Limitations in Information Management

As mentioned in the chapter “On Intelligence,” a “rule-based system” is a computer system built under the cybernetic framework. The evolution from “rule-based systems” to “classical machine learning” and later to deep learning represents a transition from a cybernetic framework to connectionism. When analyzing traditional corporate organizations, it can be concluded that they resemble a composite system combining elements of “rule-based systems” and “classical machine learning.” In this system, managers (decision-makers) collect operational information to analyze and formulate execution rules and performance plans for grassroots employees (executors). Grassroots employees carry out tasks based on the rules and performance plans set by management. The cybernetic characteristics are evident in this system, with processes such as order management and process supervision being given relatively high importance. A lack of order management or process supervision may even be regarded as mismanagement or operational failure.

Due to the cybernetic nature of traditional corporate structures, when companies face increased scale, complex market environments, or rising competition—situations where information entropy increases—the difficulty of management rises exponentially. Studies on the operational efficiency of classical machine learning in relation to data volume and type show that its effectiveness plateaus after data reaches a certain threshold. However, since traditional companies are a hybrid of “rule-based systems” and “classical machine learning” decision-execution systems, their performance in scenarios with significantly increased data (and information entropy) may not only stagnate but could even regress significantly, manifesting as the commonly referred to “large organization disease.”

In the top-down structure of traditional companies, most organizational individuals (executors) typically have limited decision-making space. These small blocks of decision-making authority are like isolated black boxes, and a small number of decision-makers in specific positions usually manage their processes and authority. The decision-making blocks of the entire organization are analogous to the various functions in a software program. They have their own information input and output pipelines, but the intermediate processes are packaged. Lower-level executors take actions based on their output (execution paths). Higher-level decision-makers use the information obtained through the information pipeline transmitted from the bottom up to evaluate the quality of their decisions and issue further instructions for decision-making. (Based on the view of “Intelligenism Organization Setting”: Information is generated by executors and transmitted to decision-makers through information pipelines)