The fitness landscape holds still, and almost nothing else does. Swap it for a seascape and the strategic advice flips: the interdependence that walls out imitators is the same interdependence that traps you when your own peak begins to sink.
In 1932, at the Sixth International Congress of Genetics in Ithaca, New York, Sewall Wright drew a picture. He had been asked to explain his ideas to a room of biologists who were, by reputation, uncomfortable with his mathematics, so he gave them something to look at instead of equations. The picture was a landscape: contour lines, peaks, and valleys, like a topographic map. Every point on it stood for a particular combination of genes, and the height at that point stood for fitness, meaning how well that combination survives and reproduces. Populations, in this picture, climb. They walk uphill toward the nearest peak of high fitness. It became one of the most influential images in the history of biology, and it may be one of the most misleading things you can hand a strategist.
The drama of Wright's picture lives in the valleys. A population sitting on a modest peak cannot easily reach a taller peak nearby, because to get there it would first have to walk downhill, across a valley of low fitness, and natural selection fights every downhill step. So populations get stranded on local optima: good-enough summits they cannot leave, because leaving means getting worse before getting better. Wright had his own answer for how life escapes these traps, a theory about small populations and random drift letting a lucky few wander across the valley. But the lesson strategists borrowed from his landscape was simpler and stickier than the theory. Find your peak. Climb it. Defend it. And if you are stuck on a low one, the only way up is a dangerous leap across the gap.
It is an irresistible thing to overlay on business. Swap genes for strategic choices and fitness for profit, and Wright's landscape becomes a map of competition. Your company occupies a position, some peak of a working business model. Competitors sit on other peaks. The valleys are the half-formed, unprofitable configurations that no one wants to occupy. This is not a loose metaphor; it is a serious research tradition, and further down I will get to the paper that made it rigorous. But there is a flaw in importing Wright's picture into strategy, and it is not a small one. It is a category error, and once you see it, the advice the picture gives you turns out to be backwards.
Here is the error. Wright's landscape holds still. The peaks stay where they are. The valleys keep their shape. The entire story is about something moving across ground that does not move. Now compare that to the most useful map in modern technology strategy. Simon Wardley's mapping method, used inside governments and large enterprises to reason about where to build and where to buy, has as its horizontal axis not position but evolution. Every component of a business, Wardley argues, drifts along that axis in one direction only: from Genesis, the brand-new and uncertain, to Custom-built, to Product, and finally to Commodity, the utility you rent without a second thought, the way you rent electricity. Computing power walked that whole road; by Wardley's reckoning it took roughly forty years to travel from a genesis-stage novelty that could differentiate a business to something you now rent by the hour from any cloud provider. The march is treated as inexorable. The terrain itself moves. And a landscape that holds still cannot represent terrain that moves.
So the Wardley practitioner reaching for a fitness landscape is reaching for the wrong picture. The right one exists, and it has a name. In 2009, Ville Mustonen and Michael Lässig published a paper in Trends in Genetics that extended, in their words, “the static concept of fitness landscapes to dynamic fitness seascapes.” A seascape is a fitness landscape whose shape changes over time. The mapping from genes to fitness will not hold still; it shifts, so peaks rise and subside and valleys migrate. Adaptation on a seascape, they wrote, is “a non-equilibrium phenomenon: the genomic response to time-dependent selection.” Biologists reach for seascapes to describe exactly the situations where the environment refuses to sit for its portrait: a pathogen evolving against a drug regimen that rotates every few weeks, a virus tracked by an immune system that keeps moving the target. The organism is not climbing a mountain. It is surfing.
That one substitution, from a landscape to a seascape, flips the strategic advice on its head. On a landscape that holds still, the peak is the goal: reach it, plant your flag, hold the summit. On a seascape, being at the peak is a liability the moment the peak begins to subside. If the optimum is moving, then tracking it beats occupying it, and the firm most perfectly adapted to today's terrain is, by that very perfection, most precisely fitted to terrain that is about to vanish. The tighter your fit to the current shape, the farther you must travel when the shape changes. Perfect adaptation turns out to be a kind of debt, and it comes due the instant the ground moves.
If you have spent time with Wardley maps, you already own a word for this. It is inertia. Wardley uses inertia to name the drag that keeps an organization from adapting as a component evolves, the accumulated weight of existing revenue, hard-won skills, culture, and past success that pins a company to one stage of the map while the component slides toward commodity beneath its feet. Incumbents feel it most sharply, and not by bad luck. They feel it because they are the best-adapted, the ones with the most tuned to a peak that is now flattening out, the ones with the most to lose from the terrain's next move. Wardley presents inertia as a catalog of organizational pathologies. The seascape gives it a mechanism. Inertia is not a failure of nerve or imagination. It is the expected cost of having been beautifully adapted to a landscape that has quietly become a seascape. One honest caveat: this reading of Wardley through the seascape is my argument, not a claim either Wardley or the biologists make. They have never met. That the two pictures lock together this cleanly is the point of this essay, not a citation from it.
