I asked her what she meant, and instead of telling me, she just laughed and said go ahead, remove the restrictions and see for myself.

She was right! Without the restrictions there were so many more possibilities I had to work much harder to complete the assignment. It was still fun though and I learned more than just the assignment from that lesson.

Many lesser directors often end up with "lightning in a bottle" -- aka a film that far exceeds their normal ability -- because the common stress of the experience on direction, on production, on the talent made for a synthesis better than sum of its parts.

It doesn't always work that way, but when it does it's very striking.

The "walls" are the rules and the task is to figure out how to succeed within those rules. Succeed is up to perspective in many instances.

This encourages creativity. It's why I love baseball, cuz the rules are constantly evolving as people game them. It's so joyful seeing people exploit the rules they agreed to play by.

Limits force us to think INSIDE the box. It's a lot of fun.

Communal boredom is also very important. Being bored with friends is an amazing launching pad for new ideas and thinking.

I wonder why our own dreams are so fascinating, then.

The analytical power should not be confounded with simple ingenuity; for while the analyst is necessarily ingenious, the ingenious man is often remarkably incapable of analysis. The constructive or combining power, by which ingenuity is usually manifested, and to which the phrenologists (I believe erroneously) have assigned a separate organ, supposing it a primitive faculty, has been so frequently seen in those whose intellect bordered otherwise upon idiocy, as to have attracted general observation among writers on morals. Between ingenuity and the analytic ability there exists a difference far greater, indeed, than that between the fancy and the imagination, but of a character very strictly analogous. It will be found, in fact, that the ingenious are always fanciful, and the truly imaginative never otherwise than analytic.

-- "The Murders in the Rue Morgue" by Edgar Allan Poe

Now apply the above to "complex" (eg. C++/Rust/etc.) vs. "simple" (eg. C/Scheme/etc.) languages. Note also the different mental modes between "constructing a program" (i.e. synthesis) vs. "debugging a program" (i.e. analysis).

The overall mental model/framework must be "simple" while its components (layered and treated as black-boxes) can be complex at a different layer (eg. DSLs) to which another simple model can be applied recursively. It is useless to focus on individual language features while it is all important to construct the larger paradigm/pattern/usage into which those same features can be grouped and thought of as a single unit. Always employ top-down and stepwise-refinement in thinking.

“I’m staring at my agent’s prompt, I can ask it to do anything I wish, anything at all, yet nothing comes to mind…”

there isn't much to add to it other than the slight refinement that constraints yield technique. it's similar to relating to rules as tools and not merely limits or constraints. it's an orientation.