The framework
Seven meta‑dimensions
A structured way to read a complex system — an organization, a rival, a conflict — across seven angles instead of one, so that you understand how it actually works before anyone decides what to do about it.
You have probably been in this moment.
Something goes wrong — in the organization, the market, the supply chain, the coalition — and the people responsible do what reasonable people do. They find the most obvious cause and address it. They bring in consultants. They restructure. They launch an initiative.
Six months later the problem is back, or it has moved somewhere else, or the fix created three new problems nobody anticipated.
That is not a failure of effort. It is a failure of diagnosis.
Most of the tools we use to understand complex situations were designed for simple ones. They assume that if you pull this lever, that thing moves. They assume the past predicts the future in a straight line. They assume that what is visible is what is real, and that the people deciding are seeing clearly. None of those assumptions holds in a living system — where everything is connected to everything else, where history shapes the present in ways nobody is tracking, and where the same intervention succeeds brilliantly in one place and fails catastrophically in another.
Think of a good physician.
They don’t ask one question and write a prescription. They take your temperature, your pressure, your pulse, your history. They listen for what you are not saying. They order tests. They are building a picture — not of your symptoms, but of your system. Only when that picture is clear do they propose a course of action. Doctors who treat only symptoms are not good diagnosticians. The good doctor gets at the invisible mechanisms driving the visible symptoms. Soundings isn’t like a good doctor. It’s like a good team of doctors.
Seven angles is what it takes to build that picture of an organization. Not how the org chart says it should operate, or how last year’s strategy deck describes it. How it behaves in practice.
Why they are not numbered one through seven
A list implies a sequence, and a sequence implies the first item matters most. This one is not shaped that way, and pretending otherwise would put an error into the diagnosis before it started.
Five of the angles are genuine peers, read in an order the engagement sets — and under denied access, in a fixed order the framework specifies rather than the analyst chooses. The sixth produces no finding of its own; it conditions the other five. The seventh reads nothing directly; it pulls the other six into a single account. Three different kinds of work, and calling them one through seven would hide the distinction that matters most.
Five peers — read in an order the engagement sets
Each is read on its own instruments and produces its own finding, with its own confidence grade and its own falsifier.
Patterns & Dynamics P&D
What does this system do again and again, whatever it says it wants?
The rhythms and repetitions you can actually observe — and what those repetitions reveal about what the system is really optimizing for, which is frequently not what anyone has written down. Behavior that recurs regardless of intention cannot be faked, and that is exactly what makes it worth reading.
Change & Evolution C&E
Is this change drift, adaptation, or decay?
Every organization is changing. Very few can tell you which of those three it is doing. This reads how the system is changing and what that change is actually made of — and whether the change process it currently runs is capable of producing the change being asked of it.
History & Memory H&M
Which of today’s constraints are old decisions nobody is still making?
What the system has been through, and how that history still shapes the present — often invisibly, the way an old injury changes how a person walks. The approval step whose original reason has been forgotten. The person still guarded about something that happened years ago.
Connections & Flows C&F
What moves through this system, along which paths, and where does it stop?
Where information and resources actually flow, because those paths are rarely the ones anyone drew on a whiteboard. What is fast, what bottlenecks, and where a single connection is carrying load the structure assumes is distributed.
Perception & Bias P&B
Where has this system’s picture of itself drifted from what is actually there?
That gap is almost always where the most dangerous blind spots live. And they are architectural rather than personal — produced by where people sit, what reaches them, how long it takes, and what the organization quietly rewards them for not noticing. Under denied access this is read first, and everything downstream inherits it.
Conditions the five
Adds no sixth finding of its own. It re‑examines the five that exist.
Scale & Nonlinearity S&N
Does this reading survive a change of scale?
The same thing that is a minor irritant at one level is a catastrophe at another. Small problems compound; large interventions dissolve into nothing. Knowing where a system is sensitive — and where it simply is not — is one of the most valuable things a leader can hold, and it is the thing a single‑level reading cannot give them.
A great deal of organizational analysis is a scale error wearing a confident face. This stage takes each of the five readings and tests it at another scale, and where it does not hold, the reading is corrected rather than the objection noted.
Pulls the other six together
Reads nothing new. Turns six conditioned readings into one account.
Process Architecture & Mechanism PA&M
How does the machinery actually run?
Who the real role players are, how they relate, what rules they are genuinely operating under, and how the system responds when something happens to it. Five elements: role players, relationships, roles, rules, and response mechanism.
Candidate mechanisms are declared openly and tested against every dimension assembled — one row per candidate per dimension — with each divergence investigated and located rather than smoothed into agreement. Where two candidates tie, none is selected and the tie is reported.
What comes out is not a list of observations. It is an explanation of how the thing runs, and the step that is rate‑limiting it — the place where, if anything is going to move, movement has to start.
Two modes
Whether the system will let you look
The same seven angles operate in two modes, and which one runs is determined by what can actually be reached rather than by preference.
A system that will let you look
You commission the reading and open the record to it. The system is not hiding from the instrument, and the diagnosis goes as deep as the record allows. This is the physician examining a patient who came in for help.
A system that won’t
The board is hostile. The competitor will not answer. The counterparty across the table, the regulator, the community burned too often to talk to outsiders. Not merely complicated — actively shaping what you see, or closed off entirely.
The second is more common than people assume. Anyone facing a hostile acquirer, a difficult regulator, an opposing party in a negotiation, or a competitor whose next move matters is operating under denied access — and the tools that work beautifully when a system consents to examination fall apart when it does not.
That mode assumes you cannot ask the system a question and trust the answer. It assumes what is visible may be precisely what someone intends you to see. It is the physician reading a patient who will not admit anything is wrong — or who has been coached on what to say. The good clinician does not give up. He reads what the patient cannot hide: the gait, the subjects avoided, what stays consistent and what does not.
When the same finding surfaces in the rhythms, the flows, the history and the response to stress, you can trust it more than any single source. When the findings do not line up, that mismatch is itself information.
It also assumes something the cooperative mode does not: that some of what you are seeing was put there for you. It distinguishes a system deceiving itself from a system deceiving you, because those look similar from outside and need entirely different handling — and it reasons from what a deception would have cost to produce, because the signals an adversary cannot afford to fake are worth the most.
Everything upstream of the gate describes.
After the seven angles have done their work, two further instruments run. The first reads the diagnosis forward: given where the system now stands, which other configurations can it actually reach, and by what sequence of footholds. The second is the gate — the disciplined point at which findings are permitted to become recommendations, and the only point at which that turn is allowed.
That boundary is the architecture’s most important feature. Nothing upstream of the gate prescribes. The instrument will tell you the option space and characterize what each option would require, but it does not choose your objective for you, and it does not convert a finding into a recommendation until it has asked whether the system in front of you can absorb one.
You do not need to know any of the underlying science to use it. You need to care about getting it right.
The framework does not stop at the organization.
It reads complex adaptive systems, and an organization is one instance. The same seven angles have been run against a failed engineering program, a geopolitical system, an electricity grid, and a published argument treated as a system of thought. What changes is the record available; what does not change is the discipline applied to it.