Hands-on Labs

Reading about computational thinking is like reading about swimming. These labs get you in the water.

Each lab is a guided, runnable exercise built around one design rule: every lab fuses two or more CT concepts, because that is how the concepts occur in real work. Nobody hands you a “decomposition task” — you get a messy problem, and decomposition, pattern recognition, and evaluation happen together or not at all.

The labs

Lab	Fused concepts	You will build
Lab 1: Support-Ticket Triage Pipeline	Decomposition + pattern recognition + algorithm design	A rule-based text classifier, measured and iterated against labeled data
Lab 2: Simulate an Epidemic	Abstraction + modeling & simulation + systems thinking	An agent-based SIR simulation exposing feedback loops and tipping points
Lab 3: Trust but Verify	Verification thinking + evaluation & debugging + data thinking	A harness that checks LLM-style outputs against invariants and golden cases

How to use these labs

Requirements: Python 3.9+ and nothing else. Every lab uses only the standard library — no installs, no API keys, no accounts. If python3 --version works in your terminal, you are ready.
Type, don’t paste — at least once. The friction is the learning. Paste on your second pass if you like.
Run after every step. Each lab is staged so the program runs (and tells you something) at every checkpoint. If a step’s output surprises you, stop and figure out why — that surprise is the lab working as intended.
Do the “iterate” steps. The first working version is the beginning of each lab, not the end. The CT skill being trained is the loop: measure, diagnose, change one thing, measure again.
Predict before you run. Before each run, say out loud what you expect. Prediction errors are where models — yours, not the computer’s — get repaired.
Expected time: 45–90 minutes per lab. They are independent; do them in any order, though Lab 1 is the gentlest on-ramp.

Working with an AI assistant? Good — that mirrors real practice. But apply the scaffolding rule from the research literature: make the AI explain, then verify its claims by running the code. Never accept a step you could not defend to a colleague.

When a lab leaves you wanting the underlying theory, the concept pages in Foundations, CT in the AI Era, and Data & Systems Thinking are the reference shelf.