The Clarity Stack surfaces questions. The Pulse System synchronises the rhythm at which those questions are brought to the altitude that can resolve them. Neither instrument resolves anything. The resolution happens inside the governance event — the forum, the review, the moment at which a question that has been surfaced and routed is converted into a binding outcome that closes it. That conversion is the work the entire governance architecture exists to make possible. And it is the work most governance architectures never describe, because they assume that a question brought to the right people will be resolved by the fact of their attention. It will not. Attention is not resolution. A question can be examined by every authority in the organisation and emerge from the examination exactly as open as it entered.
The Decision Architecture Method is the structural description of what must happen inside a governance event for it to terminate ambiguity rather than circulate it. It is not a facilitation guide. It does not describe how to run a meeting, how to manage personalities, how to build consensus, or how to ensure that every voice is heard. Those are techniques for conducting a discussion. The method described here is concerned with a different and more specific thing: the structural conditions under which a discussion produces a decision, and the structural failures that cause a discussion to produce a record instead. The distinction between the two is the distinction between a governance architecture that moves and one that performs the motions of moving while the questions remain where they were.
The method has four stages. Each stage has an entry condition that must be satisfied before the next stage can begin. Each stage has a characteristic failure that occurs when the stage is skipped or performed incompletely. And each failure produces a recognisable downstream condition — a specific way in which the question survives the governance event that was supposed to end it. The stages are not a procedure to be followed in sequence by people in a room. They are a description of what must be structurally true for the governance event to terminate the question, and the method’s value lies in naming precisely where governance events fail to be true.
The first stage is the entry condition. Before a question enters the governance event, something must already be true about it: it must be a question the forum has the authority to answer, framed in a form that admits an answer, brought with the context required to answer it. This sounds elementary. It is the stage skipped most often. Questions arrive at governance forums in three defective forms, and a forum that accepts them in these forms has already failed before the discussion begins. The first defective form is the question the forum cannot answer — the question whose resolution requires authority that sits at a different altitude, brought to this forum because this forum is the one that meets, not the one that decides. The second defective form is the question that is not yet a question — the open-ended concern, the area of discomfort, the sense that something needs attention, brought to the forum in the hope that the forum will convert the discomfort into a decidable proposition. The forum cannot. Converting a concern into a question is work that must be done before the governance event, not during it. The third defective form is the question brought without the context required to answer it — the decision that depends on information the forum does not have and cannot generate in the time available.
The failure at the entry stage is the failure of admission. The forum admits a question it cannot resolve, and the discussion that follows is not a decision process but a discovery process — a gradual collective realisation that the question cannot be answered here, with the information available, by the authority present. The discussion consumes the governance event. It produces a record — a note that the question was raised, a recommendation that it be examined further, an action to gather the missing information or escalate to the altitude that can decide. The question leaves the forum exactly as open as it entered, and the governance architecture has spent a governance event confirming that the question was not ready to be decided. The entry condition exists to perform that confirmation before the event rather than during it. A question that fails the entry condition should not enter the forum. It should be returned to the point where the missing element — the authority, the framing, the context — can be supplied.
The second stage is the signal requirement. Before opinion is permitted to enter the governance event, the forum must establish what is true about the system’s current state in the domain the question touches. This is the stage that distinguishes a decision from a debate. A debate begins with positions. A decision begins with the picture — the current state of the system as the operational truth instruments report it, the constraints already in force, the decisions already made that bound the question, the deviations already surfaced that bear on it. The signal requirement is the structural insistence that this picture is established and agreed before anyone advocates for an outcome. The picture is not a matter of opinion. It is a reading of the instruments. When the forum establishes the picture first, the subsequent advocacy operates within the bounds of what is actually true. When the forum permits advocacy first, the picture becomes contested — each advocate describing the system in the terms that support their preferred outcome — and the forum spends its energy adjudicating between competing descriptions of reality rather than deciding what to do about the reality.
