The One Inch Frame
A toddler pushes a glass toward the edge of a table. Stops. Pushes again. Stops and looks. Pushes it off. This is not destruction. It is the oldest experiment in the world.
Marc Andreessen has observed — in conversation with Andrew Huberman — that “the sheer number of people in the world who are capable of doing new things is just a very small set of people. You’re not going to have a hundred of them in a company… You’re going to have 3, 8, or 10, maybe.” He illustrates the point with IBM’s famous Wild Ducks program: a tiny aristocratic class of eight IBM Fellows who were explicitly allowed to break every rule, pull resources, report only to the CEO, and invent the next breakthrough product — while the 6,000-person divisions were expected to follow the existing system. It is a useful observation about organizations. But it mistakes the finish line for the race.
A new product is what creativity looks like after years of accumulation. It is not what creativity is.
The Confusion
We have a habit of defining creativity by its most dramatic outputs: the invention, the breakthrough, the product that changes a market. This is like defining running by the Olympic final. The Olympic final is real. But running is something millions of people do every day, at every level, and those people are genuinely running — not performing a lesser version of something that only counts when Usain Bolt does it.
When we reserve the word “creative” for people who have shipped a product or made a famous work of art, we do two things. We miscount the creative people around us. And we make creativity sound like a trait you either have or don’t, rather than a practice you either develop or neglect.
But the stronger objection runs in the other direction too. If every adaptive act counts as creativity, the word stops meaning anything. Following a recipe isn’t creative. A pilot running a preflight checklist isn’t being creative — and shouldn’t be. Execution is real. It matters. It keeps planes in the air. The question is what distinguishes the creative act from the competent execution of a known process, and the answer turns out to be precise: creativity happens at the margins, where the script runs out.
The stronger objection to this essay is worth stating plainly: Andreessen is making a managerial argument, not a philosophical one. He needs to know who can staff the Wild Duck pen. For that purpose, the 3-to-10 threshold is the right tool. Nothing in what follows disputes that practical count. What it disputes is treating that count as a complete account of creativity — because organizations and individuals both pay a price when they do.
What Comes Before the Product
Bell Labs produced the transistor, the laser, Unix, and information theory itself. None of these started as products. They started as questions. Claude Shannon asked what information actually was — not how to make a faster telegraph, but what “information” meant as a mathematical object. The transistor came from physicists asking what was really happening at the surface of certain materials.
DARPA, at its best, works the same way. DARPA program managers don’t fund products. They fund the development of capabilities that don’t yet exist, because they’ve identified something real that current tools cannot do. The product comes later, from someone else, standing on the foundation DARPA helped build.
Skunk Works built the U-2 and the SR-71 Blackbird in a separate facility, away from Lockheed’s main operations, precisely because the main organization would have required the new aircraft to meet existing specifications. The breakthrough required freedom from the inherited assumptions — but that freedom was used for years of disciplined engineering before anything flew.
Newton’s line about standing on the shoulders of giants is famous enough to be cliché. What gets missed is that most of the time, the shoulders we stand on are our own. Growth requires change. But the useful kind of change — directed change — turns motion into discovery. You test something, you observe what happens, you update, you try again. Each test is a small creative act. The eventual product, if there is one, is built from accumulated small acts, most of which no one outside the lab ever sees.
If you wait to count creativity only at the product stage, you are counting only the last step of a long staircase.
What Creativity Actually Is
Here is a more useful definition: creativity is the practice of finding out what is actually possible, as distinct from what you’ve been told is possible, through edge-testing — pressing on apparent limits to see what holds and what doesn’t.
This definition has three parts, and all three matter.
Finding out what is actually possible. Not assuming. Not accepting the received account. The world is full of constraints. Some are real in the way that gravity is real: you cannot work around them. Others are real in the way that a locked door is real: they stop you unless you have a key, or find a window, or determine that the door was never locked to begin with. Creativity requires distinguishing between these two types. The only way to distinguish them is to press.
