Utilities Privatisation
This review has its own page due to the significance. The document was produced in under one hour, initially using a leading AI Assistant to review the Privatisation of Utilities, focusing on Water. I then used The Weavers to explore further insights. It was prompted by the latest news on Thames Water. The conversation covered Thames and United Utilities as their models were different. I wanted to compare the response provided using an AI Assistant against that provided using The Weavers System.
UTILITIES PRIVATISATION: WHAT THE FRAME COULD NOT SEE
A Weavers Assessment of the UK Utilities Privatisation Experiment
A Report — with a Controlled Comparison Against an Unassisted AI Analysis
David Sutton CITP MBCS July 2026
| Provenance and method This report was produced with Claude Fable 5 working with the Weavers Main document v51, including Appendix 1 (the thirty-plus insights of the UK Industry 4 Transformation Strategy), attached as working context, combined with the model’s own knowledge and verified current reporting. The comparator input was a conversation with a leading AI assistant that had no access to the Weavers: a capable, well-sourced, balanced analysis of utilities privatisation, conducted below the line. The report therefore functions as a controlled comparison: the same subject, examined with and without the instrument. The practitioner held the recognition function throughout, and contributed one piece of first-hand evidence — the NORWEB and Vertex history — whose significance the unassisted analysis recorded but did not see. Figures drawn from the comparator conversation are reported as its sources gave them and should be read as contested estimates rather than settled fact; figures describing the July 2026 situation are from current public reporting, listed at the end. |
| Executive Summary The unassisted analysis is good below-the-line work: it marshals the evidence of financial engineering, names the debt mechanics and the lenders, balances failure against genuine gains, and derives four sensible lessons. Every lesson, however, is frame-internal: better ownership models, better regulatory powers, better dividend rules. The frame it cannot leave is the one the entire national debate is trapped in — that this is an argument about ownership and regulation. The Weavers assessment finds the deeper structure. Privatisation transferred the assets but, more consequentially, it transferred the understanding: the state entered a dependency cascade — outsourcing first the doing, then the knowledge, then the direction of an essential service — and its present terror of special administration is the moment of discovering what Phase 3 means. The same cascade ran in miniature inside the companies, documented here through firsthand evidence from the Vertex and NORWEB restructurings. The regulatory architecture was a vine: three regulators and two departments, each individually reasonable, none owning the whole, formally acknowledged only in 2025 — thirty-six years in. Compliance was certified at specification level while the system failed at failure level, to the point where a licence condition has been visibly breached since 2024 because enforcing it would trigger the failure nobody owns. And the golden thread from privatisation’s stated purpose to its actual mechanics was severed at birth — a severance the currently proposed rescue, which seeks survival by suspending performance obligations, would re-run rather than repair. The report closes with six additional lessons, four candidate insights for the register, and five inversions. |
1. The Comparator: Capable Analysis Inside the Frame
The unassisted AI analysis deserves an honest account, because its quality is what makes the comparison meaningful. Asked what the evidence shows and what lessons can be learned, it assembled the substantive case: water companies privatised with no debt and carrying over £70 billion of it three and a half decades later; dividends exceeding £80 billion against shareholder equity that shrank in real terms; roughly a third of customer bills absorbed by financing costs; leveraged buyouts loading acquisition debt onto the utility’s own balance sheet; dividend recapitalisation; offshore multi-tier structures obscuring true leverage from the regulator; inflation-linked debt cheap in the 2010s and ruinous in the 2020s. Asked who lent, it correctly identified bond markets and pension funds, banks, sovereign wealth funds, and — at the distressed end — the hedge funds now negotiating to own the sector’s largest company. It kept the counter-evidence honestly in view: real capital investment, world-class drinking water quality, the telecoms success. And it derived four lessons: natural monopolies cannot mimic markets; risk must not be privatised in profit and nationalised in failure; financial engineering must be checked; fragmented planning needs a public system operator.
All of this is true, useful — and entirely inside the frame. Every remedy is an ownership or regulatory remedy; every lesson assumes the question is how to configure ownership and oversight. The analysis also received, from the practitioner, a piece of firsthand evidence it processed only as corporate history: that when the Vertex subsidiary — the group’s call centres and IT services — was sold, the water company had to spend years and hundreds of millions rebuilding those functions from scratch. The unassisted analysis summarised this accurately, drew the mild conclusion that multi-utility conglomerates create distraction, and moved on. It did not see that it had been handed evidence of a general law — the same law that explains the state’s own predicament. That is not a defect of the product. It is the signature of below-the-line work: capable within the frame, unable to see what the frame excludes. This is common to AI assistants not directed to do otherwise.
2. Where the System Stands — July 2026
The sector’s largest company carries close to £20 billion of debt, is funded only until the final quarter of 2026, and has not held the investment-grade credit rating its licence requires since July 2024. A creditor consortium including distressed-debt specialists has proposed a recapitalisation — new equity of around £3.35 billion and several billion more in fresh debt — on terms reported to include waived fines until 2030 and modified pollution and performance targets. The environment secretary has objected to those terms; over a hundred MPs have signed a letter demanding special administration instead; and the decision now rests with an incoming government. No administration order has been made at the time of writing. Meanwhile the Independent Water Commission concluded in 2025 that the fragmented regulatory model itself had failed, recommending a single integrated regulator — the system formally acknowledging, thirty-six years after privatisation, what the Weavers would have named on day one.
