Methods · Creation, transfer & long horizon
F19
Constraint invention, including the playful kind
Invent restrictions on your own work that force you to understand it better than doing it the easy way ever would.
When to use it
- Easy versions of the task stop teaching you anything.
- You want restrictions that force deeper understanding of the material.
- You are ready to invent the brief, not only obey one.
What to ask AI
Type something like: “I'm going to explain [concept] to you. First, invent three constraints that would make this genuinely hard for me, one practical, one structural, one absurd. Don't explain the concept. Then make me do it under each.”
Ask for constraints that force contact with the real ideas, not pure avoidance.
Keep the hard part on your side: you perform under the constraints.
How it helps
Constraint invention means inventing restrictions on your own work that force you to understand it better than the easy path allows.
The playful kind still has to touch the real idea, silly is allowed; empty is not.
Clever output with nothing learned means the constraint entertained you without changing your grasp.
How to run it
- Name the concept or task.
- Ask for three constraints: practical, structural, absurd, without an explanation of the content.
- Perform the task under each constraint yourself.
- Afterward, say what each constraint revealed about the idea.
- Keep the one constraint that taught the most for a serious revision.
Example exchange
I'm going to explain natural selection. Invent three constraints that make this hard: one practical, one structural, one absurd. Don't explain the concept for me.
Practical: eighty words and no biology jargon. Structural: you may not use the words fitness or adaptation. Absurd: explain it as instructions for a board game. Do the practical version first.
In plain words: living things differ, some leave more offspring, and those differences can add up over generations if they are passed on.
You smuggled fit through suited. Retry the practical version. What does the structural ban force you to say instead of adaptation?
Copyable prompt
I'm going to explain [CONCEPT] to you. First, invent three constraints that would make this
genuinely hard for me, one practical, one structural, one absurd. Don't explain the concept.
Then make me do it under each. Afterward ask what I learned about the idea, not whether it was clever.
The Tell
If you produced something clever and learned nothing, the constraint was entertainment.
Ask what you can now say about the idea that you could not say before, if the answer is nothing, invent a harder constraint that touches the mechanism.
Principle evidence
Strength of the underlying learning idea, not a claim about AI products.
The underlying learning idea is rated weak. Inventing one's own constraints adds metacognitive design to constraint-driven creation. Plausible for engagement and deeper processing; empirical base is thinner than for assigned constraints.
AI delivery evidence
Whether an AI tutor delivers this method well is a separate question.
AI-invented constraints vary wildly in usefulness. You must reject ones that avoid the idea. Delivery evidence is speculative.