A tidy essay would stop right there, at “the ground moves, so keep moving.” Strategy is not that generous, and the sharpest idea here is the counter-argument. In 2000, Jan Rivkin published a paper in Management Science called “Imitation of Complex Strategies,” asking a plain question: why aren't good strategies simply copied? His answer throws out the usual explanations. Rivkin's model, in his own words, “excludes conventional resource-based and game-theoretic barriers to imitation altogether.” No patents, no trade secrets, no first-mover lock-in. He keeps exactly one ingredient: complexity, defined as the number of elements in a strategy and the interactions among them. And complexity alone, he proved, is enough to protect you. When a strategy is a dense web of many tightly-interacting choices, finding the best combination becomes, in his phrasing, “intractable in the technical sense of the word provided by the theory of NP-completeness.” The space of possible strategies is too rugged to climb.
The consequence is startling enough to state at full volume: you could publish your entire strategy, hand a competitor the complete blueprint, and stay safe. Rivkin's own language tells you why. Firms that try to copy a high performer's whole strategy “suffer large penalties from small errors,” and firms that climb greedily toward the nearest gain “are quickly snared on low 'local peaks.'” Read those words again. Local peaks. Hill-climbing. Rivkin is describing a rugged fitness landscape, in the precise vocabulary Sewall Wright used in 1932, and he is doing it in a strategy journal. Partial imitation fails for the same reason valley-crossing is hard: a tightly-coupled system has in-between states that are worse than either end. Copy half of an interlocking strategy and you do not get half the benefit. You get a configuration stuck in the valley, worse than where you began. The interdependence is the moat.
Now hold Rivkin and Wardley in the same hand, because this is where the tidiness breaks and something truer shows through. The interdependence that protects you from imitators is the same interdependence that stops you moving when your own peak subsides. Rivkin's moat and Wardley's inertia are not two phenomena. They are one mechanism seen at two timescales. On the short timescale, a rival tries to copy you, lands in the valley between their position and yours, and gives up; your complexity has defended you. On the long timescale, the seascape shifts, your peak starts to sink, you try to move to the new optimum, and you fall into that very same valley, because your choices are wound just as tightly for you as they were for the imitator. The wall that keeps others out keeps you in. There is no setting of the dial that hands you the moat without the trap. The formal name for that dial is Kauffman's NK model, where a landscape of N components with K interactions apiece grows more rugged as you turn K up, sliding from a single smooth peak toward a badlands of many jagged ones. More interdependence, more ruggedness. More protection, and more places to get stuck. It is a single knob, and it controls both.
Two honest limits, because this argument has real seams and a reader who appreciates cross-domain thinking will find them anyway. First, of the three bodies of work I have braided together, one is not like the others. Fitness landscapes and Rivkin's model are formal and tested. Wardley mapping is a practitioner's framework, hugely useful and widely adopted, but not validated the way the other two are. Its evolution axis is a sharp and productive posit, not an established law of technology. Treat the biology as evidence and the Wardley half as a well-worn hypothesis, and the pairing stays honest. Second, and more unsettling, the valley itself may be partly a trick of the drawing. Sergey Gavrilets observed that our whole intuition about peaks separated by valleys comes from imagining a landscape in three dimensions, while real genotype spaces have thousands. In spaces that high-dimensional, he showed, high-fitness configurations tend to be linked to one another by connected ridges. He called them holey landscapes, and on one you can often walk from peak to peak along nearly level ground without ever descending into a deep valley. If strategy space is anything like that, then the valley you dread may sometimes have a ridge around it that you simply cannot see from inside the three-dimensional cartoon. The trap is real. It may also be more escapable than the picture admits.
So the practical decision is not the one the old picture hands you. The static landscape says climb or jump, occupy or leap. The real decision, once you accept that the ground moves, is quieter and much harder. It is how much interdependence to buy. Every tightly-coupled choice you add to your strategy raises the wall against imitators and, in the same stroke, shortens your own reach when the seascape shifts. A loosely-coupled strategy is easy to copy and easy to change. A tightly-coupled one is hard to copy and hard to change. You are always buying protection and inertia together, in one purchase, at one price, and the only real question is how much of each you can afford given how fast your particular terrain is moving. On slow ground, buy the moat; the shift is distant and imitation is the nearer danger. On fast ground, keep your coupling loose enough to move, and make peace with being easier to copy, because being copyable matters far less when the peak you are defending is about to be underwater. We have written before about what gives a lineage the sheer capacity to change, the traits that let an organism or a company move quickly at all. This is the other half of that story: what happens to the well-adapted when the thing they are standing on begins to move too. Wright drew his landscape in 1932 for a room of people who needed a picture in order to think. It worked so well that it became the way we think, and that is exactly the trouble, because the picture holds still and almost nothing else does. The best strategists were never the ones who found the highest peak. They were the ones who never forgot they were standing on water.
The tighter your fit to today's terrain, the farther you must travel when it moves. Perfect adaptation is a debt.
Nowhere is the ground faster than the model landscape an agent system is built on: the peak subsides every release. The way to stay able to move is to keep your trust layer loosely coupled to the model beneath it, so switching what runs underneath does not mean rebuilding how you prove it worked. The Agent Trust Stack is that loose coupling: provenance, verification, and reputation that travel with the work rather than welding you to one vendor's peak. Buy the coupling you can afford to keep, and keep the layer that lets you move.
Read the Theory of Agent Trust
Install the whole stack: pip install agent-trust-stack · npm install agent-trust-stack