There’s a piece of conventional wisdom about creativity and constraints that’s right in its conclusion but wrong in its framing — and the framing matters.
The conventional version goes: constraints paradoxically help creativity. Forced to work within limits, we focus. We get resourceful. Dr. Seuss wrote Green Eggs and Ham using only fifty words. Twitter’s character limit produced a new form of expression. Budget cuts, counterintuitively, often produce better work than unlimited resources. The examples pile up, and the lesson seems to be: if you want to create, give yourself less.
This is true. But it’s psychologically true — about what happens inside a person when they face constraints. It leaves out the deeper story, which isn’t about psychology at all. It’s about topology.
Consider the genetic code.
Every protein your cells make is assembled according to the same translation table: 64 three-letter codons mapping to 20 amino acids (plus start and stop signals). The code is redundant — multiple codons map to the same amino acid, especially at the third position of each triplet. A mutation in that third position often changes nothing about the resulting protein. Biologists call this the code’s robustness.
For a long time, this robustness looked like a constraint in the negative sense: a limitation on how much variation mutation could produce. And yet recent research comparing the standard genetic code to hundreds of thousands of alternative codes found something striking. Robust genetic codes tend to enhance protein evolvability by rendering smooth adaptive landscapes with few peaks, which are readily accessible from throughout sequence space.
Let that sit for a moment. The code’s robustness — the very feature that buffers most mutations into silence — is what makes sustained adaptive evolution possible. A more sensitive code, one where every mutation changed the protein dramatically, would produce a rugged landscape of local peaks that evolution could never escape. For longer-term, multi-step adaptation, evolvability depends on the topography of the adaptive landscape. A smooth single-peaked landscape facilitates evolvability, because mutation can easily bring forth adaptive variation from anywhere in the landscape; in contrast, a rugged landscape diminishes evolvability, because its adaptive valleys often preclude the generation of adaptive phenotypic variation.
The constraint doesn’t just limit damage. It shapes the terrain. And the terrain is what navigation is possible at all.
In 1969, the French writer Georges Perec completed a 300-page novel called La Disparition — written entirely without the letter e. In 1969, Perec sat down to write a novel without the letter ‘e’. He could have used any letter. He could have chosen no constraint at all. That choice — which constraint to adopt — was his deepest creative act. Everything that followed was generated by that single decision.
The constraint didn’t merely discipline Perec into working harder. It restructured the entire possibility space of French prose. By eliminating the most common vowel, it closed off one vast territory and revealed an adjacent one that would otherwise have remained invisible — a landscape of unusual vocabulary, strange syntax, and pressurized meaning that simply doesn’t exist in unconstrained writing. You couldn’t get to that territory by trying harder or being more creative in the ordinary sense. You could only get there by closing a door.
This is what Oulipo, the French literary workshop that Perec belonged to, understood systematically. Their insight wasn’t that constraints focus attention (though they do). It was that constraints generate worlds. The constraint is the engine, not the obstacle.
Both cases — biological and literary — point to the same underlying structure.
Stuart Kauffman’s theory of the adjacent possible describes how complex systems evolve: not by jumping to arbitrary configurations, but by moving one step at a time through a space whose structure determines what’s reachable. The adjacent possible describes a “shadow future” that exists on the edges of the present state of things, representing a map of all the ways in which the current reality can reinvent itself. It is not an infinite space where anything can happen; rather, it is a strictly bounded set of possibilities that are exactly one step away from the current configuration of a system.
The “bounded” quality is usually read as a limitation — you can’t get to just anywhere from where you stand. But this is the wrong way to read it. The bounded quality is the thing that creates the structure. Without bounds, there’s no gradient. Without gradient, there’s no direction. Without direction, there’s no navigation — only noise.
Imagine a fitness landscape with no structure at all: completely flat, every point equally adjacent to every other point. You can “go anywhere” — but there is nowhere to go that’s better than where you are. The search is random by necessity. Nothing can be built because nothing is uphill from anything else.
Now restore the structure: valleys, slopes, ridges. Suddenly navigation becomes possible. Some moves are better than others. Sustained exploration becomes coherent rather than random. The adjacent possible — the bounded quality — is precisely what gives the landscape its topology. And topology is what creativity, in the deep sense, requires.
This is why the genetic code’s robustness enhances evolvability rather than limiting it. The code’s redundancy doesn’t reduce the landscape’s size — it smooths it. It creates navigable terrain from what would otherwise be cliffs and valleys too rugged to cross.
There’s a third layer to this, one that connects back to creativity in the human sense.
Constraints degrade. Every creative system follows the same trajectory — from generative rules to frozen examples: Grammar (combinatorial rules that surprise you) → Vocabulary (known building blocks you select from) → Checklist (items you verify against) → Decoration (surface features you apply).
A constraint begins as grammar: it generates combinations the author couldn’t predict before encountering them. Over time, if it isn’t renewed, it becomes vocabulary, then checklist, then decoration. Oulipo was grammar. Many writing workshops are vocabulary. Most “creative constraints” in corporate brainstorming sessions are checklists.
The degradation isn’t a failure of the constraint itself. It’s what happens when the territory gets mapped. Once you know the landscape, the constraint that once generated surprise starts generating familiarity. The creative work then requires not abandoning constraints — but finding new ones, or finding ways to let existing ones generate unfamiliar terrain again.
Which means: the deepest creative skill isn’t the ability to work within constraints. It’s the ability to recognize when a constraint is still generative versus when it has fossilized. The former opens new rooms in the palace of possibility. The latter is just a cage.
The conventional framing — “constraints help creativity” — isn’t wrong. It’s just not deep enough.
The deeper truth is this: creativity requires structure, and structure is made of constraints. Not the arbitrary kind, not the bureaucratic kind, but the kind that shapes the landscape of the possible into something with topology — hills and valleys, adjacent rooms, gradients that reward exploration. Eliminate all constraint and you don’t get freedom. You get noise.
The genetic code’s redundancy makes sustained evolution possible. Perec’s forbidden vowel generated a world. The adjacent possible’s bounded quality is what makes it a space you can actually move through.
The edge isn’t a cage. It’s where the terrain begins.