Staying in close touch with reality is imperative to succeeding in an environment of complexity. In order to continuously improve enterprise outcomes, a set of effective feedback cycles is required. These feedback loops must cut across multiple levels of the organization and, importantly, break across the organizational boundary to ensure that the enterprise is advancing customer value and its market position.
Despite its perceived simplicity, the topic of feedback cycles gets a lot more complicated as we enter into the real world of cognitive biases and organizational impediments of all sorts. These prevent the management and engineers from being able to adequately see reality and therefore make decisions that expose the organization to unreasonable risks. There are two key problems with feedback loops as they often appear in practice:
- The set of feedback loops is largely compartmentalized and incomplete, and…
- The existing feedback loops are largely dysfunctional
Let’s start with “1”. This problem is a typical result of anti-patterns such as Reductionist Mindset and Cargo-Cult Agile. As a result such flawed “divide-and-conquer” mentality, the system is perceived as a hierarchy of components, with Agile practices applied within the components, intended to produce system-level outcomes. Given such “adoption architecture”, there’s no surprise that feedback cycles are contained within components (teams, levels or the organization, etc.) and seldom cut across multiple levels. This often leads to various setbacks of operating in an open-loop system (see Semiconductor Organization anti-pattern). This is the reason why such a mindset is so dangerous.
Speaking of “2”. Even when instantiated, feedback cycles very often do not perform according to their purpose. They are simply not used as a learning mechanism to provide qualitatively new inputs but rather as a tool to indulging confirmation bias. See Feedback Loop Markers to learn how to approach this problem.