As distinct from what you’ve been told is possible. Organizations, institutions, relationships, and bodies all have inherited ideas about what falls within bounds. These ideas are sometimes accurate. Often they are not. They were accurate for a different time, or a different person, or they were never accurate and simply went unchallenged. The creative orientation treats these inherited accounts as hypotheses, not facts — worth testing, not worth assuming.
Through edge-testing. This is what converts orientation into practice. A disposition toward questioning, without actual tests, produces only skepticism. The test is what generates information. And the information from the test is what makes the next step possible.
The creative person is not primarily someone with unusual ideas. They are someone with an unusual stance toward limits — specifically, the habit of checking whether those limits are what they appear to be, and the discipline to read the results accurately when they press. That habit is what Bell Labs researchers had. It’s what Skunk Works engineers had. And it’s what the toddler with the glass has.
Two Kinds of Not-Knowing
Before the test comes orientation. And orientation takes two forms that are easy to confuse.
Exploration is mapping the landscape — finding out what’s out there, what the territory contains, what problems are worth solving. The Bell Labs researcher who asked “what is information, really?” was exploring. He didn’t yet know what he was looking for. He was widening the space of what was visible.
Experimentation is a different operation. You already know what domain you’re in. Experimentation asks: within this landscape, is this particular thing possible? Can I do it this way? Does this process work as assumed, or does reality push back?
Creativity needs both. Exploration without experimentation produces interesting observations that never generate new capability. Experimentation without exploration produces incremental refinement within inherited assumptions — efficient, sometimes excellent, but bounded by the original map. The path from Shannon’s question about information to the transistor to Unix to the internet ran through decades of both: widening the landscape, then testing specific possibilities within it. The MVP at the end of that chain was built on an enormous invisible foundation of both kinds of work.
Organizations that value only what can be measured tend to fund experimentation while starving exploration, because experimentation produces legible results on legible timescales. This is not irrational — but it depletes the foundations that experimentation requires.
The Three Practices
If creativity is the practiced habit of edge-testing and reading the results accurately, it can be developed. Three practices build it.
Test small, test deliberately. You do not need to attempt a breakthrough to practice creativity. You need to attempt something slightly different from what you did last time, in any domain, and notice what happens. A musician who tries a chord substitution they’ve never used, a cook who adjusts a recipe, a manager who runs a meeting differently: these are genuine edge-tests. Not because they will lead to great art or a product launch. Because they produce information about what is actually possible, which the actor did not have before. The improvising jazz musician has been testing for years. The ability to do something genuinely new in a performance is built from those accumulated tests, not separate from them. This practice serves people who have some latitude to absorb a failed test; it is harder to execute when the cost of any deviation is immediate and severe — a limitation returned to below.
Read the resistance accurately. Not every test produces a positive result. A lot of them hit walls. When you press on an apparent limit and something pushes back, the practice requires distinguishing between different kinds of resistance. Some walls are real: the physical, biological, logical constraints that do not yield regardless of effort or approach. These are worth knowing — not as defeats, but as orientation points. Others are institutional: they represent the interests, habits, or accumulated caution of an organization, not the actual structure of what’s possible. Others are self-imposed: old habits of thought that feel like external constraints. One useful discriminator: if changing your approach changes the outcome, you are not facing a hard constraint. If no approach changes it, you likely are. The test tells you the wall is there. Reading the result accurately tells you what kind of wall it is — and what the appropriate response is.
Direct the change rather than react to it. Change happens regardless. The uncreative response is to be carried by change — to react to new situations with old methods, to let the environment determine what you do. Directed change means using the information from tests to choose the next step, rather than having the next step chosen by accumulated habit or external pressure. The scientist who designs the next experiment based on what the last one showed is directing change. So is the cook who tweaks the recipe based on last night’s feedback, rather than simply salting more aggressively next time. Both are working. Only one is generating new information.
The Tolerance Band
These three practices operate within what might be called a tolerance band — the actual range of deviation an individual or organization can absorb before a test becomes catastrophic rather than informative.
