I have been making two arguments simultaneously, and I am increasingly aware that they pull in different directions. The first argument, developed across the formal apparatus posts and the statechart work, is that service design would benefit from more formal state-space thinking - that making states explicit through structured representations reveals dynamics that action-oriented tools conceal. The second argument, developed through the institutional critique of technomagic and increasingly through fieldwork observations about how design understanding gets simplified for governance, is that formalisation is the mechanism through which contingent, situated design understanding gets flattened into fixed institutional categories. The first argument says: formalise. The second says: formalisation is dangerous. This post asks whether the tension is a contradiction or something more productive.
Formalisation as closure
The institutional story is familiar enough. A designer produces a rich, situated account of a service domain - states, transitions, dependencies, edge cases, the whole texture of how a system actually works and fails to work. This account enters a governance process: programme boards, business cases, delivery frameworks. The governance process requires simplification: what are the deliverables, what are the milestones, what are the success criteria? The rich model is compressed into a set of fixed categories - approved/rejected, on-track/at-risk, delivered/outstanding - that the governance apparatus can process. The statechart becomes a status report. The designer's domain model becomes a dashboard.
This is reification in the sense Berger and Luckmann (1966) described: the forgetting of human authorship. The categories that were provisional - constructed by the designer to understand the domain, held lightly, subject to revision as understanding deepened - become thing-like. They are treated as properties of the domain rather than as descriptions of it; they lose their contingency and acquire institutional weight. The framework that governs organisational action is not the designer's situated model but a politically determined simplification of it.
The mechanism is not mysterious. Institutional governance requires categories that are stable enough to plan around, accountable enough to report on, and simple enough to be understood by people who were not present for the design work that produced them. Hajer's (1993) "story-lines" operate here: once a particular framing of the problem has been accepted and institutionalised, it acquires its own momentum; the formalisation that was supposed to serve understanding becomes a constraint on understanding, because the institution has invested in treating the categories as settled.
Formalisation-as-closure is the institutional default. It is not that institutions are incapable of holding complexity; it is that their governance mechanisms - stage gates, benefits realisation, milestone tracking - select for simplification. The formal model enters the governance process as a tool and exits as a framework. The distinction matters: a tool is something the user controls; a framework is something that controls the user.
Formalisation as inquiry
But this is not the only thing formalisation does. The statechart work on the data access service did not produce reification; it produced revelation. The formalism - states, transitions, preconditions - forced questions that the existing journey maps had not forced: what triggers this transition? What happens if the application is neither approved nor rejected but suspended? Is there a path from "suspended" back to "active", or is suspension a terminal state? The answers to these questions surfaced design problems that had been invisible in the action-oriented representation. Dead-end states, unassigned transitions, parallel processes operating on different timescales - all of these were present in the service but absent from the journey map, because the journey map's representational vocabulary did not include the concepts needed to express them.
Formalisation in this register is not closure but opening. It makes the implicit explicit precisely so it can be examined, contested, and revised. The statechart was not treated as a settled description of the service; it was treated as a hypothesis - a provisional model whose value lay in the questions it forced rather than in the answers it provided. The formalism was a tool for seeing, not a description of reality. In the terms developed in the Service Aesthetics series, formalisation-as-inquiry redistributes the sensible: it makes perceptible what the previous representation had rendered invisible, opening new territory for design action.
Suchman (1987) made the classic argument that plans are not programs for action but "situated resources" - artefacts that help people coordinate but do not determine what they do. The same applies to formal state-space models: a statechart is a situated resource for design inquiry, not a specification of how the service works. Treated as inquiry, the formalism resists reification rather than producing it; it keeps the designer asking "is this right?" rather than accepting that it is settled.
The distinction that matters
The tension, then, is not between formalisation and its absence but between two uses of formalisation: closure and inquiry. The same statechart can serve either purpose depending on how it is held.
In formalisation-as-closure, the model is the output: governance receives a specification and operates on it. The model's provisionality is lost; the designer's uncertainty about whether the categories are adequate is not transmitted, because the governance apparatus does not have a vocabulary for "provisionally adequate". In formalisation-as-inquiry, the model is a moment in an ongoing process: it is constructed to be tested, revised, and reconstructed as understanding deepens. The model's provisionality is its point - it is a snapshot of current understanding, not a claim about the domain's permanent structure.
