The Adjacent Possible
Lateral thinking, pattern recognition across fields, and staying useful when everything accelerates
The full phrase is “jack of all trades, master of none, though oftentimes better than master of one.” The second half got dropped somewhere, which tells you something about what expertise culture values and what it prefers not to examine.
There is a concept in evolutionary biology called the adjacent possible. The biologist Stuart Kauffman used it to describe the set of things that could exist next, given what already exists. Every biological configuration opens some doors and closes others. The space of what can become is not infinite; it is constrained by the shape of what is. But it is always larger than where you currently stand.
The idea migrated into other fields because it describes something recognizable across them: technology, institutions, ideas, careers. You cannot leap to a configuration that requires ten intermediate steps you haven’t taken. But you can move laterally to find the adjacent door that no one else is looking at, because everyone else is staring straight ahead.
Most professional development advice is about going deeper. Get more specialized. Accumulate credentials in a narrowing cone. The expertise economy rewards this, at least for a while, because depth is easy to verify. It can be credentialed, ranked, and hired. Breadth is harder to credential. It looks like dilettantism to colleagues and like insecurity to yourself.
But depth without breadth produces a specific and underdiagnosed failure mode: the inability to recognize when the problem you’re solving has changed shape. The specialist keeps applying a refined tool to a problem that has moved. The generalist has a cruder set of tools but a wider scan angle. They notice when the problem has migrated into an adjacent field where they happen to have some footing. This is not an argument against expertise. It is an argument that expertise should be paired with what some researchers call “structural holes” awareness, the capacity to see which domains are not talking to each other, and to understand that the gap between them is where value tends to accumulate. The translator between two isolated clusters of knowledge is often worth more than the deepest expert in either one.
Lateral thinking, at its core, is the practice of noticing when a problem in domain A is structurally identical to a solved problem in domain B, and then doing the work of transposing the solution carefully, with attention to what does and does not transfer. The “carefully” part matters. Bad lateral thinking is analogy without rigor, and there is a lot of it around. Someone notices that immune systems and cybersecurity systems both involve threat detection and concludes that every insight from one maps directly onto the other. Sometimes it does. Often it doesn’t. The discipline is in the mapping, not the noticing. Noticing is cheap. The structural analysis of what actually carries over, and what dissolves under scrutiny, is the work.
A cleaner example: epidemiologists studying disease transmission developed, over decades, a sophisticated vocabulary for modeling how behavior changes in networked populations under conditions of incomplete information. When researchers began applying those models to the spread of financial contagion after 2008, the transfer was not automatic. The structural similarity was real, but the boundary conditions differed enough that naive borrowing produced bad predictions. The people who got it right were those who held the epidemiological framework loosely enough to ask which assumptions were load-bearing and which were artifacts of the original domain. That kind of disciplined porousness is harder than it sounds.
Cross-field pattern recognition at its best is usually preceded by extended immersion in at least two domains, not shallow survey reading but genuine engagement with the problems each field considers hard. It involves tolerance for the discomfort of being a novice in one domain while expert in another. And it almost always requires a willingness to be wrong in public, because lateral moves are more falsifiable than depth moves. When you claim to have found a connection no one else has seen, you are making a prediction, and predictions can fail.
Consider someone working at the intersection of public health, institutional governance, and emerging technology policy. None of those fields were designed to talk to each other, and in many settings they still don’t. The public health people tend to think technology governance is someone else’s problem. The technology people tend to think public health is a domain-specific application of principles they already understand. The institutional people tend to think both groups are naive about how change actually moves through bureaucratic systems. The most useful capacity a person in that position can develop is not deeper subject matter expertise in any single one of them, though that matters too. It is the ability to carry a problem’s shape intact across a domain boundary and ask what it looks like from over there. Every field has its own grammar, its own presuppositions, its own sense of what counts as a serious concern. You have to learn enough of each grammar to hold the translation without losing fidelity in either direction.
It is worth pausing on the danger in that self-description. “Translator across domains” is also exactly the kind of identity that flatters a certain type of ambitious generalist, the person who has decided that his inability to sit still in any one field is actually a strategic asset. That story is seductive and sometimes true. It is also available as a consolation to people who are simply restless, or who mistake breadth of exposure for depth of understanding. The cross-domain thinker who has read widely but thought shallowly can become very good at producing the feeling of insight without the substance of it, pattern-matching at the surface while the real structural work goes undone. The check against this is not modesty as a posture. It is the willingness to be tested, to make the prediction explicit, and to find out whether the transfer actually held.
This is exhausting, and often unrewarded in the short term. Fields protect their borders. The person who has clearly mastered one domain is easier to trust than the person who claims to move between several. But the long-term pattern is different. The translators end up doing the work that no one else can do when conditions change, and conditions always change.
When things move fast, the temptation is to absorb more, read more, skim more, follow more sources. This produces the sensation of being informed while often reducing actual comprehension. The signal-to-noise ratio collapses. You end up with surface familiarity and very little structural understanding. Investing in frameworks that travel is harder and slower, but it compounds. A framework tied to one domain becomes obsolete when that domain shifts. A framework describing a structural relationship, the dynamics of information asymmetry, the logic of commitment devices, the way incentive structures produce emergent behavior, remains useful across a wide range of conditions and a long stretch of time. You are not predicting what will happen. You are developing the capacity to recognize the shape of what is happening faster than people reasoning from domain-specific models alone.
This is close to what superforecasting research reveals about why some people make better predictions than others. It is not raw intelligence or domain expertise. It is the capacity to hold multiple models simultaneously, update them based on evidence, and resist the pull of any single framework’s gravitational field. The best forecasters are, functionally, lateral thinkers operating under epistemic discipline. They borrow from every domain that bears on the question and discount their own prior beliefs at a rate the average expert finds uncomfortable.
The adjacent possible is always there, and it is always larger than you can see from where you’re standing. The question is whether you have built the habit of looking sideways, into the neighboring room, at the problem your field has been ignoring because it technically belongs to someone else’s discipline. The people who do that well are not simply more synthetic or more curious. They have learned to preserve the shape of a problem while crossing a border, which is a stranger and more demanding skill than it sounds, and one that the systems they move through rarely reward cleanly. They end up useful.