Systems and Models

Models of Systems §

• From the outside, the system appears opaque and we can only infer or speculate on how the system works through how it behaves through the use of models.

• From the inside, the system appears to be a part of a larger system. An observer from inside the system must be aware that it is prone to bias, and that it is also a component of the system using its subjective abilities to observe the system

• Information is the news of differences. It talks about facts in the system which are relevant for the specific analysis being conducted.

• Knowledge is the base of patterned expectation. We use knowledge to assess information.

• A crucial feedback loop is how information reinforces knowledge and knowledge gives more ways to interpret information. The tradeoff between flexibility and rigidity exists.

• Concepts are neural-based models of other systems that we observe.

  • More complex concepts are derived and constructed from simpler concepts. Neurologically, this involves forming links between neurons.
  • Concepts are both abstract and real. They are abstract because they are models, but they are also real because they are encoded in our neurons. Concepts are systems.

• A pattern is any set of components that stand in an organized relationship with one another from one system instance to another.

  • Spatial patterns exist when there is a consistent set of spatial relations between components.
  • Temporal patterns exist when there is a consistent temporal relation between components.
  • Maps are abstract representations that link between a pattern in one feature space to a pattern in another.
    • Two objects mapped to each other are isomorphic.
    • Maps are represented through mathematical relations.
    • Mappings are developed as one experiences associations in real life This is the essence of learning.
  • A property is a feature that constitute a pattern.
  • A characteristic is a property of an object that is always present in that kind of object.
    • The properties of an object are actually based on properties of the system itself or its subsystems
    • Properties exist in a hierarchy depending on the level of observation.
    • Properties are measurable. Measurement is comparing one system to another via the forces of interaction between them. Consistency assures us that measurements remain consistent and characteristics that are observed will be consistent.
    • Measurement operates not on “exact” instantaneous metrics but on differentials. That is we extrapolate information from measurements using differences between the compared systems.
  • Features are arrangements of component parts where such arrangements have properties that can be differentiated by measuring devices
    • Features come in hierarchies. Macro-level features can usually be decomposed into micro-level features.
  • Classes are groupings of features that imply a generalization.
    • Classification of groups of objects with similar features is a means of increasing the efficiency of mapping and, hence, pattern recognition and selection.
    • Classification means we are able to abstract away certain features that are shared by all objects of a particular class.

• Patterns are sets of relations that organize the set of features at any given level in the feature hierarchy, into a map.

• Repeated experience furnishes our fundamental means of recognizing the regularity of relationship that constitutes pattern or organization

Systems as Models §

• Systems are conceptual models for the world around us. They usually have a strong congruence with the world. However, they often fall short of representing the world fully

  • Systems fool us by presenting themselves as a series of events.
  • We are less likely to be surprised if we see how events accumulate into dynamic patterns of behavior.
  • Behavior based models are more useful than event based ones because they facilitate understanding of the underlying systems and have more predictive power. However, there are a few pitfalls
    • They overemphasize flows and underemphasize stocks.
    • In trying to find statistical links that relate flows to each other, we are trying to find something that does not exist.
    • The predictive power of such models are better for short-term rather than long-term performance.

• Systems feature non-linear relationships These nonlinearities make systems more difficult to understand.

  • These non-linearities can also change the relative strengths of the feedback loops.
  • Non-linearities can cause dominance shifts among the feedback loops of the system..

• Systems rarely have boundaries. They rarely exist within a closed system.

  • Systems as models simplify this by delineating a system in study with other systems that very likely do not influence the system in a significant way based on what is being studied.
  • In the end, there is no single legitimate boundary that can be drawn around a system.
  • Too narrow of a boundary can lead to surprises as external variables that do influence the system are not included in the model.
  • Too large of a boundary results in complicated analyses.
  • Framing Effect applies in this case as we may be prone to setting the boundaries based on existing boundaries.

• In systems, many causes can produce many effects Any physical entity with multiple inputs and outputs is surrounded by layers of limits.

  • At any given time, the input that is most important to a system is one that is most limiting.
  • Changes in limiting factors tend to be the most impactful.
  • As the system evolves, the input that is limiting also changes. One factor may become limiting or non-limiting in the future.
  • There will always be limits to growth. They can be self-imposed. If they aren’t they will be system imposed.

• Systems always feature delay. It is surprising how long things change in a system. Stocks are delays, and most flows have delays.

  • It is important to consider the scale of the delays that are important. Consider delays at the right level of granularity.
  • When there are long delays in feedback loops, some sort of foresight is essential. To act only when a problem becomes obvious is to miss an important opportunity to solve the system.

• Human-based systems typically feature bounded rationality. People act in their best interests based on the information they have even if it is to society’s detriment.

  • Perfect information is rarely present in a system.
  • We have our own share of faulty biases which render it difficult to act rationally.
  • One remedy to bounded rationality is to introduce more information.
  • The right feedback also helps actors behave rationally as they impose corrective behavior. Incentives and constraints also help.

Links §

• Shapiro, MacDonald and Greenlaw - Ch. 3.4
• Thinking in Systems by Meadows - Ch. 3 - 4
• Characterizing Systems
• System Dynamics