Skip to content
Create an account for full access.

Division of labor and system levels

Human roles, and now roles of AI, as well as roles of collective agents (organizations) in projects for the creation and development of various systems specialize at different system levels: recursively at each system level, an engineer's thinking is applied, who assembles a system acceptable to the suprasystem (higher level) - the target system (current level), using emergence from the interaction of collected subsystems (lower level). Each agent playing the role of such an engineer specializes in the emergence of a particular level, because their mastery cannot be infinite: closing mastery of two levels of equally high level is much more difficult, three even more. Moreover, for different systems (such as growing rapeseed and steel smelting), mastery in the methods of creating and developing these systems is significantly different in content, and one cannot quickly learn and master it to the level of fluent application. **The solution to this problem was the **division of labor: different agents specialize in different methods/ways of working, tied to rather narrow types of systems, which are quite narrowly distinguished at different system levels. Labor deals with some subject areas, and subjects are closely tied to system levels, because changing the system level causes the properties of subjects to change, thereby changing the work method for changing systems in order to achieve or remove the properties of that system level. The more knowledge about a subject area, the harder it is to master that subject area by one agent: thus, one subject area is defined as finer, certain types of systems are distinguished, within them certain additional system levels, within them certain individual emergent properties - and individual agents begin to specialize in achieving these emergent properties. This is called **deepening of labor division. For example, with the deepening of labor division, instead of one role of "engineer," a set of very different roles suddenly emerges - visionary, developer, architect, internal development platform engineer. Each of these roles has its preferences in areas of interest, so agents playing these roles will have to negotiate, and if in a very small project someone will perform all these roles together, one can doubt whether they have enough mastery to perform each of these roles well, as well as suspect a conflict of interests (many roles with conflicting interests in one subject area by one agent-performer of all roles. For example, in the "Systems Engineering" course, it will be more detailed explained what conflicts between the roles of visionary and developer, developer and architect, they are in a productive conflict. And for one engineer-agent, performing all these roles, there will inevitably be a conflict of interests).

Division of labor is everywhere. In the group of physical work methods, doctors treat body tissues, trainers create types of physical/motor behavior (types of movement methods) and their styles (variants of methods) and train adepts at the level of organism operation as a whole, cultural ambassadors care about cultural diversity of movements at the community level (a multitude of organisms that perform certain movements), and "body therapy" is basically about trying to solve mental problems through inducing the body into certain states (usually - relaxation and spontaneous movement). And we clearly didn't mention all the system levels in our example of social dances!

What we are talking about different roles, and different agents performing these different roles in relation to moving human bodies - this is division of labor. This division of labor continues at lower system levels and at higher levels. For example, geneticists can identify people with special cell metabolism, which allows for pre-selection of people resistant to oxygen deficiency for participation in competitions - they will not be limited where other athletes will struggle due to lack of oxygen in muscle cells during exercise. High-performance sports is increasingly turning into a competition of molecular biologists and geneticists nowadays, and competition among pharmacologists has also begun - there is a rise in sports with authorized doping, enhanced games, a technical solution to human organism limitations. Creating an athlete will require knowledge of molecular biology, genetics, and pharmacology, as well as coaching.

Changing properties of lower-level systems leads to changes in properties of upper-level systems - precisely through emergence. And every kind of emergence requires mastery to achieve these properties: understanding what affordances can be used as constructs to achieve the desired behavior, perform the desired function in the suprasystem.

Systems thinking is deeply connected with division of labor: at each system level, there are specialists skilled in achieving the system effect of that level - they achieve it by knowing how to assemble a system (target or "our," referring to very different systems) from its subsystems. But that's not enough, they also know where the assembled system can be used so that specialists at the higher level can use their system as a subsystem. Every agent as a qualified master-creator understands their system effect and links three levels of system breakdown with their work - and together, specialists can create and develop very, very complex target **systems and all the necessary elements for their creation and development described by the system creator graph.

This is true for engineering, but it is also true for other areas of activity, which are, in fact, also "engineering," although not entirely conventional. Nevertheless, we recommend thinking about them "engineer live," as it usually greatly simplifies understanding what is happening in projects creating non-technical systems.

For example, in the field of body practices (methods of movement in sports, yoga, Eastern martial arts, dances, stage movement, etc.), the engineering approach significantly accelerates understanding, eliminating numerous metaphors and artistic distractions from the essence, as well as providing normative/cultural/engineering division of labor. Just as hundreds of people today make a movie "the engineering way," many people (and even more broadly - agents, because not just people) can help turn someone into a full-fledged participant in a dance culture (in slang - turn someone into a good dancer, but we remember that a dancer is not yet an artist, and an artist is not yet a participant in a community practicing a certain subculture). At the same time, a dance performance turns out to be multi-level, its complexity can be overcome by using systems thinking in relation to it.