The toddler’s tolerance band is high: a broken glass is absorbed by a parent, and the information gained (gravity is real, limits exist, glass breaks) is enormous relative to the cost. A junior employee’s band may be low: visible failure carries career risk, which suppresses testing and narrows the zone of exploration. The same creative act lands differently depending on where you sit.
This is not a counsel of despair about constraints. It is a practical observation: creative practice, for most people in most contexts, means right-sizing the test to the band. Small tests, read accurately, gradually expand the band — because accumulated small wins build the credibility and understanding that make larger tests survivable. The Skunk Works engineers didn’t start by designing the SR-71. They started somewhere. Everyone does.
The organizational implication is pointed: organizations that narrow the tolerance band through punishment of visible failure don’t eliminate creativity. They push it underground, slow its accumulation, and eventually lose the people who most need room to work. The question worth asking in any institution isn’t “how many people can ship an MVP?” It’s “how many people are actively edge-testing what is possible in this domain, how accurately are they reading the results, and what happens to them when a test fails?”
The MVP Problem
Measuring creativity by whether someone can launch a minimally viable product imports a specific set of assumptions about what counts as output. A product is something that can be shipped, sold, or used. By this measure, Shannon’s work on information theory — published as a paper, not shipped as a product — was not the creative act. The product that eventually used it was the creative act. This gets things exactly backwards.
The MVP is a measure of exploitation: taking a developed capability and bringing it to market. What precedes it is exploration and experimentation — the phases that make exploitation possible. An organization that optimizes purely for the exploitation phase without attending to the conditions that produce it is spending down a resource it is not replenishing. The Wild Duck program at IBM was Watson Jr.’s imperfect attempt to protect that resource inside an institution whose natural momentum was consuming it.
More importantly, the MVP standard creates a false binary. Either you can ship something new, or you’re not creative. This means that most of the creative work in any organization is invisible — not because it isn’t happening, but because it hasn’t yet reached the stage where it becomes measurable by that standard. The staircase is there. The product is just the last step.
What This Means in Practice
For the individual: creativity is not a gift you were born with or weren’t. It is a set of habits that can be built. The central habit is the disposition to edge-test — to press on apparent limits, read what you find, and use that information to direct the next step. Not always, not recklessly, and not in contexts where the right move is disciplined execution of a known process. A pilot’s preflight checklist is not the place for creative deviation. But when the situation exceeds the checklist — when the problem is genuinely new — the pilot who has been edge-testing all along has resources the pure executor does not.
For the organization: the question is not how many people can ship an MVP. The question is what conditions make edge-testing possible and safe enough that the people who can do it will do it, and what they learn will accumulate rather than disappear. Organizations that answer this question well — that expand rather than contract the tolerance band — aren’t just managing innovation. They are replenishing the foundation that everything visible is built on.
The MVP is the last page. Creativity is everything that happened before the book was even opened.
Open Questions (Ω)
Ω: Threshold — empirical — At what point does accumulated small-scale edge-testing generate genuine novelty, as distinct from incremental refinement? The claim that quantity of tests produces qualitative change is plausible but not demonstrated here.
Ω: Suppression vs. Selection — underspecified — Whether organizations that suppress creativity are selecting against creative people or against creative behavior depends on empirical facts about how deeply habits can be modified by environment. The essay assumes the latter (conditions over traits) without fully defending it.
Ω: Whose tests count — structural — The tolerance band framework implies that people with very little slack face structurally different constraints, not merely psychological ones. The practices recommended here index to people who can absorb failed tests without catastrophic cost. A more complete account would address how bands can be expanded from positions of very low initial latitude, and whether that expansion is available to everyone or depends on factors outside the individual’s control.
Ω: Exploration without product — empirical — The essay treats exploration and experimentation as both necessary for creativity. But some productive creative traditions appear to operate primarily in one mode (pure mathematics, certain strains of conceptual art). The relationship between the two phases may be domain-dependent in ways that complicate the general claim.
cafebedouin.org 