The Illusion Was Rational

Chapter 3: The Board Narrative and the System

The board pack presents parallel descriptions of different aspects of the organisation using different measures, none of which connect investment to system behaviour. Each section is internally consistent and globally incoherent; the gap is bridged socially by translation roles whose existence is the cost of structural incoherence.

The CEO of a mid-sized European insurer funded a three-year technology modernisation programme at € 40 million, hired a respected CTO, and approved a new organisational structure. Two years in, with the core measures she cares about unmoved, she commissions an external review, and the consultants produce 120 slides recommending another reorganisation, a new operating model, a governance framework. She approves the recommendations, because the structure provides nothing else to approve. The problem is not CEO incompetence; she is governing through representations that cannot be reconciled. This was the rational outcome of a governance model designed before software mediated value creation, and the representations she receives were built to be approved, not tested against the system.

Every board of a software-dependent corporate receives a narrative, as a board pack, management reports, a quarterly business review, containing financial performance, programme status, risk registers, and strategic progress. Each section is produced by a different function and is internally consistent. The board reads finance, technology, operations, and risk as a composite picture of one organisation, but they are parallel descriptions of different organisations, each in its own language, with its own measures and its own definition of success. The CFO's section reports cost, revenue, and margin against budget. It does so accurately, but it does not connect expenditure to system behaviour: a € 40 million programme appears as a cost line, with nothing to show whether it changed anything in the system that creates value. The CTO's section reports deployment frequency, a new application, a stronger recruitment brand, without connecting any of it to whether the organisation can bring a product to market faster. The COO's section reports processing times and incident counts without explaining whether performance is constrained by the system or by the people running it. Each section is locally consistent; taken together they are globally incoherent, because the organisation built no mechanism to reconcile these representations against the system that implements them.

Software-dependent corporates govern through representations, strategy documents, operating models, architecture diagrams, portfolio views, KPI dashboards, and these serve a legitimate purpose: no CEO can understand every system in a large organisation, and representations provide the necessary abstraction. The problem begins when a representation replaces the thing it represents. A portfolio view that shows “Customer Engagement” as a product with a P&L and a roadmap is a representation of a business capability; if the portfolio treats it as something that can be funded, measured, and held accountable, it has confused the label with the system. The product called Customer Engagement has no coherent boundary in the code: it is a name applied to features scattered across services maintained by different teams. The label exists; the product does not.

How does this incoherence survive? The gap between the board narrative and the system is bridged socially, by people whose job is to translate between the two. These translation roles, programme managers, delivery leads, portfolio analysts, business partners, are so pervasive that they are invisible. They are not performing unnecessary work: without them, the gap between narrative and reality would produce immediate, visible failure. But the synthesis they provide is expensive, fragile, and entirely dependent on the continued presence of the people performing it.

Boards were designed to govern through representations, and that is appropriate: a board cannot and should not understand every service and data flow in a large organisation. The governing question is what happens when the representations cannot be tested against the system they describe. When they are accurate, governance works. When they are disconnected from reality, the board approves a € 40 million programme not knowing the system cannot deliver the business case without structural changes the programme does not include. The board is not incompetent; it is governing through artefacts that reach it only after the structural contradictions have been smoothed away. The constraint is structural: nothing in the current model requires any representation to be testable against the system, so the board keeps approving descriptions that no one is obliged to reconcile.

...

See also: Full contents · Preview chapters · Illusions of Work