When a CEO tells me they wished their employees were more capable at “self-organising”, I usually ask what exactly they mean. The answer matters for more than academic precision; leaders drawn to the notion of decentralisation routinely turn to self-organisation as the mechanism that will deliver it. The appeal is clear: Give people more autonomy, reduce bureaucracy and allow coordination to emerge naturally rather than through layers of management. In other words, let them “organise themselves”.

But this appeal rests on a confusion. As I wrote in a recent paper, decentralisation and self-organisation are not the same thing, and the difference comes down to a simple question: What does the structure of influence look like?

Decentralised vs. self-organising systems

In a decentralised organisation, influence is distributed in an egalitarian manner. Decisions are made by many people rather than concentrated at the top. But behaviour across the entire system is shaped by the organisation’s structures and processes that affect everyone: shared norms, incentives, governance mechanisms, hiring practices, cultural expectations, operating principles and so on. 

Influence, in this sense, is still global. Open-source software communities, Wikipedia, scientific communities, worker cooperatives and many decentralised autonomous organisations (DAOs) sit in this family. More specifically, firms like entertainment software and technology company Valve Corporation and materials science company Gore fall under this category. Although formal hierarchy is limited or even absent, a global structure of influence is very much present.

In a truly self-organising system, no person, team or overarching structure guides the system as a whole. Instead, patterns emerge solely from how individuals interact with one another (i.e. local interactions). Consider a flock of birds: Each bird responds only to nearby birds, with no leader and no overarching mechanism directing the flock. Yet, coordinated behaviour emerges. Termite mounds, ant colonies and unmarked “lanes” forming spontaneously among pedestrians work the same way.

It’s easy to confuse the two ideas because they often appear together in the same examples in the natural world – and the difference becomes obvious only when you look closely in the social world.

Unblurring the line

Markets offer a good illustration of a highly decentralised system, with millions of participants making independent decisions. But market participants are influenced by property rights, contracts, regulations, currency, courts and social norms – system-wide structures that shape behaviour and outcomes. Therefore, markets may be decentralised, but they are not purely self-organising. Managers who opt for the "we will just let things sort themselves out" approach tend to forget how much policy scaffolding the market actually depends on.

The same principle applies inside organisations. Valve Corporation is often cited as an example of radical decentralisation. Employees choose projects, form teams voluntarily and move between initiatives with considerable freedom. Yet, the company still depends on hiring practices, cultural norms, shared expectations and organisational principles that influence behaviour across the entire organisation. Moreover, the founder played a role in designing this system and continues to contribute actively, particularly through hiring. 

As such, the resulting order doesn’t emerge solely from local interactions, but depends on structures that shape how those interactions unfold. Unlike the “invisible hand” of pure self-organisation, this is closer to a hidden hand: a global influence structure embedded in the design of the system rather than exercised continuously through overt command. The hand, though less visible than in a traditional hierarchy, is doing the same kind of work. In its absence, the result may be chaos, or an undesirable order.

DAOs make the point more starkly. These are organisations built explicitly to eliminate hierarchy, with governance handled through token-weighted voting on a blockchain. They are perhaps the most ambitious attempt in recent times to engineer pure decentralisation through self-organisation at scale. However, large-scale evidence on thousands of DAOs shows something striking: Systems designed to eliminate hierarchy reliably reconcentrate influence, often through the accumulation of governance tokens by a few people. When you design for pure self-organisation, you usually don’t get it. Instead, you get a structure shaped by global influence – in other words, a structure in a different costume.

Real lessons from biology

Why are the analogies of self-organising flocking birds and termite colonies misleading for managers?

Flocking birds, ant colonies and immune systems are genuinely self-organising, and also genuinely adaptive. But this is due to a process that doesn’t transfer to human organisations: evolution through natural selection. The order in an ant colony is impressive because of how the local rules that ants follow were fine-tuned over a long time, such that maladaptive colonies died out. What we observe today, and find so fascinating, is the small surviving subset of designs that produces useful global order. The invisible hand looks competent because natural selection has hidden its failures.

Unfortunately, organisations don’t have the luxury of a patient selection process over an extended timeframe. When people interact locally without coordinating structures, the resulting order may or may not be productive. Self-organisation can bring innovation, adaptability and collaboration. On the flip side, it can also lead to silos, the duplication of effort, polarisation and decision paralysis. Economist and policy expert Thomas Schelling showed how even mild local preferences can generate segregated neighbourhoods – even if that outcome is unintended by any individual. Online communities often self-organise into echo chambers rather than productive, deliberative spaces. The order is real. The adaptiveness is not guaranteed.

A matter of design

This is where organisational design becomes critical, and where the role of leadership in decentralised organisations is often misunderstood. Leaders don’t necessarily need to direct day-to-day decisions. Their more important role is to create the conditions under which decentralised decision-making can succeed: defining boundaries, establishing rules, shaping incentives, building culture, determining membership and creating mechanisms that help people coordinate.  

Done well, the hand becomes hidden, not absent. This means the need for constant intervention is reduced as the system becomes capable of coordinating itself within carefully designed constraints. The challenge of organisation design for decentralised systems, then, is not whether to eliminate system-wide global influence altogether, but to determine the extent to which influence can be distributed evenly yet remain effective.

Even the financial market, one of the most decentralised systems ever created, depends on institutions that shape behaviour across the entire system. There is little reason to expect organisations to succeed through pure self-organisation alone.

So, the next time someone in your company proposes a “self-organising” solution, ask the following three questions:

1. Who controls entry into the system, and what kinds of people, capabilities and behaviours does that selection process produce? 
2. How does the system distinguish adaptive from maladaptive patterns once they emerge, and how quickly can it amplify the former and suppress the latter? 
3. What structures, norms, incentives or actors are shaping behaviour across the system, even if nobody refers to them as “authority” or “hierarchy”?