Institutions often preserve their past in the rules they continue to follow. A procedure introduced after an earlier failure may survive for decades, a requirement created to prevent a particular mistake may become part of ordinary administration, and a sequence of approvals, checks or consultations can continue shaping behaviour long after the people who originally learned the lesson have left. This persistence can be one of the great advantages of institutional memory because it allows an organisation to benefit from earlier experience without requiring each generation to rediscover every lesson for itself. Yet there is an important complication hidden inside that apparent success: a rule can survive more effectively than the understanding that once made the rule sensible, leaving the institution able to remember what it is supposed to do while gradually losing its ability to explain why.

At first, this condition may not resemble forgetting at all. The procedure remains visible, employees continue to follow it, compliance may be high and the relevant forms, workflow steps or technical controls remain embedded in everyday operations. Something learned in the past is therefore still influencing behaviour in the present, which is precisely what institutional memory is expected to achieve. The difficulty is that behavioural persistence and rationale persistence are not the same thing. If the reasoning behind a procedure disappears, the institution loses part of its ability to interpret what the procedure means, recognise when the conditions that justified it have changed and decide intelligently whether the inherited rule should still possess authority.

Consider a safeguard introduced after a serious operational failure. The institution studies what happened, identifies a vulnerability and creates a new requirement intended to prevent the same problem from recurring. Over time the safeguard becomes routine, the original incident recedes from organisational memory and successive generations of staff learn the requirement simply as part of the process. The procedure may remain entirely appropriate, but the institution’s relationship with it has nevertheless changed. What began as a reasoned response to experience has gradually become inherited compliance, because people can reproduce the behaviour generated by the original lesson without necessarily being able to reconstruct the lesson itself.

This creates two failure possibilities that look similar from the outside but have very different implications. In the first, the rule remains useful even though its rationale has been forgotten. The institution is still doing something sensible, but it no longer understands the knowledge embodied in what it does. That loss becomes important whenever the procedure must be interpreted, defended, adapted or applied to an unusual case, because the people responsible can see the visible instruction without seeing the cognitive architecture behind it. In the second possibility, the original conditions themselves have changed and the procedure has survived beyond the problem it was designed to solve. What once represented institutional learning may then become an obsolete solution whose continued existence reflects persistence rather than present judgement.

Institutions are particularly exposed to this second condition because they accumulate procedures as they accumulate history. Rules emerge in response to particular incidents, technologies, organisational arrangements, political priorities and social circumstances, while the administrative mechanisms that preserve those rules often last longer than the environments that produced them. A reporting requirement, for example, may have been introduced when information moved slowly between organisational units and managers needed a particular document to maintain coordination. Years later, shared information systems may have transformed that environment while the reporting requirement remains untouched, leaving staff to reproduce a solution to a coordination problem that no longer exists in its original form.

None of this means that an old procedure should be regarded as an obsolete one. Indeed, the opposite mistake can be equally serious. A safeguard that has successfully prevented a particular failure for many years may eventually make the failure itself disappear from organisational experience, at which point the continuing absence of the problem can paradoxically make the safeguard look unnecessary. If nobody remembers why the control was created, removing it may recreate precisely the vulnerability that earlier institutional learning had managed to suppress. Institutional memory therefore has to protect organisations from two symmetrical errors: preserving inherited arrangements automatically because their origins have been forgotten, and discarding inherited arrangements casually because their success has made those origins difficult to see.

The ability to reconstruct historical purpose becomes crucial precisely because it allows an institution to distinguish between these possibilities. Rather than asking only what the rule requires, an institution needs some capacity to recover what the rule was originally trying to accomplish, what failure or constraint produced it, which assumptions made it reasonable and whether those assumptions still hold. This does not require every employee to become an organisational historian, nor does it imply that every minor procedure should carry an elaborate documentary genealogy. It means that consequential institutional arrangements should retain enough interpretive context for their rationale to become recoverable when circumstances make that rationale relevant again.

Reasons are often harder to preserve than rules. A procedure can be copied into a new manual, incorporated into a workflow or transferred into software with relative ease, while the reasoning that produced it may remain scattered across old reports, meeting records, informal recollections or assumptions that once seemed too obvious to document. Over time, the visible instruction can therefore become increasingly durable while the invisible explanation fades. The institution continues to reproduce the outcome of earlier cognition, but the connection between that outcome and the experience from which it emerged becomes progressively more difficult to recover.

Digitalisation can intensify this asymmetry because technical systems are exceptionally good at preserving behavioural constraints. A software application can require a second approval, prevent a transaction without a particular check or force users to complete a field before proceeding, thereby reproducing an institutional decision with remarkable consistency. Yet the interface may offer little indication of why the constraint exists, what historical failure produced it or which assumptions originally justified it. The institution may therefore become better at remembering what must happen while becoming worse at understanding why it must happen, because the historical judgement has migrated from an interpretable organisational practice into an opaque technical requirement.

Artificial intelligence can raise a related issue when automated processes inherit classifications, thresholds or decision rules whose institutional origins are poorly understood. Automation can make inherited choices more consistent, scalable and consequential without making the reasoning behind them any easier to reconstruct. The central concern is not that technological systems necessarily erase institutional memory, but that persistence without interpretability can allow historical assumptions to exercise increasing influence at precisely the moment when the institution becomes less capable of judging whether those assumptions remain appropriate.

The answer is not continual reconsideration of everything inherited from the past. Institutions require stability, and a memory system that forced every generation to reopen every earlier decision would undermine one of the principal reasons for having institutional memory in the first place. The more useful principle is that persistence should preserve the possibility of interpretation. An institution should be able to rely routinely on inherited arrangements while retaining enough of their rationale to interrogate them when circumstances change, anomalies appear or revision becomes necessary. In that sense, the past remains cognitively useful not merely because it constrains present behaviour, but because present actors retain some capacity to understand what the past encoded.

This distinction reveals why successful institutional memory cannot be measured simply by asking whether something from the past has survived. Behaviour may survive while understanding disappears, and a procedure may remain perfectly intact while the institution gradually loses the capacity to distinguish accumulated wisdom from accumulated inertia. The problem is not yet one of deliberate institutional unlearning, because losing the rationale behind a rule is different from deciding that earlier knowledge should no longer govern present action. It is a more basic failure of memory architecture in which the institution has preserved the consequence of earlier cognition while allowing too much of its meaning to disappear.

A durable institutional memory therefore requires more than continuity. It requires enough continuity of meaning for the present to remain capable of interpreting what it has inherited, because only then can an institution distinguish a safeguard worth preserving from a solution that has outlived its problem. An institution has not fully preserved a lesson merely because it continues obeying the rule that the lesson once produced; the deeper achievement is to preserve enough of the reasoning behind that rule for future generations to understand why the past still deserves to influence what they do.