UNDER CONSTRUCTION — WORK IN PROGRESS
This page forms part of a developing public proof of concept. Content, structure and illustrations may change as the project develops.

WORK IN PROGRESS — WHY PUBLIC SYSTEMS FAIL

Why Public Systems Fail

The citizen experiences the whole system. The institutions experience only their individual parts of it.

Public-sector failure is often not caused by one incompetent person or one broken organisation. It can emerge when several organisations each perform their own bounded task while nobody owns the complete outcome.

The public sector is already partly digital, partly analogue and entirely human. A missing letter, an outdated database entry, a misunderstood telephone conversation and a perfectly functioning computer system can combine to produce one failed outcome.

Primary visual slot: When Nobody Sees the Whole Picture.png


1. Fragmented ownership

Specialisation is essential. Police investigate. Prosecutors make charging decisions. Courts determine guilt or innocence. Complaints bodies review conduct. Oversight bodies monitor performance. Policy teams design reform.

The difficulty begins when each organisation sees only its own boundary and no one is responsible for looking across the interfaces.

A system cannot reliably learn from failures it has never organised itself to see.

2. Local completion, system failure

An organisation can complete its process correctly while the citizen is still left with an unresolved consequence.

This is one of the central distinctions in Rewiring the Public Sector: process completion is not the same thing as outcome resolution.

The paperwork may be complete. The failure may still be present.

Supporting visual slot: The Orphaned Outcome.png

3. Disconnected records

Information can be correct in one place and wrong in another. It can be duplicated, stale, incomplete, contradictory or detached from the context that originally gave it meaning.

Once records are separated across organisations, each copy can begin to acquire a life of its own. A later authoritative outcome does not automatically propagate back through every database, form, note and decision that depended on the earlier information.

That is why record integrity is not simply about whether a document exists. It is about whether the right information reaches the right place, at the right time, with its provenance and meaning intact.

Infographic slot: Disconnected Records in the Archive.png

4. Data without reconciliation

More data does not automatically produce a better system.

If records from different sources are never reconciled, the system can accumulate contradiction rather than knowledge. A database may be technically functioning while the organisation acts on an incomplete picture.

This matters particularly when one organisation’s output becomes another organisation’s input. Errors can cross institutional boundaries and gain authority simply because they have been repeated.

5. Analogue bureaucracy inside a digital world

Digitisation does not necessarily remove bureaucracy. It can reproduce old processes on faster technology.

A public service may contain sophisticated databases while still depending on letters, scanned documents, manual re-keying, telephone calls, separate spreadsheets and individual judgement to connect the system together.

The problem is therefore not simply that some processes are analogue. It is that analogue and digital elements often coexist without a deliberate architecture for the interfaces between them.

Don’t digitise the failure. Understand the system first.

6. Weak feedback

Open-loop systems act but do not reliably check what happened next.

A complaint may be answered. A recommendation may be accepted. A procedure may be changed. A letter may be sent. None of those actions, by themselves, demonstrate that the original failure has been removed.

Without independent verification and effectiveness review, an organisation can record corrective action as complete while the defect remains embedded in the system.

7. The handover problem

Many serious failures occur not within one organisation but at the boundary between organisations.

At a handover, responsibility can become ambiguous, context can be lost and assumptions can be transferred without challenge. The receiving organisation may reasonably trust information because it came from an apparently authoritative source, while the originating organisation assumes its responsibility ended once the information was passed on.

The interface therefore deserves as much design attention as the internal process.

8. Cascade amplification

Small errors do not always remain small.

An initial classification, omission or assumption can pass through several records and decisions. Each subsequent system may add its own authority or consequence. What began as a minor defect can become a substantial outcome because the error has been amplified across the network.

This is why early verification matters: the cheapest point at which to correct an error is before it becomes somebody else’s input.

9. Measures without explanation

Performance data can show that something changed without explaining why it changed.

Targets, percentages and trends are useful only when the underlying measures remain comparable and the causes of variation are understood. Otherwise an organisation can mistake movement in a metric for improvement in the system.

The engineering response is variance explanation: ask what changed, why it changed, what else could explain it, and whether the measurement system itself has altered.

10. Systems designed with no headroom

A service operating permanently at maximum nominal capacity has little room for variation, unexpected demand, staff absence, recovery or error.

In engineering, spare capacity is often intentional. In public administration it can be misread as inefficiency.

Unused capacity is not necessarily waste. Sometimes it is resilience.


The recurring failure pattern

The same pattern appears repeatedly:

  1. A process is divided across several organisations.
  2. Each organisation optimises or completes its own part.
  3. Information crosses an interface with some context lost.
  4. No one owns the complete downstream consequence.
  5. The citizen experiences the cumulative result.
  6. The organisations see only their individual records.
  7. A complaint or review examines the parts rather than the whole.
  8. Action is recorded, but effectiveness is not independently verified.
  9. The system remains capable of producing the same failure again.

Rewiring the Public Sector begins by making that whole system visible.