Second part 2. The bottleneck of responsibility. When data is transferred to the information system, then someone must make a decision. But if the decision concerns something significant, for example punishments (fines, punishments, etc.), then it must be accepted by people, and these people are few. A bottleneck appears. To remove it, system administrators have an instinctive desire to give decision-making to robots, to automate everything, to absolve themselves of responsibility. But this is direct discrimination against citizens; They will inevitably fight, because they have their own agency and interests. 3. Loss of precision, work on areas. The third problem is that life situations are infinitely varied. It is impossible to describe all this diversity and respond differently to each one. Therefore, different situations in the ontology, in the "situation catalog" of those who create control systems, are combined into groups, i.e. enlarged. It turns out to be a paradox: first, digitalization reaches every small element, and then they still need to be grouped. But sometimes these groups turn out to be strange, built on wrong assumptions, unreliable or distorted data. Then the effects exerted on each group may be harmful or wrong. 4. Excess data. When building large digital systems, the default assumption is that the more data, the better. In fact, excess data is a nuisance for digital systems and their developers. If there is too much data, then the systems are "choked" with it, they cannot draw the right conclusions and they start to fail. For example, the US attacks on Iran this year are believed to have been planned using data collected by the Palantir system. The analysis of the data made it possible to plan the first attack against Iran's top management. Then a real high intensity war began, the enemy did not respond to the attacks as expected. And the system is simply choked with huge data in real time.