The decisive organisation requires an architecture function that is designed differently from the one most organisations currently operate. The difference is not size and it is not seniority. The reflex, when an organisation concludes that its architecture function is not producing what it needs, is to make the function larger or more senior — to add architects, to elevate the chief architect, to give the function a more prominent place in the governance hierarchy. Neither change addresses the actual defect, because the defect is not that the function is too small or too junior. It is that the function is designed for the wrong work. The architecture function that most organisations operate is designed to produce and maintain artefacts. The architecture function the decisive organisation requires is designed to maintain the conditions under which better choices are continuously possible. These are different jobs, requiring different scopes, different instruments, different relationships to delivery and governance, and different measures of success — and an organisation that makes its artefact-producing function larger or more senior has produced a larger or more senior artefact-producing function, not the function the decisive organisation needs.
The distinction has a precise form, and it returns to a figure introduced in the discussion of the repository: the librarian. The architecture function designed to produce and maintain artefacts is a library, and its architects are librarians. Their work is to acquire, catalogue, organise, and preserve the documents that record the organisation’s architecture — the principles, the standards, the reference architectures, the decision records, the diagrams. They maintain the collection. They ensure it is complete, current, and accessible. They are valued for the quality and comprehensiveness of the collection they maintain, and they measure their success by the collection’s condition. This is honourable work and it is the wrong work, because the collection is not the architecture. The collection is the record of the architecture, and a function whose purpose is the maintenance of the record has confused the evidence of architecture with its substance. The librarian maintains a perfect record of decisions that may or may not be governing anything, and the perfection of the record is no evidence at all that the decisions are alive in the system the record describes.
The architecture function the decisive organisation requires is not a library and its architects are not librarians. Their work is not the maintenance of a collection but the maintenance of a set of conditions — the conditions, described throughout this book, under which ambiguity expires at the rate the organisation requires. The redesigned function maintains the authority design, ensuring that holders continue to hold and that the positions that should close questions retain the decision rights, cost-absorption, and signal access that closing requires. It maintains the decision infrastructure, ensuring that the closures holders produce are captured, held current, and made accessible. It maintains the Clarity Stack, ensuring that principles remain current, decisions remain held, and constraints remain translated. It maintains the Pulse System, ensuring that the rhythm synchronises the altitudes against current pictures. The redesigned function’s work is the continuous maintenance of these conditions against the delivery pressure and organisational drift that continuously erode them, and its success is measured not by the condition of any collection but by whether the conditions it maintains are producing the velocity they exist to produce.
The scope of the redesigned function follows from this purpose, and it is both narrower and deeper than the scope of the librarian function. It is narrower because the redesigned function does not attempt to document everything. The librarian function expands toward comprehensiveness — it wants the collection to be complete, to record every system, every decision, every standard, every relationship, because completeness is the librarian’s measure of quality. The redesigned function resists comprehensiveness, because comprehensiveness is not its purpose and the effort spent pursuing it is effort taken from the maintenance of the conditions. The redesigned function documents what the conditions require to be documented — the live decisions, the current constraints, the active principles — and nothing more, because everything more is collection-maintenance that does not serve the conditions. And it is deeper because the maintenance of the conditions requires the function to engage with the substance of decisions rather than merely their recording. The librarian records a decision without needing to understand whether it is the right decision or whether it is being honoured. The redesigned function must understand both, because its job is to maintain the conditions under which the decision was made well and is being honoured, and it cannot maintain conditions it does not understand.