The failure at the signal stage is the failure of grounding. The forum proceeds to decide without establishing the picture, and the decision it produces is grounded not in the state of the system but in the relative persuasiveness of the people in the room. The most articulate advocate wins. The most senior voice prevails. The decision is made — it is real, it is binding, it is recorded — and it is made against a picture that was never established, which means it may be the right decision for a system that does not exist and the wrong decision for the system that does. The signal requirement is the defence against this failure. It does not guarantee a good decision. It guarantees that whatever decision is made is made against the truth of the system rather than against the rhetoric of the room.
The third stage is the compression mechanism. This is the stage at which the forum moves the question from a set of options toward a single binding commitment — and it is the stage at which governance events most reliably fail, because compression is uncomfortable and the forum has many ways to avoid it. Compression means narrowing. It means that of the available options, one is chosen and the others are closed. It means that the trade-off the question contains is accepted in a specific form, with specific costs consciously taken on, owned by a named holder who will be accountable for the consequence. Compression is the opposite of the forum’s natural tendency, which is to expand — to add options, to identify additional considerations, to request further analysis, to widen the question rather than close it. Every expansion feels like diligence. Every expansion is a deferral. The forum that responds to a hard trade-off by commissioning more analysis has not been more rigorous. It has declined to compress, and dressed the declination in the language of thoroughness.
The compression mechanism is the structural counter to this tendency. It establishes that the purpose of the governance event is to compress — that the event is not complete when the options have been examined but when one has been chosen, and that the examination of options is instrumental to the choice rather than an end in itself. The mechanism operates through a specific discipline: the forum names, before it begins to compress, what would have to be true for each option to be selected, and then reads the established picture to determine which of those conditions actually holds. Compression is not the imposition of a preference. It is the application of the established picture to the named options until the picture eliminates all but one. When the picture does not eliminate all but one — when two options remain viable against the current state of the system — the compression is performed by the holder, who accepts the trade-off in the form their authority permits and owns the consequence. What the mechanism does not permit is the third path: the path where the picture is not applied, the holder does not choose, and the question is widened until it no longer fits inside the governance event and must be carried to the next one.
The failure at the compression stage is the failure of closure. The forum examines, considers, deliberates — and adjourns without compressing. The question is not closed. It is matured. It returns to the next governance event better understood, more thoroughly analysed, surrounded by more context, and exactly as undecided as before. This is the failure that produces the appearance of progress without its substance. Every governance event advances the question’s documentation and none advances its resolution. The decision aging record reads the question as moving — it is being actively discussed, it appears in successive forum records, it is visibly receiving attention — while the question itself is stationary. The compression mechanism is the defence against this failure. It insists that the governance event produce a narrowing, and it treats the absence of a narrowing not as prudent deliberation but as a failure of the event to do the one thing it exists to do.
The fourth stage is the binding output. A decision that is compressed inside the forum but not bound outside it is not yet real. The binding output is the structural requirement that the compressed decision be recorded in the decision layer in a form that makes it operational — held against an owner, translated into the constraint it implies, given an expiry threshold at which it will be revisited, and made accessible to every practitioner whose work it governs. The compression produces the decision. The binding makes it consequential. Without the binding, the forum can compress perfectly — choose decisively, accept the trade-off, own the consequence — and the decision will dissolve in the interval before the next governance event, because nothing captured it in a form that delivery could act on. The practitioners who needed it never received it. The constraint it implied was never established. The decision was made and then lost, and the question it answered will return to a future forum as though it had never been decided.
The failure at the binding stage is the failure of transmission. The decision is real in the room and absent everywhere else. This is the failure that most resembles success, because everyone present at the governance event experienced the decision being made. They remember the discussion, the compression, the moment the holder accepted the trade-off. From inside the room, the question was resolved. From outside the room — where the work is done, where the constraint should operate, where the practitioner consults the decision layer and finds nothing — the question is open. The binding output is the bridge between these two realities. It is the structural insistence that a decision is not complete when it has been made but when it has been transmitted into the infrastructure that holds it and the environment that must act on it.
