> ## Content Index > Fetch the complete content index at: https://www.ariet.net/llms.txt > Use this file to discover other available public pages before exploring further. # Chapter 2: What we design, and what we don't - URL: https://www.ariet.net/chapter-2-what-we-design-and-what-we-dont/ - Published: 2026-10-06T16:29:35.000Z - Updated: 2026-10-06T16:29:35.000Z - Author: Richard Thackeray - Tags: Book ## So far > "The purpose of a system is what it does." Stafford Beer. Chapter 1 made a single argument: operating model redesigns keep disappointing not because they are badly designed, but because they are aimed at the wrong layer. The formal model specifies structure, process, governance and incentives. Beneath it, the adaptive system reads the actual signals in the environment and responds to those, regardless of what the model says. The NHS National Programme for IT is one clear example: technically coherent architecture, adoption failed because the conditions underneath it were not addressed. We have been designing the wrong thing, with significant effort and intent. This chapter looks more closely at what the formal model contains, how large the unspecified layer really is, and why we keep adding to the formal layer even when we know it isn't working. ## And onto chapter 2…… There is a moment, familiar to anyone who has done this kind of work, when the operating model leaves the room where it was designed and enters the organisation it was designed for. It has been through workshops, reviewed by stakeholders, signed off by the right people. It represents weeks or months of careful thinking. And then it meets people. What happens next is not what the model anticipated. Not because people are obstructive, though some will be. Not because the design was wrong, though parts of it may be. But because the model was built on an assumption that does not hold: that organisations change when you change their formal design. They don't. Or not in the way the model assumed, and not at the pace the plan projected. --- What the formal operating model specifies is a thin layer of intended behaviour. Structures: who reports to whom, how teams are organised, where accountability sits. Processes; how work is supposed to flow, what steps it passes through, who approves what. Governance: how decisions are made, how performance is reviewed, what gets measured. Incentives; what the organisation says it will reward. These things matter. They are not nothing. An organisation with no clarity on any of them will struggle in ways that are obvious and correctable. The formal layer is necessary. It is also, by itself, insufficient. Because beneath it, and significantly larger than it, is everything the formal model does not and cannot specify. Think about what an operating model specifies against what a day at contains. The model might specify thirty pages of structure, process and governance. It says nothing about the conversation that happened before the meeting, the relationship that determined whose idea got heard, the unspoken rule about what it is safe to raise with which senior leader, the norm that has developed around how feedback is given (or avoided), the workaround that three teams independently invented and never compared, the decision that gets made in a corridor because the forum that should make it takes six weeks to convene. These are not edge cases. They are the substance of organisational life. The formal model touches a fraction of them. The rest are governed by the adaptive system underneath: by what people have learned to expect, by the signals the environment sends about what is safe and what is not, by the accumulated experience of what happens when you do what the model says versus what you learn to do instead. The disproportion matters because the thing that gets designed is the small part, and the thing that gets left to emerge is the large part. And what emerges is not random. It is shaped by the conditions the organisation has created, by its history, by the behaviour of the people with the most influence over those conditions. It is, in other words, designable. But not through the formal model. --- Human systems are adaptive. This is one of their greatest strengths and, in the context of operating model design, the source of most of the disappointment. The formal model specifies what the organisation should do. The adaptive system underneath determines what it does. And the signals the system reads are not produced by the model; they are produced by everything the organisation does, has done, and has learned to expect. Those signals were present long before the new design arrived, and they do not update simply because a new design has been announced. Stafford Beer saw this clearly, writing about it fifty years ago. Modern institutions, he observed, accumulate policy and process (each addition rational in isolation, each one a response to a real problem) until the mechanisms designed to help become the thing people have to survive. The compensation layer grows. The original purpose gets buried. And the people inside the system develop the particular exhaustion that comes from spending most of their energy navigating the system rather than doing the work the system was supposed to enable. We read that observation and recognise it. We have been building the thing he described ever since. The reason we keep building it is not ignorance. It is that each addition to the formal layer is individually rational. A risk surfaces; a control is added. An accountability is unclear; a governance forum is created. A process produces inconsistent results; a standard is mandated. Nobody in the room, when any of these decisions is made, is wrong to make it. The problem is cumulative and systemic, not individual and intentional. The organisation that emerges from a decade of these decisions is one where the mechanisms of assurance have become the primary work. Where people spend more of their time in governance than in the work governance is meant to assure. Where the energy available for adapting and responding and thinking has been consumed by the energy required to report, comply, escalate and document. Mechanical control is cheap to add and expensive to live with. Each unit of control costs very little in the moment and compounds quietly over time. The cost arrives later, diffused across the system, hard to attribute to any individual decision. Which is precisely why it keeps being added. --- The implication is uncomfortable, and worth stating. If behaviour is determined primarily by signals rather than formal design, and if those signals are produced by something the operating model does not specify, then the operating model is not the right place to intervene. It is necessary but not sufficient. And in organisations that treat it as sufficient (that commission the design, implement the structure, and then wonder why the behaviour hasn't changed) it becomes an expensive distraction from the work that would make a difference. The question is what that work is. What produces the signals? What determines whether the adaptive system underneath the formal model orients toward the organisation's purpose or away from it? What is the layer that the operating model consistently leaves out? Every practitioner who has worked in this territory has their own version of the same moment. The strategy that landed differently than intended. The restructure that produced the same dynamics in new boxes. The values programme that the organisation went through the motions of, while privately everyone understood that the real values were the ones that got you promoted. These are not failures of execution. They are the adaptive system behaving exactly as adaptive systems behave: reading the signals in the environment and responding to those, regardless of what the formal model says. The question this chapter is opening is not rhetorical. What produces the signals? What determines whether the system underneath the formal model orients toward the organisation's purpose or gradually away from it? What is the layer that operating model design has never quite learned to include? The next chapter begins to answer it. --- ## A moment to pause ### If you are living inside this system: the gap between the formal model and the lived experience is something you know from the inside. The policy that nobody follows, the process that everyone has a workaround for, the values statement that bears no relationship to what gets rewarded. That gap is not a failure of the people in the system. It is the adaptive system doing what adaptive systems do. The question worth sitting with is what the signals around you are conditioning you toward, and whether that is what the organisation needs. ### If you are leading: the signals your organisation reads are not the ones you intend. They are the ones produced by everything you do and don't do, every day. The operating model you signed off tells people what you say matters. What you pay attention to, what you protect, what you let slide; those tell people what matters. The gap between those two things is where the drift begins. ### If you are working to change this: the question this chapter leaves open (what produces the signals, what is the layer the operating model consistently leaves out) is the question the rest of the book answers. This chapter has described the problem. The next chapter begins to explore what lies beneath it.