Why Systems?

The world is made of interacting systems, and most people only see the surface symptoms of those systems.

Systems thinkers try to uncover hidden structures, connect data to decisions, identify root causes, and test assumptions in order to understand and improve the systems that are all around us

1. Real-world problems are messy

In school, problems are structured: clean datasets, known methods, clear goals.

In the real world, the system itself is unclear.

You often face:

• messy or incomplete data

• stakeholders with conflicting beliefs

• poorly defined problems

The real job is not running models—it’s figuring out what the problem actually is.

Systems insight:

Before optimizing anything, you must understand the system.

2. Symptoms are not the problem

People usually propose solutions to visible symptoms, not underlying causes.

Example from the slides:

• Client thinks slow deliveries → hire more drivers

• Data shows drivers wait hours at warehouse

• Root cause → broken sorting machine

Systems insight:

Problems typically emerge from interactions between parts of a system, not from isolated failures.

Fixing symptoms often makes the system worse.

3. Humans are unreliable sensors

Surveys capture opinions and beliefs—but people often misreport their own behavior.

Example:

• People claim they go to the gym frequently

• Swipe data shows they rarely do

Systems insight:

A system must be studied through observed behavior, not just stated intentions.

4. Data analysis is iterative, not linear

Real analysis looks like a loop:

1. Define the problem

2. Clean data

3. Explore patterns

4. Interpret findings

5. Discover flaws → repeat

You constantly uncover:

• bad assumptions

• missing variables

• misinterpreted patterns

Systems insight:

Understanding a system requires continuous refinement of your mental model.

5. Let the system reveal patterns

Analysts often begin with a theory and try to prove it.

But the correct process is the opposite:

• explore the data

• discover patterns

• then build explanations

The slides emphasize avoiding confirmation bias and letting the data guide hypotheses.

Systems insight:

Systems thinking requires curiosity before certainty.

6. Correlation isn’t causation

Systems often contain hidden variables.

Example:

• ice cream sales correlate with sunburn

• the true cause → summer heat

Systems insight:

Observed relationships may be driven by unseen forces within the system.

7. The real value is translating insights into action

Data alone has little value.

Value increases through stages:

1. Data

2. Information

3. Insight

4. Actionable recommendation

The analyst’s role is to translate complexity into decisions.