What Each Body of Research Found
On why documentation is written the way it is. Jay Galbraith’s information-processing view of organisation design frames documentation as one of several possible responses to rising coordination demand as an organisation grows and its tasks become more interdependent and uncertain. On this view, documentation exists to reduce the information-processing burden on any one person by encoding shared knowledge in a form that does not depend on any individual being present to transmit it directly. This explains why documentation becomes more prevalent as organisations scale — it is substituting for the direct, informal coordination that worked when the organisation was smaller — but it says relatively little about whether the documentation, once written, actually gets followed.
On the gap between documented and actual practice. This is the specific territory examined by Erik Hollnagel, Robert Wears and Jeffrey Braithwaite, in a white paper jointly published by the University of Southern Denmark, the University of Florida and Macquarie University. Their central argument is that things generally go well in complex operational settings not because people strictly follow documented procedure, but because people make sensible, situational adjustments to the actual demands in front of them — adjustments that a static document, written in advance for a generic situation, cannot anticipate. Their finding implies something uncomfortable for the assumption that documentation drives consistency: the document describes one version of the work, and the work, as actually performed, routinely and often necessarily diverges from it, in ways that are frequently invisible until something forces the gap into view.
On why documentation stops being maintained. Nelson Repenning and John Sterman of MIT Sloan School of Management, in their research on the capability trap, offer a third angle. Their finding is that organisations under sustained pressure default to visible, immediate effort over the less visible, structural work of maintaining systems — including documentation — over time. Keeping documentation aligned with an evolving reality requires ongoing attention that produces no immediate, visible payoff, which is exactly the kind of investment their research shows gets crowded out under pressure. This offers a direct explanation for why documentation, even when accurate at the point it was written, so reliably drifts out of alignment with practice as time passes.
Where the Three Bodies of Research Agree
All three converge on a shared implication, even though each approaches the question from a different angle: documentation is not, on its own, a mechanism that produces consistency. It is, at best, a snapshot — a representation of the work at the moment it was written, produced in response to a genuine coordination need (Galbraith), immediately and continuously subject to divergence from actual practice (Hollnagel, Wears & Braithwaite), and prone to falling further out of alignment over time as the ongoing maintenance it requires competes with more visible priorities (Repenning & Sterman). None of the three bodies of work suggests documentation is worthless. All three suggest that the common assumption — write it down, and consistency follows — treats documentation as a solution when the evidence more consistently treats it as one input into a considerably more dynamic and ongoing situation.
Where They Diverge
The three perspectives diverge on where the more productive point of attention actually sits. Galbraith’s framing points toward structure and coordination design — the question of whether documentation is even the right tool for a given level of task uncertainty, or whether some other coordination mechanism would serve better. The work-as-imagined-versus-work-as-done research points toward the value of understanding actual practice directly, rather than assuming the document is an accurate description of it. The capability-trap research points toward organisational incentives — specifically, why ongoing maintenance work so reliably loses out to visible, immediate effort, regardless of how good the original documentation was. These are not contradictory findings, but they do not converge on a single, unified account of what documentation is actually for, which is itself a meaningful finding: the role documentation plays in consistency appears to be genuinely multi-dimensional, not reducible to a single mechanism.
What Remains Unresolved
None of the three bodies of research, individually or together, resolves the practical question of how an organisation should decide when documentation is worth writing, updating, or retiring — nor should this article attempt to answer that question, since doing so would move from evidence synthesis into prescriptive methodology. What the research does establish, fairly clearly, is that the common, largely unexamined assumption — that documentation is inherently a consistency mechanism, and that inconsistency therefore indicates a documentation gap — does not hold up well against any of the three bodies of evidence surveyed here. Inconsistency can just as easily indicate a coordination structure mismatched to actual task uncertainty, a document that never reflected real practice in the first place, or a maintenance process that lost out to more visible priorities. Distinguishing between these before assuming the answer is “write more documentation” appears, on the evidence, to be a meaningfully different and more useful starting point.
There is a further implication worth drawing out from the synthesis as a whole. If documentation is genuinely a snapshot rather than a mechanism, then the organisational question worth asking is not simply “is our documentation accurate” but “how large is the gap currently, and is it growing.” A snapshot taken recently, in a relatively stable operational environment, may still be reasonably close to actual practice. The same snapshot, taken years ago in an environment that has since changed substantially — new systems, new people, new external conditions — is very likely to have drifted considerably further, even if nobody has updated it and it therefore looks, on paper, exactly as authoritative as it did the day it was written. Age and stability of the underlying operation, considered together, may be a more useful signal of how much a given piece of documentation can currently be trusted than anything about the document itself.
A final point worth drawing from the synthesis: documentation’s role appears to differ by what it is actually being asked to do. Where documentation exists mainly to transmit context that would otherwise require a person’s direct presence — Galbraith’s framing — its value holds up reasonably well even as it ages, because the underlying context it encodes may not have changed much. Where documentation exists to specify precise operational steps in a fast-changing environment, the Hollnagel, Wears and Braithwaite research suggests it is far more exposed to becoming misleading quickly, because the situational adjustments real work requires accumulate faster than any static document can be revised to reflect them. The same word — “documentation” — is doing meaningfully different work in each case, which may be part of why its relationship to consistency is so much less straightforward than the common assumption suggests.
The same pattern appears across many kinds of organisation; see Other Sectors for how it is read elsewhere.
Questions worth considering
- When documentation and actual practice diverge in this organisation, has anyone examined why — is it a coordination-structure mismatch, an inherently generic document, or a maintenance gap?
- How long, on average, does documentation in this organisation stay aligned with practice before drift becomes noticeable?
- Is documentation currently treated as a finished artefact, or as something requiring ongoing, resourced maintenance?
This perspective is part of Operations.
Your team is facing similar challenges?
Subscribe to The Riddle Ledger
Perspectives are sent as they are published, with no promotional email, and you can unsubscribe at any time.