Angelov, Boyan. (2021, January 5). Homeostatic Project Management (Version 1.0). Zenodo. http://doi.org/10.5281/zenodo.4419707
Note: The foundation for some of the ideas developed here is laid by Stafford Beer in The Brain of the Firm and Raul Espejo in Organizational systems: Managing complexity with the viable system model.
In this blog post, I’ll explore an idea for the improvement of traditional project management. Let’s go through a few terms first (some of them are shown on the legend), and provide examples:
- Environment. Department, company, country - normally on a larger scale.
- System. The same as environment, but one scale lower (i.e a team or department)..
- Subsystem. A further scale lower, such as an individual contributor.
- Stressor. Software bugs, political issues, competition, finances, innovation lag.
- Management. Project manager, C-level.
- Sensor. It can be an individual contributor, or an automated process, detecting stressor impacts on the subsystem(s).
- Impact. Effect of a stressor on a subsystem.
- Variety. The complexity of an environment, system or subsystem, represented by its number of possible states.
- Black swan event. A statistically rare event, which becomes inevitable due to the law of large numbers; examples are the economic crises and wars.
Now that we got those terms out of the way let’s cover the traditional project management process (on the left side of the diagram). In this configuration, we have a system with a semi-permeable membrane, relatively open to environmental interference from stressors. Every subsystem has its own sensor, which is also stressor-specific. Those sensors transmit when there is an issue (effect) to the management, which then alleviates the stressor. This configuration is of a high variety, due to the many possible states. If a black swan event occurs, it’s effects might go undetected for some time since there is no suitable sensor. Thus no response occurs from management, potentially leading to the destruction of the system.
On the right side, on the other hand, we have a so-called “ultrastable” system. There are several notable differences. The most important one one is the “homeostat”, which effectively replaces the management layer. Instead of each subsystem dealing with the environmental stressors on its own and transmitting data to management (which can lead to information loss, as seen in Communication 2.0), the homeostat helps this system self-regulate. This is achieved by focusing on the only thing which matters: the daily operations’ cadence and quality.
If there’s a deviation in this metric, the homeostat sensor detects it (this measurement must frequently occur, i.e.in daily standups) and take regulatory/alleviation action. Simultaneously, the focus of subsystems is increased (they have lower variety) since they have full autonomy of their work - no potential interference from a bureaucratic layer. Another source of increased focus is the non-permeable membrane. Because the subsystems do not need to pay attention to the environment with their sensors, they can focus on the task at hand. This helps the ultrastable system to keep its total variety low. Even a black swan event can be potentially dealt with on time because the homeostat will detect it. In an optimal case, the homeostat sensor’s role can be filled by the project manager who responsible for setting the metric, measurement, and potential real-time adjustments. We can call this improved way to do project management “homeostatic”.
So why are we not seeing such systems implemented everywhere? This is probably due to the difficulty in creating a measurable metric for the homeostat. In non-knowledge work, such as a factory, it is relatively easy to develop such a metric - it could simply be the system’s productive output. In knowledge work, however, this becomes more abstract. Potential metrics can include issues closed, features complete, code quality, production model performance, automation gain. Setting such a metric requires a lot of preliminary work, and high variety management.
Some of those ideas might sound a bit abstract, especially if you are new to the fields of systems science and operations research. I’ll be expanding on this idea and adding some case studies in future posts.