When governments want to preserve what they know, they tend to create places in which knowledge can be stored. They build databases, document procedures, maintain archives, publish guidance, create knowledge-management systems and increasingly organise enormous quantities of information in digital repositories. These resources are essential. Yet they can also encourage a misleading picture of institutional knowledge: the idea that if we could identify the right repository, we could point to the place where the government’s knowledge actually lives.
In reality, the knowledge required for government to function is scattered across a much wider institutional landscape. Some of it certainly lives in documents and databases. But some exists in the experience of civil servants who have encountered the same problem many times before. Some is embedded in procedures that encode lessons learned years ago. Some resides in specialised professional communities, in relationships between units, in established ways of interpreting ambiguous rules, or in an employee’s knowledge of whom to contact when the formal process does not provide an obvious answer. Government knowledge does not live in one place because government itself does not know through one place.
Consider a public programme that has operated for many years. Its official documentation may describe its legal basis, eligibility rules, objectives and administrative procedures. Those documents contain real institutional knowledge, but they are unlikely to contain everything required to make the programme work. Front-line employees may know which parts of an application routinely confuse citizens. Legal specialists may understand why apparently arbitrary distinctions exist. Data teams may recognise patterns that are invisible at case level. Managers may know which operational compromises were introduced after previous failures. Long-serving staff may remember why a procedure was changed even when the original decision has disappeared from everyday documentation.
None of these locations contains the complete knowledge of the programme. More importantly, the forms of knowledge they contain are different. A database can preserve records but not necessarily the practical judgement required to interpret an unusual case. A written procedure can specify what should happen but may not capture every condition under which experienced staff know that escalation is necessary. An expert can understand a specialised domain deeply while remaining unaware of operational knowledge elsewhere in the institution. Institutional knowledge is heterogeneous as well as distributed.
This makes the familiar distinction between explicit and tacit knowledge useful, but still incomplete. Some knowledge can be written down relatively easily: rules, definitions, records, analytical findings or procedural instructions. Other knowledge is harder to separate from experience: recognising an unusual pattern, understanding the practical meaning of an exception, anticipating how another organisation is likely to respond, or knowing which apparently minor issue deserves attention. Yet even this distinction does not tell us everything, because knowledge can also become embedded in institutional arrangements themselves. A routine may preserve a lesson even when the people following it no longer know the history that produced it. A reporting requirement may encode an earlier recognition that certain information must reach a particular decision point. A role may preserve specialised responsibility across successive generations of employees.
So asking where government knowledge lives requires us to look beyond repositories and individual minds. It can live in people, records, routines, roles, procedures, relationships and institutional arrangements. Different kinds of knowledge persist through different carriers, and those carriers have different strengths and vulnerabilities. Documents can survive personnel turnover but become outdated. Experienced employees can interpret ambiguity but eventually leave. Databases can preserve enormous quantities of information but only in forms that systems are designed to capture. Informal networks can move knowledge rapidly but may disappear when relationships change. Procedures can reproduce valuable lessons while also preserving assumptions that no longer make sense.
This distributed geography has practical consequences. An institution may possess important knowledge without knowing where it is. A department may commission research into a problem that another unit has already studied. New employees may recreate solutions that experienced colleagues developed years earlier. A policy team may search databases for evidence while the most relevant operational knowledge remains with front-line staff. Conversely, an institution may rely heavily on experienced individuals without recognising how much organisational capability would disappear if they left. What the institution knows and what the institution can locate are not necessarily the same thing.
That distinction becomes increasingly important as governments invest in digital knowledge systems and artificial intelligence. Digitisation can make enormous quantities of explicit information easier to search, combine and analyse. AI systems may make previously inaccessible documents discoverable or identify relationships across large bodies of recorded information. These are potentially powerful developments. But they do not mean that institutional knowledge has suddenly migrated into the digital system. The system can work only with knowledge that has become available to it in some usable form, while significant institutional understanding may remain embodied in experience, practice, relationships or context that was never captured.
The risk is not an argument against documentation or technology. On the contrary, institutions need strong records, accessible information and better ways of preserving knowledge. The mistake is to assume that building a sufficiently comprehensive repository will solve the institutional knowledge problem by itself. A repository can contain information without containing the institution’s full capacity to understand what that information means. Knowing which rule applies may require context. Understanding why the rule exists may require institutional memory. Recognising when the rule produces an unexpected consequence may require experience. Deciding what to do about it may require knowledge distributed across several different parts of the organisation.
For leaders, this suggests a different kind of knowledge map. Instead of asking only what information systems the institution possesses, they can ask where different forms of knowledge actually reside. Which capabilities depend heavily on particular individuals? Which lessons have been embedded in routines? Which important decisions rely on informal networks? Which knowledge has been documented but is difficult to find? Which information exists in one part of the institution but would be valuable elsewhere? Where would knowledge disappear if a team were reorganised or experienced staff departed?
These questions reveal that institutional knowledge is not simply an inventory. It has a geography. Some knowledge is concentrated, some dispersed; some is durable, some fragile; some is formally accessible, some practically invisible. Understanding that geography is essential because an institution cannot deliberately preserve, connect or mobilise knowledge whose location and form it does not recognise.
A government’s knowledge therefore lives wherever institutional experience has been preserved in a form capable of influencing future action: in people and records, but also in routines, roles, procedures, relationships and organisational arrangements. The challenge is not to force all of that knowledge into one place. It is to understand its distribution well enough that valuable knowledge does not become invisible simply because we have been looking for it in the wrong places.