compromised mechanical efficiency. Operators—flawed, unpredictable, slow—had become the bottleneck in a system otherwise engineered for perfection. Something had to be done.
The problem was deceptively simple: the productivity of highly sophisticated production devices was being compromised not by mechanical failure, not by software bugs, but by the human beings entrusted with their care. The inventor observed patterns, tallied metrics, and assembled data with the same grim focus one might expect of a mortician performing an autopsy. Some operators consistently caused disruptions—slower output, misconfigured components, degraded performance. But the factory marched on blindly, allowing each to repeat their offenses indefinitely.
To the inventor, this was intolerable.
Thus emerged the idea: a system to monitor performance, identify low performers, and restrict their access to the machine. A performance restriction not dictated by a manager’s hunch or emotion, but by the machine itself. The production system would remember everything: every component placed, every nozzle configured, every deviation, every delay. It would detect lowered productivity, identify the culprit, and automatically apply restrictions to prevent future harm. If the operator failed again, the restrictions would become harsher. Eventually, access would be revoked entirely.
And so the application was filed.
It was described with meticulous care. A production system, with a production device like a mounting machine, and a management computer. The computer would display a screen offering work modes, tracking the operator’s choices. A memorizing means would store both the operator’s performance information and restriction rules. A detecting means would monitor for decreased productivity. An identifying means would link the decline to a specific operator, and a restricting means would limit their access accordingly.
The idea seemed elegant. The inventor dreamed of a future where the machines not only produced but judged, learned, evolved. A future purged of human inefficiency. He waited for validation.
Instead, the reply arrived from the European Patent Office.
The letter was brief, and yet infinitely complex. The application had been refused. The words on the page had a metallic taste, like steel.
“The subject-matter claimed in all requests lacks an inventive step. Furthermore, certain claims contain added subject-matter and lack clarity.”
No signature. No empathy. Just the faceless judgment of the system.
The inventor appealed. He appealed not just with documents, but with his spirit. His words were passionate yet precise, an engineer’s plea dressed in a lawyer’s suit.
“This is not mere management,” the inventor wrote. “The restriction is individualized. It is not a blanket policy applied by a human supervisor, but a direct response to a technical condition, applied selectively and dynamically. It improves productivity by ensuring that only those operators capable of maintaining system efficiency are allowed to interfere with the machine.”
He argued that the cited prior art did not offer such a solution. There was no matrix of operator-specific performance data. No dynamic adjustment of restrictions based on technical input. No intelligent link between machine and man.
He requested oral proceedings, if necessary. He filed auxiliary requests—six of them—each with added detail, narrowed scope, precision added like screws to a circuit board. He hoped one might be acceptable.
But the Board of Appeal had already begun its own inquiry.
The walls of the Board’s chambers were lined with shelves of decisions, dossiers, and doctrines. They knew of COMVIK, the precedent that loomed like a shadow over all hybrid inventions. And from within that labyrinth, they drew their blade.
They agreed with the Examining Division. The invention was nothing more than an automation of a managerial decision. The system monitored performance and imposed restrictions, yes—but that was merely a non-technical requirement, they said. A manager could do the same with pen and clipboard. Automating it did not make it technical.
“According to the COMVIK approach,” the Board wrote, “non-technical features, even if novel, cannot contribute to inventive step. The technical problem to be solved is merely how to implement these requirements on a conventional computer system.”
The inventor argued again. “But the system differentiates between operators and tasks. It builds a matrix. It adjusts dynamically based on individual performance. It avoids future failures by learning from the past. Surely that is technical?”
But the Board remained unmoved. “The restriction of work based on individual performance is still a managerial act. That it is individualized does not change its non-technical nature. The claim defines no technical manner of detecting poor performance, nor how the restriction is executed in a novel way. The technical features are merely standard computer functions. The implementation would have been obvious to the skilled person.”
Also the auxiliary requests were discarded like worn-out components:
- The first was dismissed without discussion, for lack of substantiation.
- The second, adding task-specific restriction, was ruled non-technical again.
- The third, with recursive performance counting, failed for the same reason: “a management decision can also be dynamic.”
- The fourth, a combination of the second and third, failed by association.
- The fifth, defining specific task types, was found to contain added matter and unclear language.
- The sixth, rich in detail, describing the mounting head, the nozzle stocker, the camera unit—was the last hope.
The Board reviewed it with care. The machine was now vividly described. The types of work were clearly technical: setting component order, configuring image data, selecting the correct suction nozzle. And yet the core remained the same.
“The claim does not define how lower performance is detected in relation to these tasks,” they wrote. “There is no direct link between the technical work and the restriction. The restriction remains a managerial decision, uninfluenced by the type of production device. Therefore, the invention lacks an inventive step (Article 56 EPC).”
And with that, the decision was final:
The appeal is dismissed.
The inventor sat alone, surrounded by blueprints and graphs, data tables and annotated flowcharts. The machine he had imagined still clicked and whirred behind him, indifferent to the news. It would continue to obey, to produce, to be monitored.
But it would never judge.
He felt not anger, nor sadness, but a peculiar numbness. As if the system had not merely rejected his application, but absorbed it—filed it, numbered it, buried it under layers of doctrine. Perhaps one day someone else would rediscover the idea, submit it in slightly different form, and receive a different answer.
Or perhaps not.
Guidance for the future? The inventor scribbled it himself, long after the Board had turned away:
- Do not mistake automation for technicality. A rule, even if digitized, remains a rule. A decision, even if made by a machine, remains managerial if its content is not technical.
- The COMVIK approach is not merely a hurdle—it is the architecture. One must isolate what is truly technical. Novelty is not enough.
- And above all: Never assume the system understands. The system processes. It does not feel. It does not care.
He signed the note simply:
“The Inventor.”
And then, he filed it away.
Based on T 2619/22 (Restricting low performers’ access to system/FUJI)