The redesigned function’s relationship to delivery is the relationship that most distinguishes it from the librarian function, and it is a relationship of presence rather than review. The librarian function relates to delivery through review — delivery produces something, the function reviews it against the collection, the function approves or requires changes. This relationship positions the function as a gate that delivery must pass through, and the gate is experienced by delivery as an obstacle, because it arrives after the work is done and can only approve or obstruct. The redesigned function relates to delivery through presence — it maintains the constraints that delivery encounters before it chooses, so that the function’s influence is felt at the point of choice rather than at the point of review. The redesigned function does not gate delivery. It shapes the choice space delivery operates in, so that delivery’s choices are made inside the constraints the architecture requires without the function needing to review them afterward. This is the difference between governing through design and governing through enforcement that the Clarity Stack chapter described, applied to the architecture function itself. The librarian enforces. The redesigned function designs, and the design makes most enforcement unnecessary, because the constraints are present at the point of choice and the choices made inside them do not need to be reviewed against the collection afterward.
The redesigned function’s relationship to governance is similarly transformed. The librarian function serves governance by preparing and presenting artefacts — the documents the forums review, the records the governance process produces, the materials that make the governance visible. It is the staff of the governance theatre, producing the props the performance requires. The redesigned function serves governance by maintaining the instruments that make governance decisive — the decision aging record, the ownership velocity index, the epistemic integrity score, the translation completeness measure. It does not prepare artefacts for the forums to review. It maintains the instruments that determine the forums’ agendas, surface the questions that require decision, and read whether the decisions are producing the velocity they should. The redesigned function’s product is not the materials of governance but the conditions of governance — the instrumented, synchronised, measured environment in which the governance architecture can be decisive. The forums do not consume the redesigned function’s artefacts. They operate inside the conditions the redesigned function maintains.
The measures of success for the redesigned function are the measures of the decisive organisation that the preceding chapter described, applied to the function’s own work. The librarian function measures its success by the condition of its collection — its completeness, its currency, its accessibility, its quality. These measures can all be satisfied while the organisation governs nothing, because they measure the collection rather than the conditions. The redesigned function measures its success by the rate at which ambiguity expires, the distribution of decision latency across altitudes, the ownership velocity index, the translation completeness, the epistemic integrity score — the measures of whether the conditions it maintains are producing the velocity they exist to produce. This is a more demanding standard, because it cannot be satisfied by the function’s own activity. A librarian function can succeed by its own measures through its own effort, by maintaining its collection well. The redesigned function cannot succeed by its own effort alone, because the velocity it is measured by is produced by the whole organisation operating inside the conditions, and the function’s success depends on whether those conditions are actually producing velocity, which depends on more than the function’s diligence. The redesigned function is measured by an outcome it influences but does not solely control, and this is uncomfortable, and it is correct, because the function’s purpose is the outcome and a function measured by anything other than its purpose will optimise for the measure rather than the purpose.
The redesign asks something of architects that the librarian function does not, and it is worth naming because it is the part of the transition that the architects themselves find hardest. The librarian function offers a form of safety: the architect who maintains the collection well has done their job, regardless of what the organisation does with the collection. The architect’s success is within the architect’s control, and the architect can be excellent by the function’s measures while the organisation fails by every measure that matters. The redesigned function removes this safety. The architect in the redesigned function is responsible for conditions whose success depends on the whole organisation, and the architect cannot retreat into the maintenance of a collection that is excellent regardless of outcomes. This is a harder professional position, and it is also the position that the profession’s own claims have always implied. Architecture has always claimed to be responsible for the organisation’s clarity, its direction, its structural soundness — claims about outcomes, not about collections. The redesigned function is the function that takes the profession’s own claims seriously enough to be measured by them, and the discomfort the redesign produces is the discomfort of a profession being asked to be accountable for the outcomes it has always claimed to produce.
The redesign changes who the function hires, and the change is significant enough that an organisation cannot staff the redesigned function from the same pool that staffed the librarian one without retraining or reselection. The librarian function selected for people good at producing and maintaining artefacts — thorough, rigorous, comprehensive, oriented toward completeness. The redesigned function selects for people good at maintaining conditions — judicious, decisive, oriented toward outcomes, comfortable with the discomfort of being accountable for a result they influence but do not solely control. These are different people, or at least people exercising different capacities, and an organisation that redesigns its function without addressing its staffing will find that the people who excelled at the librarian work struggle with the condition-maintenance work, because the work rewards a different orientation than the one they were selected and developed for. This is among the hardest parts of the redesign, because it is not a matter of restructuring the function on a chart but of changing what the function’s people do every day, and some of the people who were excellent at the old work will not be excellent at the new, through no fault of their own, because they were selected for a different job than the one the redesigned function asks them to do.