The distinction maps onto the planning/design distinction itself. Planning uses formalisms as given frameworks to operate within; design uses formalisms as provisional instruments to construct and reconstruct the framework. The same representational tool - a statechart, a state table, a transition diagram - can be used in either mode. What determines which mode is operative is not the tool but the institutional context in which it is used, and the designer's capacity to resist the pull toward closure.
The ontological question
There is a deeper question here that the series has been circling without confronting: what is the ontological status of the state spaces designers construct? Are they discovered, invented, or negotiated?
Dourish (2004) argues that ontological structures - categories, classifications, the fundamental entities that a system recognises - are not designed in any straightforward sense. They emerge from practice; they are "enacted" rather than specified. This complicates the series' claim that state spaces are "constructed, not given", because it suggests that the construction is not a free act of design but is constrained by the domain's own structure and by the social practices through which that structure is enacted. You cannot simply decide what the states of a rehabilitation service are; the states emerge from how rehabilitation is practiced, governed, experienced, and contested.
Burns and Hajdukiewicz (2017), working in the ecological interface design tradition, provide a different and I think more useful framing. Their argument is that work domains have constraint structures - physical, functional, and organisational - that exist independently of any particular worker or designer. The designer's task is not to invent these constraints but to make them visible through interface design, so that practitioners can perceive the constraints they need to act upon. Applied to state-space construction, this means that the designer is not freely constructing an ontology but revealing a constraint structure that already shapes the domain, albeit implicitly. The state space is partly discovered (the rehabilitation service has real states that exist whether or not anyone has named them - a person either has employment or does not, a referral either exists or does not) and partly constructed (the decision to group certain conditions into one "state" and distinguish them from another is a modelling choice with consequences). The variables are constrained by the domain's reality; the granularity and boundaries are constructed by the designer's modelling decisions.
Folkmann (2013) describes this as "the dialectic between the openness of the possible and the closure of form". Design operates in the tension between what is possible (the open space of potential representations) and what is actual (the domain's constraint structure). State-space construction is a specific instance of this dialectic: the possible representations are many; the domain's constraints narrow the field; the designer's judgement determines which representation, from the narrowed field, best serves the purposes at hand.
A more nuanced position
The series' claim - "constructed, not given" - remains correct but is now better articulated as: partly discovered (the domain has a constraint structure that shapes what states are possible), partly constructed (the selection of which constraints to represent, at what granularity, through which formalism, is a design decision), and partly negotiated (different stakeholders with different conceptual spaces will construct different state spaces from the same domain, and resolving these differences is a political process). This three-part formulation - discovered, constructed, negotiated - is more nuanced than "constructed, not given" and more accurate to what the series has described. It accommodates Dourish's concern about emergence, Burns and Hajdukiewicz's concern about constraint structures, and the series' own arguments about the politics of representation.
The designer's task, then, is to keep formalisation in the inquiry register - to resist the institutional pull toward closure, to maintain the model's provisionality, to treat the state space as a hypothesis under continuous test rather than as a settled description. This is what evaluative capacity - the designer's practical judgement about when a model is adequate and when it distorts - makes possible. The evaluative capacity is not an add-on to the formal apparatus; it is what makes the formal apparatus a design tool rather than a bureaucratic one. Without it, formalisation produces reification; with it, formalisation enables inquiry.
References
Berger, P. and Luckmann, T. (1966). The Social Construction of Reality. Penguin.
Burns, C. and Hajdukiewicz, J. (2017). Ecological Interface Design. CRC Press.
Dourish, P. (2004). Where the Action Is: The Foundations of Embodied Interaction. MIT Press.
Folkmann, M.N. (2013). The Aesthetics of Imagination in Design. MIT Press.
Hajer, M. (1993). Discourse Coalitions and the Institutionalisation of Practice: The Case of Acid Rain in Great Britain. In Fischer, F. and Forester, J. (eds), The Argumentative Turn in Policy Analysis and Planning. Duke University Press.
Suchman, L. (1987). Plans and Situated Actions: The Problem of Human-Machine Communication. Cambridge University Press.