Skip to content
Get in touch

Practical systems modeling

From a model to explainable work.

Keep the question connected to the behavior someone can inspect.

Trace one engineering decision

This worked pantry example reveals the final records. It does not complete your exercise.

  1. Question: How many distinct people participated in each role, and overall, during September 14–20?
  2. Modeling choice: Keep one Person identity connected to volunteer participation and distributions collected. Household remains a separate thing.
  3. Records: Retain who participated, what occurred, when, and the relevant household and source record.
  4. Behavior: Inspect a person or a total to see the records behind it.
  5. Validation: Ask the pantry what recipient means, what completes a distribution, which period applies, and how identity is established.

Roles alone do not prove activity during a period. This fictional report counts the people collecting food, not everyone in their households.

Open a total. See its support.

Ana · one person, both roles

V1: volunteered September 14. D1 and D4: collected for H1 on September 14 and 18. Noor confirms that these records refer to the same Ana; a matching name alone would not establish identity.

2 volunteers · 3 participations

Ana: V1. Ben: V2, V3. Two distinct people across three recorded activities.

3 recipient people · 4 distributions

Ana: D1, D4. Cora: D2. Dev: D3. Repeat collections do not create new people.

3 households

H1: D1, D4. H2: D2. H3: D3. Four distributions served three households.

1 person in both roles · 4 unique people

Ana appears in both groups. Combine Ana, Ben, Cora, and Dev once each. Adding 2 volunteers to 3 recipients would count Ana twice.

Leaving Ana out of a role she performed understates activity. Assigning Ben a recipient role unsupported by these records overstates it. Errors like these can lead to audit failure. This exercise demonstrates traceable counts; it does not certify audit compliance.

Your workshop notes

Reading a room or example is exploration. A completed capstone includes your own evidence-linked revision. Free-text reflection is preserved without automated grading.

Read my saved record
No interactive notes loaded.

Carry these questions into your work.

What is the purpose? Which identities must stay stable? What happened, and when? Which claims have sources? What does the new evidence change? What should we confirm with the people doing the work?

An engineer and AI need the same explicit story. Keep the sources and context with the model so speed does not erase the reasons behind the design.