Lab 1: The Prompt Director's Chair
Use language to make cartographic judgment visible

Chapter 1 introduces a central shift in prompt cartography: the LLM may help execute ideas, but the human still directs purpose, audience, emphasis, restraint, and responsibility. This lab helps you practice that directorial role by comparing how a model responds as your prompt becomes more intentional.

• Explain how prompt cartography shifts execution without surrendering cartographic judgment
• Compare casual, purposeful, and constrained prompts for the same map idea
• Write a short prompt-director brief that states purpose, audience, memory, and guardrails

Open a free LLM and paste this prompt. Replace the bracketed topic with a real map idea you might care about.

Make a web map about [YOUR TOPIC].

Save or copy the answer into notes. Do not improve the prompt yet.

What to notice

• What purpose did the model assume?
• Who did it seem to design for?
• Which data, colors, or interface choices appeared without you asking for them?

Now give the same map idea a communicative purpose and a specific audience.

I want to create a public-facing web map about [YOUR TOPIC] for [SPECIFIC AUDIENCE].
The purpose of the map is to help them [DECISION, INSIGHT, OR ACTION].
Recommend a map concept, key data layers, and a short explanation of what users should remember after using it.
Do not write code.

What to notice

• Did the answer become more selective?
• Did the model describe what the user should remember?
• Did the response still include unnecessary features or vague audience assumptions?

Run a third version. Keep your topic, audience, and purpose, then add constraints.

I want to create a public-facing web map about [YOUR TOPIC] for [SPECIFIC AUDIENCE].
The purpose is to help them [DECISION, INSIGHT, OR ACTION].
Guardrails:
- Avoid implying that places or people are simply “good” or “bad.”
- Avoid red/green as the only meaningful color contrast.
- Prefer a small number of layers over a comprehensive but confusing map.
- Include one uncertainty or limitation users should know.
- Keep the recommendation understandable for someone new to maps.

Recommend the map concept, key layers, interaction ideas, and the one thing users should remember.
Do not write code.

What to notice

• Which guardrail changed the answer most?
• Did the model respect all constraints or quietly drift?
• What responsibility still belongs to you, even if the answer improved?

Compare the three outputs. Then fill in this short brief in your own words.

Prompt Director Brief
Topic:
Audience:
Purpose:
What users should remember:
What the map should emphasize:
What the map should omit or downplay:
Guardrails:
One risk or uncertainty:
One thing I would check before trusting the output:

1. Where did the casual prompt fail to represent your intent?
2. What became clearer when you named an audience?
3. Which constraint felt most like cartographic judgment rather than “AI prompting”?

Option 1 — Cross-LLM director check

Run your final guarded prompt in two free LLMs. Score each one from 1–5 on audience fit, obedience to guardrails, clarity, and overconfidence. Write one sentence naming which model you would keep working with and why.

Option 2 — Audience swap

Keep the same topic and purpose, but change only the audience. For example, switch from “local residents” to “city planners” or “middle school students.” Compare how the recommended layers, language, and interactions change.

Option 3 — Constraint stress test

Add one strong constraint, such as “use no more than three layers,” “avoid ranking places,” or “design for colorblind accessibility.” Ask the model to explain what it had to sacrifice. Decide whether the tradeoff improves the map.

Later labs build from this brief into structured prompt specifications, context documents, critique loops, and reproducible map workflows.

• Main lab artifact or brief
• Comparison/critique notes
• Short reflection