Feedback loops in system thinking

I’ve written recently about solving problems by using system thinking. It was pretty easy. Right now it will be more difficult — we will read about the feedback loops in system thinking, in particular about the reinforcing and balancing loops. This article should have been the first one, but let it be a sequel.

Systems thinking is not linear, it happens not in a straight line — it happens in cycles, loops, contours. All parts of the system are connected, so a change in one part generates waves of changes that reach all other parts. Thus, the action returns to the starting point in a modified form: so we have a loop. It is called a feedback loop. All our experience is formed as a result of such actions. The phrase “feedback” is often used to designate any reaction, but in reality, it means the perception of the result of our actions affecting subsequent actions, i.e. two-way communication. Our experience is formed as a result of actions of this kind of feedback loops, although we are used to seeing the only one-way influence. The presence of feedback is an integral characteristic of a system. No feedback means no system.
There are two main types of feedback:

• Reinforcing feedback: a change in system state which serves as a signal to enhance the initial change. In other words, the system provides a big change in the same direction.
• Balancing feedback: a change in system state which serves as a signal to start moving in the opposite direction in order to restore the lost balance.

You take an action, get a reward and repeat the action — here’s how the feedback loop works.
A reinforcing loop is one in which an action leads to a result that affects the same action, which leads to positive or negative progress of this action. But eternal growth is impossible. Eventually, the second type of feedback comes into play and stops the growth. It is called balancing (opposing changes) feedback. This form of feedback opposes change and maintains the stability of a system that would otherwise be destroyed by the action of reinforcing feedback. Balancing feedback directs the system to the goal, to such a state when the balancing feedback mechanism is turned off. When the system approaches the goal, the balancing feedback weakens, the system changes more slowly; and vice versa — when balancing feedback is removed, it becomes stronger, and the system changes quicker. Balancing loop helps the system to maintain balance.
Even easier: effect of the balancing loop is the pursuit of a specific goal, the effect of the reinforcing one is a constant progressive effect (both in positive and negative).

Yes, I promised to talk about IT. Let’s look at an example of how this works for us. So, in the IT project life cycle, there are four typical feedback loops (examples taken from Diti Das “Systems Thinking — A Process For Problem Solving in SDLC”). Product development loop (Balancing Loop), Product Management Loop (Reinforcing Loop), Employee Productivity Loop (Balancing Loop), and Product Maintenance Loop (Reinforcing Loop).
For example, the balancing Product Development loop describes that when there is an increase in customer requirements, the need for restructuring the current software functionality and design increase. The Reinforcing Product Management loop (Positive) suggests that quick management actions should be taken to implement changes in project schedule and prioritization. The reinforcing Employee Productivity loop (Negative) indicates if there is an urgent customer requirement and the project progress falls behind schedule, the team needs to work overtime to satisfy customer deadlines. The Product Support & Maintenance loop (Reinforcing Loop) shows the post product development support & maintenance activities that involve deploying the software to meet customer requirements.

Actually, why I have started this topic at all. Feedback loops are a powerful tool in the manager’s hands. Let me remind you that the initial change in the variable (process, etc) stimulates its further change in the original direction. Thus, if we succeed in changing the variable in the direction we need (reinforcing loop), we can start the process throughout the whole context, and since the variables enter several contexts (aka contours) at once, we can launch the same series of cascade effects that will now work for us. Just remember that all systems are endowed with a balancing feedback mechanism that ensures their stability. But — in order for balancing feedback to work, measurement is necessary (f.e., to define when should we switch to balancing loop). This measurement must be accurate enough for the feedback to work adequately.

The next time you solve a really difficult problem, try to depict cause-and-effect relationships by connecting them in loops. You may do this rather quickly or you may need a few days. Eventually, you will find that you have made a logical diagram that allows you to see the system, or at least most of it (several contexts/contours). You may want to involve the team in the discussion or to create a mental model (for example, mind map). Anyway, this diagram, even a schematic one, will be a powerful push to making the right decision.