There is a familiar complaint about large institutions: nobody seems to understand how the whole thing works. Ask one department about a problem and it refers you to another. Speak to a specialist and they understand their part extraordinarily well but cannot explain what happens elsewhere. Senior managers may know the formal structure while front-line employees understand operational realities that never appear on an organisational chart. Technical teams know the systems, legal teams know the rules, policy teams know the objectives, and service teams know what happens when all of those things encounter the public. No one seems to know the whole.
It is easy to interpret this as evidence of institutional failure. Surely a well-functioning organisation should understand itself. Surely someone, somewhere, should possess the complete picture. If nobody does, the institution can appear fragmented, incoherent or even irrational. Yet this expectation becomes much less plausible once we consider what complex institutions are actually required to do.
A national government, a health system, a university, a large municipality or an international organisation operates across domains that no individual could realistically master. Legal interpretation, budgeting, procurement, information technology, policy design, operational delivery, evaluation, security, communications and countless forms of specialist expertise coexist inside the same institutional environment. Each domain contains its own knowledge, practices, uncertainties and professional languages. The complexity that makes institutions necessary also makes complete individual understanding increasingly impossible.
This is not an accidental defect. It is partly the result of specialisation, one of the fundamental mechanisms through which institutions become capable of doing things that individuals cannot do alone. A public health agency does not need every epidemiologist to understand procurement law, every procurement specialist to understand disease modelling or every data engineer to understand the political history of health regulation. Requiring everyone to know everything would not make the institution more intelligent. It would make complex institutional action impossible.
Specialisation therefore creates a paradox. Institutions become more capable by distributing knowledge, but the same distribution makes it less likely that any individual will understand the institution as a whole. As organisations acquire more expertise, they may actually become harder for any one person to comprehend. The absence of a complete knower can therefore be a consequence of institutional sophistication rather than evidence of institutional stupidity.
This distinction matters because otherwise we risk diagnosing the wrong problem. Imagine discovering that no employee in a large public organisation can describe every step involved in delivering one of its most complex services. That fact alone tells us surprisingly little about whether the organisation works well. Perhaps each part understands what it needs to understand, relevant information moves effectively between specialised units, responsibilities are clear and the institution reliably produces good outcomes. In that case, nobody knowing the whole may be entirely compatible with high institutional capability.
The real problem begins somewhere else. Fragmentation becomes dangerous when the institution cannot compensate for it. A specialist may understand one part of a problem without knowing that another unit possesses information that would change the analysis. Two departments may act rationally from their own perspectives while producing an irrational result together. A decision-maker may receive several technically correct pieces of advice without anyone recognising that they depend on contradictory assumptions. An operational failure may repeatedly occur at the boundary between units because each unit sees only the portion for which it is formally responsible.
Here, the issue is not simply that knowledge is distributed. The issue is whether the institution can make distributed knowledge work as a system. Complex organisations do not need a single person who understands everything, but they do need ways of recognising when different fragments of knowledge must meet. They need mechanisms through which one part can discover that another part knows something relevant. They need enough shared language to communicate across specialist boundaries, enough institutional visibility to locate expertise, and enough coordination to assemble partial perspectives when a problem cannot be understood from one position alone.
This also helps explain why organisational charts can be misleading representations of institutional cognition. An organisational chart tells us where formal responsibilities sit. It does not necessarily tell us where knowledge resides, how information travels or which relationships allow different forms of expertise to be combined. Two units that appear adjacent on a diagram may barely communicate, while employees located in distant parts of the institution may maintain informal relationships through which critical knowledge moves rapidly. The formal institution and the cognitive geography of the institution are related, but they are not identical.
For leaders, this creates a particularly important temptation. Senior positions often generate an expectation of comprehensive understanding. The higher someone rises in an organisation, the more they may be expected to “see the whole”. But leadership cannot eliminate the underlying distribution of knowledge. A minister, chief executive or senior civil servant receives compressed representations of institutional reality: reports, briefings, dashboards, meetings, indicators and advice. These tools can make complexity more manageable, but they cannot reproduce inside one person’s mind everything known across a large institution.
The goal of institutional intelligence should therefore not be to create an imaginary all-knowing centre. Centralising every piece of knowledge would often destroy the advantages of specialisation while overwhelming those expected to process it. A more realistic objective is to build an institution in which partial knowledge can remain distributed without becoming institutionally disconnected. Different parts should be able to know different things while the organisation retains the ability to connect those things when a decision or problem requires it.
This changes how we interpret one of the most common experiences of institutional life. Discovering that nobody understands the whole organisation need not immediately alarm us. In a sufficiently complex institution, it may be exactly what we should expect. The more useful questions are whether people understand the parts for which they are responsible, whether the institution knows where relevant expertise resides, whether important information can cross organisational boundaries, and whether fragmented perspectives can be combined when circumstances demand a wider view.
A complex institution does not need anyone to know everything. It needs an architecture that allows what different people know to become collectively useful without requiring any one of them to possess the whole. The challenge is not to eliminate fragmentation, but to prevent necessary cognitive specialisation from becoming institutional blindness.