Knowledge Drift

What it is

Knowledge drift is the gap that opens when published statements, internal documentation and actual product behaviour stop matching. It is gradual rather than sudden: a plan is renamed, a limit is raised, an integration is deprecated, and the pages describing the previous reality stay exactly as they were. No single edit causes it, which is why no single person is ever obviously at fault for it.

The name is in circulation but not settled. At least one knowledge base vendor now publishes a definition of knowledge drift in roughly this sense, and a semantic-layer vendor applies it to the data stack rather than to documents; the same failure is also called documentation drift or content rot. The useful parallel is configuration drift in infrastructure, where a deployed resource is judged to have drifted once its actual property values no longer match the expected ones. Machine-learning work uses the phrase for at least two other things: drift in a classifier's concept hierarchy, and a model's drift away from its original knowledge after exposure to false information. This register uses the content sense and treats it as a borrowed frame rather than a settled industry definition.

Why it matters

Drift is expensive precisely because it is quiet. It is the failure mode Knowledge Company was built to surface. It surfaces as a prospect quoting a price you no longer offer, a support ticket about a feature that moved, or an AI assistant confidently describing a two-year-old version of your product. Knowledge Company's answer is structural rather than editorial: connect content surfaces to the facts they depend on, so a change to the fact raises a flag instead of waiting for a human to notice the mismatch. That is a position rather than a consensus: the documentation literature prescribes clear ownership and integration into the normal review workflow, on the argument that documents without owners go stale.