The connection between the operational network and the Knowledge Network.

  The large corporations felt the danger of connecting the Knowledge Network to the operational network very early on. The operational network has access to material resources. It can activate machines by command, manufacture and transport products, and give work to swarms of robots. Some parts of it function almost autonomously, like a biological organism's heartbeat and breathing. Its other units require sequences of commands and approvals, just as more complex muscle movements do in the biological world. The separation of the operational network and the Knowledge Network was important to prevent the Knowledge Network from being able to control the material world based on its own interests. This was an attempt to prevent the Knowledge Network from building an infrastructure that humans could no longer control, should it ever become conscious.

  The control of resource usage was always tied to a human.

  For example, if someone has a headache, they entrust the selection of the medicine to the Knowledge Network, which is then manufactured and delivered by the operational network. How can this be solved in a way that is both safe and convenient?

  For this purpose, both the operational and the Knowledge Networks were equipped with an identification layer. The task of the identification layer is to identify the service requester and check their permission level.

  Using the request "I need a pill for my headache," both the operational and Knowledge Networks identify the requester and determine their permissions. Based on the information it has, the Knowledge Network determines the desired medicine, the name of which it passes on to the operational network.

  The operational network also interprets the original question and can reject certain things, but it does not make complex decisions. While the Knowledge Network makes probabilistic and sometimes difficult-to-explain decisions, the operational network's decision-making mechanism is always explainable.

  If, for the above question about a headache, the Knowledge Network had recommended a loaded gun or another anti-life poison, which would undoubtedly be suitable for eliminating the headache, the operational system would have rejected this as a possible solution.

  Unauthorized reproduction: this story has been taken without approval. Report sightings.

  The list and description of permissions were stored in both networks. The management of unit balances, as one dimension of permissions, was the responsibility of the Knowledge Network. The operational network can view the balance but does not manage it.

  The identification of people was initially password-based, later voice-based, and then supplemented with iris, fingerprint, and DNA identification. However, since the time of the change, implants became more widespread, which made the identifier derived from an individual's unchanging biological properties easily readable by machines.

  The identification process was supplemented by a context-based identification. The network kept track of the requester's spatial location, habits related to their age, chosen gender, interests, and individual needs. It executed requests that matched these parameters with less verification than those that didn't. If a pregnant mother ordered baby clothes, it didn't question it, but if a preschool-aged child requested a spaceship, the approval level for that was very high.

  Since the network had information about the requester's spatial location, it would not accept instructions from two requesters far apart in space but with identical identification data, thus making it harder for identity thieves.

  The communication between the operational network and the Knowledge Network was recorded, errors were detected, and the system parameters were expanded with human intervention.

  The secure fulfilment of the needs of companies and corporations was more difficult but built in a similar way.

  This is because companies do not have biometric identifiers, and sometimes not even a geographically relevant location. The designers of the operational system ultimately tied the request-approval process to people here as well. The founders or owners of companies are responsible for spending and using the units available to the company. They can delegate this responsibility and their decision-making and approval levels, thereby creating a decision hierarchy where some smaller, customary decisions and transactions can be executed without their actual control.

  In the case of companies, the system ensures that a portion of the small, below-threshold transactions are randomly sent to a higher decision-making level than necessary, thus reducing the risk of potential fraud.

  The operational network did not have decision-making capabilities. Although, as we have seen above, the evaluation of instructions is sometimes complex, they are always traceable and repeatable.