When Knowledge Management Becomes a Graveyard for Knowledge

Walk into the knowledge base of almost any large organization and you meet a quiet paradox: an enormous quantity of stored knowledge sitting beside an enormous reluctance to use it. The wiki has ten thousand pages, and people ask the colleague at the next desk instead. The lessons-learned database is filled dutifully at the end of every project and opened at the start of none. The organization has succeeded almost completely at capture and failed almost completely at use.

This is not for want of effort or investment. Knowledge management as a discipline has spent decades building repositories, taxonomies, portals, and search. The intent was sound and the theory respectable: Walsh and Ungson taught that organizations retain knowledge across several stores, and it seemed natural to strengthen the one store we can engineer directly—the external, documented archive. If memory can be written down, surely more writing means more memory.

The flaw hides in that last step. It assumes storage is memory. It is not. Memory is the capacity to bring the right knowledge to bear at the right moment; storage is merely a precondition for it, and not a sufficient one. A document that cannot be found, is not trusted, or arrives stripped of the context that once made it meaningful is not institutional memory. It is a sedimentary layer. The archive grows steadily while the usable memory does not—and can even shrink, as the signal is buried under the sheer volume of the record.

Part of the difficulty is that the most valuable knowledge resists capture in the first place. Polanyi's observation that we know more than we can tell applies here with full force. The judgment of a seasoned engineer, the feel for a difficult client, the instinct for when a rule should be bent—none of this survives compression into a wiki page. What the system captures is the explicit residue of understanding, and it quietly trains the organization to mistake the residue for the thing itself.

Worse, the system rewards contribution and retrieval asymmetrically. People are asked, sometimes required, to contribute; almost no one is made responsible for whether anything is retrieved, curated, or retired. So the archive becomes write-only. Obsolete entries are never removed, because removal is nobody's job—a refrain these essays keep returning to—and the accumulating mass of stale, unverified, and duplicative material steadily erodes trust. Eventually searching the system feels less reliable than asking a person, and the expensive repository becomes ceremonial: maintained, audited, and consulted by no one who has a choice.

The outcome earns its blunt name. A knowledge base that ingests everything and surfaces nothing, that preserves the explicit husk while the living judgment walks out the door at five o'clock, is not a memory. It is a graveyard—orderly, well maintained, and full of things that used to be alive. The tragedy is that it looks like success. Every metric the initiative was given—pages created, documents stored, compliance with contribution policy—glows green while the underlying capability quietly starves.

There is a self-reinforcing quality to the decline that makes it hard to reverse. Once trust in the archive drops below a threshold, the best contributors stop contributing—why write a careful entry no one will read?—so the average quality of new material falls, which lowers trust further, which deters the next contributor. The system does not fail all at once; it hollows out gradually, keeping its full appearance of activity while losing its substance. Administrators, seeing usage fall, respond with more mandates to contribute, which adds volume without value and accelerates exactly the decline they meant to arrest. The intervention and the disease become indistinguishable, and the graveyard grows better tended and less visited with each passing quarter.

The lesson is not that knowledge management is worthless. It is that capture, by itself, is not a remedy, and can even be a trap, because it produces the reassuring appearance of memory without the substance. A store without judgment about what to keep, how far to trust it, and when to let it go does not preserve organizational knowledge so much as embalm it. Which points to a harder question these systems were never built to answer: what would it take to build not a warehouse of knowledge but a memory—one that knows what it holds, how much to trust each piece, and when to release what no longer serves? A better repository will not get us there. Something with judgment in it is needed, and judgment is precisely what a filing cabinet cannot supply.