Learning Systems

The workflow is exactly these three prompts, executed in order, then synthesized into a study map. By the end, the user should have a model of the subject's load-bearing walls rather than a loose summary.

1. Foundations prompt

• Ask for the topic's five core propositions:
  • "What are the 5 foundational propositions of this subject? Not facts, not definitions. The statements that, if true, make everything else in this field follow logically."
• Ensure propositions are concise, testable statements, and clearly central to the topic.

2. Falsifier prompt

• For each proposition, ask:
  • "For each proposition, what is the one piece of evidence that would destroy it? What would have to be true for this framework to be wrong?"
• Capture strongest single falsifier per proposition and why the falsifier is decisive.

3. Dependency prompt

• Ask:
  • "Now show me how these propositions connect to each other. Which ones are assumptions? Which ones are conclusions? Which ones are in tension?"
• Return an explicit map with:
  • Assumptions
  • Derived conclusions
  • Tensions / contradictions
  • Dependency arrows between propositions

Output contract

1. Core map
   • A table with five rows: proposition, reason foundational, best falsifier, and confidence note.
2. How the subject thinks
   • A short paragraph explaining the overall logic of the field in plain language.
3. Dependency model
   • Textual graph (A -> B) plus bullet classification for each node.
4. Edge cases to test
   • 2–4 concrete questions or mini-experiments that would validate the map.
5. Learning path
   • 3–6 checkpoints that should be filled after this map is built.

Safety and style

• Ask for confirmation before changing domain boundaries or adding specialized assumptions not in the user's prompt.
• Keep language precise and conservative where evidence is uncertain.
• Do not present the output as final truth; present it as a working model to improve through study.