Captured Governance

Suppressing internal feedback is one thing. Blocking external information is another. L11 is something else entirely: actively degrading the oversight function itself.

What is captured governance at DHCW?

Oversight obstruction differs from information suppression in kind, not degree. Suppression hides the searchlight’s view. Obstruction disables the searchlight. Once an oversight role is hollowed out, the capacity does not regenerate: the institutional knowledge walks out with the person, and the replacement — if there is one — inherits a capture-ready role rather than a functioning one.

The delay is permanent. What is removed does not come back on its own.

How It Manifests at DHCW

The captured assurance machinery. Before the Welsh Government level, the oversight functions internal to DHCW had already failed. The board approved without scrutiny: the knowledge graph documents 51 distinct instances — including a £20M Kainos framework where the Chair later admitted “I should have looked. I don’t know how these appear on our website as contracts,” a £226M Microsoft Enterprise Agreement passed in a single sentence with no questions and no vote, and full business cases for LIMS, LINC, and RISP all moved into private session under “commercial sensitivity.” The Performance and Delivery Committee generated zero corrective actions across eighteen consecutive months from May 2024 to May 2025, even after Welsh Government installed Olivia Shoraks as a permanent observer. Inside that same window, the data centre cooling failover failed twice — in July 2024 and in June 2025, near-identical incidents twelve months apart, each taking 32 services offline. No corrective action specific to the data centre supplier’s maintenance regime is visible in any PDC output across the window between the two incidents; the contractual scrutiny flagged at the June 2025 board — that the provider may not have met contractual standards — does not appear in any published assurance output. All three sub-committees were blind to imminent Level 3 escalation in March 2025 — none raised concern signals in their final pre-escalation meetings. Audit Wales, present from month six, issued a “good governance, stable and cohesive board” assessment four months before Level 3 was triggered.

Forty-six months after DHCW’s founding, Ifan Evans told the Performance and Delivery Committee: “This persistent ambiguity, let’s call it, which we usually phrase as ‘we need to clarify roles and responsibilities’… PDC is an oversight and assurance function, not a programme management function.” DHCW was simultaneously being measured against milestones for programmes it did not control — most visibly EPMA, where local organisations are the accountable delivery body — while having no mechanism to hold those organisations to account. The gap was named, repeated, and never closed.

Welsh Government as co-conspirator, then sudden judge. Welsh Government created the conditions for failure, then escalated DHCW for the predictable consequences. WG pressured DHCW to soften red RAG ratings on programmes (Evans, PDC November 2024: “Government was asking… why then are those programs rated red amber?”). It cut DPIF from £33M planned to £28M allocated for 2024-25. It refused capital funding for e-referrals and the integration hub in November 2025, then set milestones requiring them. It compressed LIMS from a four-year programme into two years. It imposed a recruitment freeze via the remit letter while demanding accelerated delivery. Funding letters were not confirmed until 25% through the financial year (July 2022). The entity issuing the corrective signal was substantively co-authoring the failure.

The neutralised oversight layer. Leadership obscured the information provided to Welsh Government’s oversight bodies. This contributed to the departure of the WG GDS director from his role. The GDS director was the Welsh Government’s principal technical counterweight to DHCW — the one person in the oversight structure with the capacity to independently assess whether DHCW’s claims about programme status were true. His departure removed that capacity. It has not been replaced. With the technical oversight function neutralised, DHCW’s information monopoly over its own oversight body became near-total. The independent digital expert appointed under the Level 3 framework has never been publicly identified — making it impossible to assess their independence, qualifications, or conclusions.

From Level 3 to Level 4. Twelve months of Level 3 monitoring produced no de-escalation and no consequences. The monitoring mechanism itself had been captured: it functioned as proof that oversight existed without functioning as actual oversight. In 2026, DHCW was escalated further — to Level 4 Targeted Intervention. The Cabinet Secretary then publicly declared the entire framework “complex, data-heavy, burdensome, lacks transparency and does not drive improvement.” The minister responsible for overseeing DHCW had publicly acknowledged that the oversight mechanism does not work — while continuing to operate it.

March 2026 end state. At the most recent board meeting on record: Phase One of Level 4 found two of forty-seven milestones missed (specifically LIMS and WRISTS); £32.9M of DPIF revenue and £13.1M of capital remained unallocated; an Accountable Officer letter had been sent stating delivery was “not possible without confirmed DPIF allocation”; recruitment was frozen by remit letter; the IQPD meetings were ceasing, replaced by a new escalation meeting chaired by the Director General from April; Rowan Gardner — the sharpest governance challenger on the board — was departing; the CFO had already departed to Swansea Bay. Glazzard’s closing assessment: “We’ve always struggled because it’s one-year funding. We should have worked it out by now, surely.”

L11 completes the self-preservation engine. L6 manages internal perception. L9 silences dissent. L10 blocks external information requests. But none of those address the most dangerous threat: active government oversight with real powers. L11 closes that gap.

This is the difference between hiding from a searchlight and disabling it.

The obstruction occurred while DHCW was under active investigation — meaning leadership was simultaneously being scrutinised and actively undermining the quality of that scrutiny. Every escalation step was taken with full knowledge of the consequences: the CEO and directors were informed of the harm and chose to proceed.

The legal significance is substantial. Misconduct in public office under the Attorney General’s Reference (No 3 of 2003) requires “wilful misconduct serious enough to amount to an abuse of the public’s trust.”

Actively undermining the government’s ability to monitor a public body — during an investigation triggered by serious delivery and patient safety concerns — meets that threshold.

The Escalation trap is visible here. External scrutiny escalates. Leadership escalates information control in response. This generates more evidence of obstruction. Scrutiny escalates further. They are escalating their way into formal accountability proceedings — criminal, regulatory, and civil — that will bypass the information fortress entirely, because Employment Tribunal judgments are public, criminal investigations have compulsory powers, and Audit Wales has statutory access.

What would a healthy alternative look like?

Oversight bodies have independent, unfiltered access to delivery data — not mediated through the people being overseen. Technical oversight roles are structurally protected: departure of the person fulfilling that role triggers automatic replacement, not a gap. Monitoring frameworks have teeth — escalation produces consequences on a defined timeline, not open-ended observation. When governance is not captured, the oversight function cannot be degraded from inside.

How does the blueprint break captured governance?

Obstruction of oversight cannot be remedied from within the obstructing organisation. Breaking L11 requires two moves together: Competent Leadership replaces the leadership actively degrading the oversight function, and the governance redesign distributes oversight across multiple actors with overlapping jurisdiction — so that capture of any single channel does not neutralise accountability entirely.