The Art of Scaling

Scale kills essence.
What spreads widely is never what it began as.
Choose reach, or choose meaning. Never both.

For much of human history, religion belonged to particular places. Gods were attached to landscapes, shrines and communities, and their stories were carried through local rituals. A belief did not have to make sense everywhere because it only needed to make sense here.

Universal religions introduced a different kind of structure: one god, one book and a framework of values that could travel beyond the place where it began. To survive that journey, the framework had to be both recognisable and adaptable. It needed enough definition to hold people together, but enough abstraction to acquire a slightly different meaning in every culture it entered.

This is one of the recurring patterns of scale. Growth is rarely a matter of making more copies of an unchanged original. The thing itself has to become portable. Context is removed, variation is narrowed and qualities that once lived in relationships or traditions are turned into rules that can be taught to strangers.

Languages follow a similar path. Thousands have disappeared as states, empires, trade and global companies concentrated communication around a much smaller number. A dominant language creates extraordinary reach. People who share almost no history can work together, exchange ideas and understand one another. The same process also displaces local languages carrying forms of memory and experience that do not translate cleanly.

The gain is real, and so is the loss. Scale connects by establishing a common layer, but a common layer can only remain common if it leaves some difference underneath.

Political ideas reveal the same tension more directly. Anarchy begins with a simple proposition: no state and no rulers, with people arranging their lives through direct agreement. In a small group, where everybody knows who made a promise and who failed to keep it, much of that coordination can remain personal.

Larger societies cannot rely on familiarity in the same way. Agriculture turned settlements into towns, strangers became dependent on one another and decisions began to affect people who would never meet. Institutions appeared because agreements had to persist beyond a conversation. They were often primitive, unequal and violent, but they solved a problem that face-to-face negotiation could no longer contain.

Pacifism changes in a related way. As a personal ethic, refusing violence can be an unusually demanding expression of courage and humility. At the scale of a state, the same refusal includes a decision about the safety of millions of other people. A pacifist government confronted by an aggressive neighbour cannot make sacrifice only on its own behalf. What is noble at the scale of one person may become irresponsible when imposed at the scale of a population.

The point is not that either idea is naive. It is that an idea changes when the number and relationship of the people involved change. Some principles survive that translation. Others depend on intimacy, shared context or voluntary participation, and begin to fracture when those conditions disappear.

The same pattern is visible in something as ordinary as a family restaurant. Its quality may depend on an owner tasting the food, a cook who has made the same dish for twenty years and a room where regular customers are recognised. None of this appears on the menu, but it is part of what the restaurant sells.

Opening a second location seems like replication, yet it immediately requires a different organisation. Recipes become instructions, judgement becomes training, suppliers are standardised and the owner divides their attention. The new branch may carry the same name and serve the same dishes while gradually losing the network of small decisions that made the original good.

McDonald’s and IKEA succeeded because they did not attempt to reproduce a local experience unchanged. McDonald’s narrowed its promise to speed, safety and predictability. IKEA built a system around range, self-service, flat-pack logistics and products designed for many ordinary situations. One reduced variation; the other organised an enormous amount of it. Both redesigned the experience around repeatability.

This suggests a more useful rule than the idea that large things must be bad. A narrow promise is easier to reproduce than a rich one. The more a product depends on taste, tacit knowledge or a particular relationship between people, the harder it is to make available everywhere. Scale does not necessarily lower quality, but it changes which kinds of quality a system can reliably provide.

Travel offers a more uncomfortable example because the value being scaled was partly a rejection of scale itself. For a generation, buying memories rather than objects became a moral as well as an aesthetic choice. Movement promised freedom from routine consumption, and crossing borders became evidence of a fuller and more curious life.

Then millions of people pursued the same escape. Barcelona began pushing back against tourists. Short-term rentals distorted housing markets in cities such as Milan. Paris filled with congestion, Amsterdam’s centre became difficult to separate from the crowd moving through it, and places such as Venice, Florence and Prague absorbed volumes their streets and monuments were never designed to carry.

No individual traveller has to behave badly for this to happen. One person can be attentive, curious and respectful; ten million people doing the same thing can still overwhelm housing, transport and public space. When enough people pursue the same experience of authenticity, the infrastructure required to serve them begins to replace the thing they came to find.

What began as a meaningful rejection of materialism did not become false. It simply produced different consequences once it became a mass social value. Intent remained personal while impact became collective.

Technology companies provide the most familiar organisational version of this problem. Over two decades, small teams became institutions employing tens of thousands of engineers. Growth required continuous hiring, onboarding systems and layers of management capable of dividing an expanding organisation into units that somebody could still understand.

Agility became a slogan, autonomy became a process and reorganisation settled into the background as a permanent activity. The structure was designed to absorb expansion. It was much less clear what that structure should do once expansion stopped.

The pandemic accelerated remote work and compressed decisions that organisations might otherwise have made slowly. A few years later, interest rates rose, capital became more expensive and growth lost its position as the default answer. Hiring slowed, layoffs arrived and companies that had spent years adding people were suddenly expected to produce more with fewer of them.

This created an awkward question: what is management for when the organisation is no longer growing? Some layers existed to recruit, coordinate and subdivide new teams. As that work receded and AI tools increased what smaller groups could produce, the distance between strategy and execution became harder to justify.

The obvious response would be to collapse layers, widen the scope of teams and bring leaders closer to the work. Organisations often resist this because a structure developed for growth quickly acquires its own history, incentives and constituencies. The company may become smaller while continuing to operate through the model built for becoming larger.

Three limits sit underneath these examples. The first is logistics: the growing burden of moving people, goods and information. The second is simplification: the need to turn a complex system into smaller models that can be repeated and managed. The third is cognition: the limit on how much context any person can carry before relying on abstractions, reports and shortcuts.

These limits reinforce one another. More logistics creates more information. More information demands simplification. Simplification removes context, which increases the chance that people optimise the model while missing what is happening in the real system. What began as something connected and understandable is divided into pieces, and the pieces gradually become the organisation’s reality.

Refusing scale is not a clean solution. The family restaurant can preserve its character and still disappear when the owner retires. A local language can carry a unique way of seeing the world and vanish with its last speakers. A small organisation may remain coherent while lacking the resources to survive a serious mistake. Smallness protects context, but it also limits access and increases fragility.

The real choice is therefore not simply between growth and purity. It is deciding which qualities are essential, which can be standardised and how much transformation the original idea can withstand. Reach always asks for translation. The honest question is whether the translated version still contains the reason the thing mattered in the first place.

We keep looking for a form of growth that preserves every local detail, tacit judgement and human relationship while reproducing indefinitely. Usually it does not exist. Something will be delegated, simplified or lost. The art of scaling begins by deciding what may change—and by recognising the point at which the thing we are spreading has become something else.