Labor is a synonym for the work method for creating and developing a system. Division of labor is a certain broad method divided into narrower methods/practices/"small types of work", while different agents of the project perform work according to different narrow methods of one broad method. For example, engineering work (broad method of work) is divided among different agents performing narrower methods that together make up engineering work - visioning, development, decision-making, internal development platform engineering (more details about engineering as a broad work method and narrower engineering methods will be explained in the "Systems Engineering" course).

Division of work is when work is divided among equally skilled agents working in some method (or even submethod) of work. For example, work needs to be done on painting a fence, 100 square meters. The method chosen - painting with a roller and paint, the role performed is "painter." Division of labor is when the role "painter" is separated from the general role "builder" and entrusted not to any builder, but to someone who knows how to paint. Division of work is when ten agents skilled in painting (with mastery of painting with a roller) are taken, and each agent paints not 100 square meters, but only 10. Only work is divided, and the method of this work remains the same, each agent performs the entire method/practice/activity, not just a part of it.

Division of labor is an evolutionary process. Work divided among agents is usually more efficient than the work of one agent, and one can simply observe how techno-evolution works, how with culture there is a deepening of labor division - labor division optimizes collective creators, the best solutions for dividing labor are remembered and then replicated among agents. But at the same time, the problem arises: how to unite divided labor in one general work of creating and developing the target system so that the system is successful?

Systems thinking allows uniting divided labor: it provides an understanding of why labor division occurs, how to explain the professionalization of agents in this or that "type of work"/"method of work." Thus, first of all, labor division occurs for work/methods related to objects of different system levels. If in an airplane we isolate the "engine as a black box," solutions for the transparent box will possibly be dealt with by different people, possibly at a lower system level, "engine specialists" - at each system level, some object features appear, new properties, often requiring a change in engineering role specialization.

If this happens at many levels with all sorts of systems at each of these levels, then we increase the chances of system success: we achieve not an ideal (it's impossible! Techno-evolution allows achieving only a quasi-optimum, a "local optimum"), but to some approximation of the ideal. Usually, it's already enough for us if the system is not ideal, but slightly more optimal than competitor systems ("to escape from a lion, you don't have to run faster than a lion, it's enough to run faster than the slowest person running from the lion").

What matters is that after creating and starting the operation of the initial system, everything else continues: continuous development of new system capabilities, new technical solution variants, implementation of new features, continuous deployment of them, "continuous everything." Before, requirements were "satisfied" once, now "system is created and continuously developed," often even by not one team's efforts, but by different teams specializing in creating and developing various system types. Division of labor helps cope with complexity: there is no need to pack all knowledge about the system and ways of working for its creation into the mastery of one agent (individual or collective).

Present-day dancers' or figure skaters' performances are immeasurably more complex and interesting than they were a hundred years ago, and learning dances or figure skating to a comparable level today can be much faster, because over a hundred years, the methods of creating and developing mastery have been significantly improved. Spaceships (which did not exist a hundred years ago), computer technology, software systems, medical robots - creating each of them requires many different kinds of agents' expertise, but with labor division, one can count on the fact that this expertise can be learned in a short period (assuming sufficient intelligence).

Division of labor associated with obtaining competencies in work on three adjacent system levels (understanding how and what to do at the current level of the target system, knowing why to do it and where and how to use the work - understanding the suprasystems, knowing how it is structured what to create it from - understanding the subsystems) allows building long chains of activities by various people, breaking down the overall super-complex activity at all system levels into many less complex ones at each system level.

However, it should be noted that at higher evolutionary/system/organizational levels of complexity than organizations/collectives of intellectual agents, with system descriptions and engineering relations, there are problems with labor division, and with describing the best methods of social engineering (in comparison with "letting things run their course," that is, relying on evolution). The frontier of systems thinking passes through the mastery of creating collective systems at the level of community and society. Dealing with system thinking in such projects should be done with caution, understanding its application in simpler systems. The level of personality is already extremely complex, an organization of many individuals is even more complex, but engineering copes with this. However, speaking about division of labor at the level of community, society, humanity - "agentism": including levels of society, community, and their AI agents in addition to humans, hence "agentism," not "humanity."