The four stages compose into a single property: a governance event that satisfies all four does not circulate the questions it receives. It terminates them. The entry condition ensures the question is one the forum can resolve. The signal requirement ensures the resolution is grounded in the truth of the system. The compression mechanism ensures the deliberation produces a narrowing rather than a maturation. The binding output ensures the narrowing becomes operational rather than remaining a memory of the room. A failure at any stage produces a governance event that consumes attention and returns the question open. The method’s value is that it names which stage failed — and therefore what to repair — rather than leaving the organisation with the undifferentiated sense that its governance is not working without the ability to say where.
The four stages must occur in order, and the order is not a matter of procedural tidiness but of structural necessity, because each stage produces the condition the next stage requires. The entry condition produces a question the forum can actually resolve, which is the precondition for the signal requirement, because there is no point establishing the picture for a question the forum has no authority to close. The signal requirement produces the established picture, which is the precondition for the compression mechanism, because compression is the application of the picture to the options and cannot proceed before the picture exists. The compression mechanism produces the narrowed commitment, which is the precondition for the binding output, because there is nothing to bind until the compression has produced the decision the binding makes operational. Run out of order, the stages do not merely lose efficiency; they lose the conditions that make them possible. A forum that attempts to compress before establishing the picture compresses against opinion. A forum that attempts to bind before compressing binds an absence. The order is the logic of the method, and a governance event that runs the stages in any other sequence is not running the method faster or differently. It is running something else that resembles the method and produces the failures the method exists to prevent.
There is a temptation, having understood the four stages, to convert them into a meeting agenda — to print them as headings, to walk the forum through them in sequence, to treat the method as a script that a facilitator follows. This temptation should be resisted, not because structure in a governance event is harmful but because the method is not a description of what the forum should talk about. It is a description of what must be structurally true for the forum to terminate the question, and the difference matters because a forum can follow the agenda perfectly and satisfy none of the conditions. A forum can have an agenda item labelled compression and never compress, because the holder present does not have the authority to accept the trade-off and the item passes with the question discussed and unclosed. The method is not satisfied by an agenda that names its stages. It is satisfied by the structural conditions its stages describe, and those conditions are present or absent regardless of whether the meeting was organised around them. The method’s value is diagnostic rather than procedural: it tells the organisation, when a governance event fails to terminate a question, which condition was absent — and the absent condition is rarely the one the agenda was missing and almost always the one the structure could not supply.
The method also has a relationship to time that the four stages do not make explicit and that the organisation operating the method must understand. The entry condition, the signal requirement, and the compression mechanism can in principle occur within a single governance event, because they are stages of reasoning that a forum can perform in sequence in the time it has. The binding output cannot reliably occur within the same event, because binding requires the decision to be recorded in the decision layer, translated into the constraint it implies, and made accessible to the practitioners it governs — and this is work that extends beyond the forum, into the infrastructure and the delivery environment. The failure of transmission that the binding stage guards against is, in practice, most often a failure of the interval after the forum, when the energy that produced the compression has dissipated and the work of binding the compressed decision feels like administrative follow-up rather than the completion of the decision. The method’s insistence that the binding output is a stage of the method rather than a clerical task that follows it is the structural counter to this dissipation. The decision is not made when the forum compresses it. It is made when the binding completes, and the interval between the two is where decisions most reliably die.
There is a precondition the method assumes and does not itself supply. Every stage of the method depends on a holder — someone whose authority makes the entry condition assessable, whose accountability makes the compression possible, whose ownership makes the binding real. The method describes what the holder must do. It does not establish whether a genuine holder is present. A governance event can be structured perfectly and still fail at every stage if the question has no owner — if it has entered the forum because it belongs to no one and the forum is where ownerless questions go to be discussed. The relationship between ownership and the rate at which questions move through the method is the most predictive property in the entire governance architecture, and it is the subject the discussion that follows takes up directly.