Subordination ​

In addition to monitoring, that is, observing important objects, their qualitative and quantitative characteristics (metrics), there is also controllability/accountability/governance. It differs from simple monitoring/tracking in that it involves not only observation but also intervention in situations that are not going well. For example, not just seeing that the soup is boiling over, but also taking it off the stove. Not only noticing that employees are doing what they want, rather than what the company needs, but also working on improving their performance discipline.

Thanks to controllability, an agent can align their perceptions of the world (model, worldview) with reality (physical world), and vice versa. This is achieved through interrupt-driven management or polling-driven management.

Interrupt-driven management means that attention is focused and maintained "as usual," and only when a problem arises, a signal about the change in state is sent and awareness takes place. Whereas polling-driven management means that attention is periodically directed to objects of interest (projects, issues, reports, and so on) and "snapshots" of their states are taken. Interrupt-driven management allows for speed (there is no need to spend many resources on polls), while polling-driven management ensures accuracy.

To optimally address issues (so that the agent acts more often in the zone of maximum optimality), awareness uses both methods, choosing the most suitable one depending on the situation. The "hardware" interrupt on danger signals is inherent to agents (whether humans, animals, or AI): humans and animals react to danger, AI signals problems that can disrupt its operation, providing people working with it awareness of issues and the need to address them. Polling-driven management in activities is more "software-based"; people have developed it by implementing various projects: developing drugs, launching rockets into space, organizing festivals, and dance battles. It is somewhat more expensive in terms of resources (constantly and rhythmically directing cameras, using their limited computational resources), but it allows for accuracy and the ability to anticipate problems.

How to use interrupt-driven and polling-driven management in activities? In TameFlow, it is suggested to combine them as follows: by creating buffers (buffers for team tasks, MOVE buffers, etc.). The buffer is divided into three zones, marked with corresponding colors:

• Green: everything is fine, work continues without interruptions.
• Yellow: a signal of possible delay has been received, work continues without interruptions. The "Yellow" signal usually indicates systematic deviations related to how the system operates (if you want to have fewer of them, you need to change the system, i.e. conduct long-term optimization).
• Red: intervention is needed and changes need to be made, for example, add work to the constraint buffer or, if it is the critical chain buffer, allocate additional resources (idle organizational units). Red indicates a deviation for a specific reason (a "fire" has been detected), the focus should be on rectifying it (short-term optimization).
• Black: a force majeure has occurred, information needs to be urgently communicated to top managers with the authority to resolve the situation. After resolving the conflict, its causes need to be identified and efforts should be made to change the organization's operations to prevent similar situations in the future.

This essentially means that as long as the buffer is "green" or "yellow," interrupt-driven management is applied. But as soon as the buffer turns "red," a switch to polling-driven management occurs, actively intervening with monitoring what is happening with the problem and its solution. This helps avoid the imposition of artificial rhythms/cadences typical of Scrum and Kanban, where there is a forced switch to polling-driven management without considering the signals of what is happening. Stable execution of work over long time scales, without a "reset" of statistics from sprint to sprint, allows for greater productivity through cumulative effects without unnecessary effort and resource expenditure.

Controllability is also important for organizing quality feedback. You can choose the frequency of project meetings by paying attention to the frequency of significant issues that require a meeting for resolution (polling-driven management). Also, establish rules for emergency communication separately: defining what situations are considered "emergency," when intervention is necessary, and which communication channels to use (interrupt-driven management).