The redesigned function is smaller and deeper than the librarian function, and the combination is counterintuitive enough that organisations resist it. Smaller, because the function no longer pursues the comprehensiveness that the librarian function pursued, and the abandonment of comprehensiveness removes a great deal of work — the documentation that governed nothing, the artefacts that were produced because production was rewarded, the maintenance of a collection whose completeness was its own justification. Deeper, because the work that remains requires the function to engage with the substance of decisions rather than their recording, which is more demanding per unit of work than the documentation it replaces. The redesigned function does less and understands more, and an organisation accustomed to measuring its architecture function by its size will read the smaller function as a diminishment, when it is in fact a concentration — the same or greater architectural capability, directed at fewer things, each engaged more deeply. The instinct to grow the architecture function in response to architectural problems is the instinct the redesign must overcome, because the redesigned function’s strength comes from its depth and focus, and growing it toward the comprehensiveness of the librarian model would dilute the depth that the redesign exists to produce.
There is a constant pressure to absorb the redesigned function back into the librarian model, and naming the pressure is necessary because an organisation that has redesigned its function will lose the redesign if it does not actively defend it. The pressure comes from the same source that produced the librarian function in the first place: the demand that the architecture function demonstrate its value through visible work, and the redesigned function’s work is largely invisible, because the maintenance of conditions leaves no artefact and the prevention of problems produces no evidence. Under the pressure to demonstrate value, the redesigned function will be tempted to produce artefacts again — to document its condition-maintenance, to generate reports of its activity, to make its invisible work visible through the production of the very artefacts the redesign was meant to abandon. This temptation is the path back to the librarian function, and it must be resisted continuously, because each artefact produced to demonstrate value is effort taken from the condition-maintenance that is the function’s actual purpose, and an organisation that yields to the temptation will find its redesigned function gradually reabsorbed into the librarian model it was redesigned to escape, producing artefacts to demonstrate a value that the artefacts themselves undermine.
The redesigned function is, above all, the steward of the instruments, and this stewardship is the concrete form its purpose takes. The instruments — the decision aging record, the ownership velocity index, the epistemic integrity score, the translation completeness measure, the Clarity Stack in its operating mode, the Pulse System — are the means through which the conditions are maintained, and someone must keep them, read them, and act on what they reveal. This is the redesigned function’s daily work: not the production of architecture documents but the stewardship of the instruments that read whether the organisation’s architecture is producing the velocity it should. The function reads the ownership velocity index and identifies the positions whose velocity is low; it reads the epistemic integrity score and identifies the dimensions under pressure; it reads the translation completeness and identifies the decisions that have not reached delivery; and it acts, on each reading, to maintain the condition the instrument reveals to be drifting. The redesigned function is the organisation’s instrument-keeper, and the keeping of the instruments is the work through which the conditions are maintained, the velocity is sustained, and the architecture function fulfils the purpose that the librarian function, for all its diligence, could never serve.
The redesigned function, correctly understood, is the architecture function the profession described all along and rarely built — the function responsible for clarity, present at the point of choice, measured by the organisation’s decisiveness rather than the condition of its records. Building it is the work of the transition, and operating it is the work of the decisive organisation. But the function that works in an organisation of one scale does not automatically work in an organisation of another, because the conditions the function maintains depend on properties — the pulse intervals, the authority altitudes, the structure of the decision infrastructure — that change as the organisation grows. The function designed for a system of moderate complexity must be redesigned again when the system becomes large, and the logic of that scaling, the properties that remain constant and the properties that must change, is the subject the next chapter takes up.