In this video, Professor Donella Meadows explains the philosophical aspects of system dynamics modeling. We try to unpack the core assumptions of system dynamics, emphasizing that a model is fundamentally a set of assumptions. In particular, it focuses on understanding how complex systems work through causal relationships, feedback loops, and the concepts of speed and level.

1. The essence of system dynamics

Professor Donella Meadows explains the philosophical aspects of the System Dynamics methodology, emphasizing that it is very difficult but important to explore the core assumptions of the field. Just as we see the world through an eye lens without being able to see it directly, system dynamics thinkers are often unaware of the basic assumptions they use. 🤯

She explains that system dynamics is a philosophy for learning about complex systems, with a particular focus on how systems change over time (Dynamics).

"System dynamics is called 'dynamics' because it deals with how systems change over time."

In addition, system dynamics (System) is said to be interested in the whole system rather than the characteristics of individual elements, and helps to explore the behavior of the entire system using a computer as a tool by embodying system philosophy.

2. Importance of Causal Linkages

Professor Meadows identifies Causal Linkages as the first important aspect of system dynamics. System dynamics models strive to include true causal relationships between elements, emphasizing that they should be able to explain the actual mechanisms by which one element causes changes in another element.

For example, in a deer model, it is a very obvious causal relationship that birth rate affects population, and this relationship should be intuitively easy to understand.

"We try to include in our models relationships that we think are truly causal. That is, when we draw an arrow from one element to another, we think we can describe the real mechanism by which element A changes element B in the real world."

She explains that a distinction must be made between this causal approach and a simple correlational approach (Correlational Approach). Correlation means that two factors can change together, but one does not directly cause the other. For example, there is a correlation between the number of doctors per 100,000 people and per capita plant-based calorie intake, but in reality, third factors such as 'industrial development' can affect both. 💡

Additionally, Nonlinearity and Delays are important considerations in system dynamics models.

• Non-linearity: Many relationships in the world are not linear, and can have a large impact initially, but then have less impact after a certain point. For example, increased food or medical care does not increase life expectancy indefinitely.
• Delay: Changes within a system are not immediate and often take time. It takes time to increase food production through agricultural investments or to increase life expectancy through building hospitals. This delay is essential to accurately understand the dynamics of the system.

3. Discovery of Feedback Loops

Professor Meadows selects Feedback Loops as one of the most important concepts in system dynamics. She explains that wherever you look you find feedback loops, and that the belief that the world is made up of closed causal chains is at the heart of system dynamics.

"I find feedback loops everywhere. This is again part of system dynamics. We expect the world to be made up of closed causal chains, and so we find them."

Feedback loops can be divided into two basic forms:

3.1. Negative Feedback Loop

• Characteristics: Tend to maintain balance and move toward a goal. When a change occurs, it acts to offset the change and stabilize the system.
• Example: A situation where beer is poured into a beer glass. When there is more beer in the glass, close the tap and stop when the target volume is reached. This leads to goal-directed behavior. 🍻

3.2. Positive Feedback Loop

• Characteristics: Tends to amplify and self-reinforce change. Once a change begins, it continues to lead the system to grow or shrink.
• Example: When interest is charged on bank deposits. The larger the balance, the more interest accrues, and the more interest accrues, the faster the balance grows. This causes exponential growth or decline. 💰

Professor Meadows emphasizes that real-world systems are complex with multiple feedback loops. Using an inherited oil reserves model as an example, we illustrate the complex interplay between exploration costs, oil prices, investments, and discoveries, showing how this loop is self-reinforcing through positive feedback.

"Whenever someone says, 'A causes B,' I always ask myself, 'So how does B affect A again?' This is ingrained in my system."

She emphasizes that system dynamics models are state-determined systems. In other words, state variables within the system drive changes in the system. Therefore, we advise that when a problem occurs, the cause should be found in the internal structure of the system rather than external factors. However, he adds that not all systems fit this model, so it is important to have the wisdom to understand the nature of the problem.

4. Levels and Rates

The two basic components of a feedback loop in system dynamics are Levels and Rates.

• Levels: Measurable elements that represent the state of the system. People make decisions based on this level. (e.g. beer in a glass, bank balance, unidentified oil reserves)
• Rates: Actions or decisions that cause a level of change. (e.g. how much you open the tap, interest on deposits, oil exploration rate)

These concepts are represented visually in a Dynamo Flow Diagram.

• Level: Displayed as a square box. 📦
• Speed: Displayed as a valve shape. 🚰

The concept of Auxiliaries also appears, which are intermediate variables used when expressing the speed equation by dividing it into several steps, and are shown as circles in the diagram. 🔵

Using an oil exploration model as an example, Professor Meadows shows in detail how unproven reserves ('level') are linked to exploration costs, oil prices, investment, and discovery rates ('rate' and 'auxiliary variables'). Here too, non-linear relationships or delays may be important factors.

5. Relationship between structure and behavior

The most important value of system dynamics is that we gain knowledge about how the structure of interrelationships generates dynamic behavior.

• Positive feedback loop: Causes either exponential growth or exponential decline. (e.g. interest-bearing bank balance) 📈
• Negative feedback loop: triggers goal-oriented behavior (goal-seeking fashion), causing the system to converge to a certain goal level. (e.g. pouring beer into a beer glass) 🎯

Professor Meadows explains that system dynamics experts have an intuitive understanding of the relationship between 'Structure' and 'Behavior', which is an important clue to modeling the complex real world and solving problems. For example, if you see a system growing exponentially, you might assume that there is a self-reinforcing positive feedback loop within it. Conversely, if you look at a stalled system, you might think that a strong negative feedback loop would keep pulling the system toward its goal. 🧠

Conclusion

Professor Meadows emphasizes that as systems become more complex, intuitive understanding alone is not enough and computer models must be used, but these basic principles serve as important guides for modelers to understand the world and build models. Next time we will cover how these concepts are translated into mathematics and computer programming languages.