A Luhmannian reading of Safety Culture Maturity Models
The poetry of safety culture vs. the reality of organisational evolution
Safety culture maturity models as reform poetry
Safety culture maturity models present themselves as linear, developmental, progressive, and normatively ordered toward a better state. From a system perspective, these models are semantic constructions that depict the present as deficient, construct a desirable future, and position reform as the bridge between the two. This corresponds to what sociologist Niklas Luhmann called the paradox of reform; i.e. the past must be portrayed as insufficient so that the future can be imagined as better—yet neither description is stable when time moves on. As Luhmann notes, “the transformation paradox … is reformulated as a tautology: it succeeds if it succeeds” (Luhmann, 2018, p. 273). Thus, the levels of safety culture are not empirical stages, although Ron Westrum originally developed them as a three-part categorisation continuum (Westrum, 2014). The levels of safety culture are rhetorical devices that turn complexity into a narrative of improvement, what Luhmann describes as “the poetry of reform … geared fully to consensual constructions” (Luhmann, 2018; p. 278).
The illusion of linear development
Maturity models assume that organisations progress through stages, that higher levels are more advanced, and that improvement is cumulative and directional. Luhmann would reject this orientation for several reasons. First of all, organisations are not linear but self-referential, operationally closed, and historically contingent. This means they do not move smoothly from stage to stage but instead reproduce themselves through decisions. Second, structural change should not be seen as cumulative, because changes “do not run as planned; in the course of time they lose their plannedness” (Luhmann, 2018, p. 283), are often forgotten, reversed, or reinterpreted, and may coexist inconsistently in different parts of the organisation. Granted, some of the maturity models’ propagators acknowledge internal variation, but the maturity framework still imposes a single developmental axis. From a systems perspective, this is a simplification that hides the simultaneity of contradictory structures.
The overestimation of intentional change
Safety culture frameworks assume that culture can be diagnosed, its level can be assessed, and it can be improved through intent. But Luhmann criticizes in reform thinking that intentions are not decisive. Even if leadership intends to move toward proactive culture, outcomes depend on how the organisation processes these signals internally, not on the intention itself. Second, loose coupling undermines control. Policies, communication, and procedures are only loosely coupled to actual behaviour. As Luhmann puts it, “the stability of the system depends rather on loose couplings … so that operations not subject to direct intervention can survive reform without being affected” (Luhmann, 2018, p. 274). Workers, teams, and subunits interpret reforms differently. Thus, a generative culture can’t be engineered, only described after the fact.
Culture as decision premises—and their limits
In Luhmann’s terms, culture is a set of decision premises; i.e. shared expectations that guide decisions. Safety culture models treat culture as if it directly influences behaviour and reliably produces safety outcomes. But for Luhmann, “decision premises and decisions … are loosely coupled,” (Luhmann, 2018; p. 280) meaning that good culture does not guarantee good decisions, and poor safety outcomes may occur even in advanced cultures. This undermines the core assumption that improving culture leads predictably to improved performance.
The role of observation: measurement creates reality
The framework constructs interview-based descriptions, consensus representations, and self-assessment tools. From a Luhmannian perspective, these are operations of observation, but not direct access to culture. As he argues, “organizational change is always observed change,” and “unobserved change is not change, because the system cannot react to it” (Luhmann, 2018; p. 273). Safety culture levels exist because organisations observe themselves through this framework. Thus, the model creates the reality it claims to measure, by stabilising certain distinctions—from pathological to generative, from weak to strong, and from immature to mature.
The function of maturity models: producing reform dynamics
Rather than judging whether these models are true, Luhmann would ask what function they serve. First of all, they create differences between current and desired states, between units (sites, teams, hierarchies), and between leaders and laggards. Second, they generate communication—workshops, audits, self-assessments, and improvement discussions. Third, they legitimise intervention; reform becomes necessary, and management action becomes justified. Thus, while safety culture models do not primarily improve safety, they do produce continuous reform activity. This aligns with Luhmann’s claim that reforms “serve not to attain their goals but to maintain this dynamic” (Luhmann, 2018; p. 278).
The paradox of success
Maturity models imply that organisations can reach proactive or generative culture and thereby achieve high reliability. Luhmann would point out, first, that success is retrospectively constructed: what is called generative culture is recognised after favorable outcomes, not before. Second, success and failure are interpretive, as the same state can be seen as a success (safety maintained), or failure (risks overlooked). As he notes, “there is a blurring between ‘success’ and ‘failure’,” and more sharply: “achieving success is much more difficult than claiming success” (Luhmann, 2018; p. 280). Thus, achieving a generative culture may be less important than claiming it.
The neglect of evolution
The strongest critique is that maturity models privilege planned change (reform), and neglect unplanned structural change (evolution). Most change arises from daily decisions, local adaptations, unintended consequences, and selective retention and forgetting. Safety culture models treat variation as something to be controlled, rather than as the source of evolution. But for Luhmann, real structural change emerges from processes that “cannot be predicted and cannot be planned but have to be treated as random” (Luhmann, 2018, p. 286), namely the uncoordinated interplay of variation, selection, and stabilisation.
The deeper illusion: control over the future
At its core, the safety culture maturity approach assumes the future can be shaped, risks can be managed, and organizations can become high reliability systems. Luhmann’s response is more radical: “the future … is unknown … but survives … every attempt to determine it” (Luhmann, 2018, p. 278). Reforms, in this sense, “treat the indeterminable as determined by them” (ibidem). Thus, labeling interventions as proactive or generative does not reduce uncertainty; it merely re-describes it in optimistic terms.
Final synthesis
From a Luhmannian perspective, safety culture maturity models are best understood not as communicative devices that transform uncertainty into structured expectations of progress. They simplify complexity into stages, translate ambiguity into measurable categories, sustain continuous reform efforts, and provide a language for claiming improvement. As Luhmann suggests, “the main result of reforms could be to ascertain the need for further reforms. Ecclesia reformata semper reformanda.” In short, safety culture maturity models are the poetry of safety transformation—while actual organisational change unfolds as messy, contingent, and evolutionary.
Reference:
Luhmann, N., Barrett, R., 2018. Organization and Decision. Cambridge/New York/Melbourne/New Delhi/Singapore: Cambridge University Press.
Westrum, R., 2014. The study of information flow: a personal journey. Safety Science. Vol. 67: pp. 58-63.