3. The Weavers Assessment
3.1 The vine: a regulatory architecture of individually reasonable failure
Water was overseen by an economic regulator, an environmental regulator, and a drinking-water quality regulator, under two government departments with different priorities. Each boundary was individually reasonable: economics is not ecology; quality is not price. Together they produced the defining pattern of the vine — invisible systemic fragmentation in which each body optimised its own remit while the whole degraded. The economic regulator held bills down and permitted leverage; the environmental regulator lacked the resources and the financial sightlines to connect pollution to balance-sheet extraction; no body owned the question that mattered: is this system, as a whole, becoming unable to perform its purpose? The 2025 Commission’s recommendation of an integrated regulator is the vine being named officially. The Weavers’ addition is the timescale: a vine is cheap to design out at the start and enormously expensive to cut down after thirty-six years of growth — and the framework’s test for any new regulatory architecture is not its organogram but whether any single function owns the failure state of the whole.
3.2 The state’s dependency cascade — the finding the ownership debate conceals
Insight 10 of Appendix 1: the dependency cascade begins with outsourcing the doing, proceeds to outsourcing the understanding, ends with outsourcing the direction — and becomes invisible as it progresses, because by Phase 3 the organisation cannot see what it has lost. Read privatisation through this insight and the picture reorganises. In 1989 the state outsourced the doing: operations, maintenance, delivery. Over the following decades it lost the understanding: the engineering knowledge migrated into the companies, the financial understanding into structures the regulator could not penetrate — the offshore tiers existed precisely to defeat oversight. And by the 2020s it had lost the direction: a hedge-fund consortium now proposes the terms on which London’s water will be financed, including which environmental obligations will be suspended, while the state’s remaining move — special administration — frightens it precisely because it no longer possesses the sovereign capability to run what it would be taking over. Insight 11 states the law this violates: you cannot regulate what you do not understand; you cannot require modifications you could not make yourself. Insight 24 states the exit condition: what survives transitions is not a particular ownership model but the retained capability to understand what you depend on, challenge it when it fails, and replace it when you must. The state retained none of the three. That — not privatisation as such — is the structural failure. Public ownership with the same hollowed capability would fail differently, not less.
3.3 The nested cascade: Vertex, NORWEB, and the destruction of understanding
The same law operated in miniature inside the companies, and the firsthand evidence is in the comparator conversation. The 1995 multi-utility merger was unwound in waves: the energy retail arm sold in 2000, the telecoms business in 2006, the electricity network itself in 2007. And Vertex — created by pooling the group’s call centres and IT services, grown into an outsourcing business serving external clients — was sold in 2007. With it went the billing systems, the customer databases, the trained agents: the operational understanding of the company’s own customers and its own technology. The water company then spent years, and substantial regulated capital, rebuilding in-house what it had owned and sold — new call centre operations, a decade-long re-architecture of IT away from the legacy systems that had walked out of the door.
The unassisted analysis recorded this as a lesson about conglomerate focus. The Weavers reads it as the same cascade the state ran, one level down: the doing was pooled, the understanding was sold, and the direction — what the company’s customer operations and technology should become — had to be repurchased at a price no disposal document ever counted. Every restructuring wave in the sector’s history has run this pattern: institutional knowledge treated as free because it appears on no balance sheet, destroyed in each transaction, and rebuilt — where it is rebuilt at all — at the bill-payer’s expense. The sector’s true asset register was never the pipes. It was the accumulated understanding of the people who knew why the pipes were laid where they were — and that register has been in continuous liquidation since 1989.
3.4 Specification-level compliance, failure-level collapse
Insight 34 distinguishes specification-level certification — the system meets its specifications — from failure-level certification: the system behaves safely when it does not perform as specified. The entire regulatory apparatus of privatised utilities was built at specification level. Price reviews certified investment plans; licence conditions certified financial ratios; performance commitments certified target metrics. Nobody certified the failure level, and the failure level is where the sector actually lives: sewage systems certified against design conditions, discharging under the storm conditions that actually occur; debt structures clever at specification level — cheap, inflation-linked — and ruinous at the failure level of an inflation surge; supplier markets in energy certified as competitive at specification level, thirty of whose participants collapsed simultaneously when wholesale prices met the failure level in 2021, at a cost in billions transferred to the public.
The current situation supplies the definitive instance. The largest company’s licence requires two investment-grade credit ratings. It has held none since July 2024 — a specification visibly breached for two years, unenforced, because enforcing it triggers the failure state that no institution owns. A specification that cannot be enforced at the point of failure was never a safeguard; it was a description of the good times. The Weavers’ recovery framework draws the conclusion: any reformed regime must certify the failure level first — who acts, with what retained capability, when a monopoly provider of an essential service stops performing — and treat specification compliance as the easy residue, not the substance.
3.5 The golden thread, severed at birth — and about to be severed again
The golden thread is the traceable connection from any decision back to the purpose it serves. Privatisation’s stated purpose was explicit: private capital would fund the investment the state could not afford. The mechanics that actually operated inverted the purpose: capital flowed out, not in — dividends exceeding £80 billion against equity that shrank in real terms — while investment was funded from customer bills and borrowed money serviced by customer bills. Real investment occurred; the counter-evidence is genuine; but the thread from purpose to mechanism was severed in the design, because nothing in the architecture required the connection to hold, and no institution was charged with tracing it. Where the thread cannot be traced, change is drift — and thirty-six years of drift with a mission statement is what the evidence describes.
The proposed rescue re-runs the severance in miniature. Survival is to be purchased by suspending the purpose: fines waived, pollution and performance targets modified, bills raised beyond the determined level. Whatever its financial necessity, a recovery whose terms suspend the system’s purpose is not recovery in the Weavers’ sense; it is the drift, renamed and refinanced. The golden thread test for any rescue, public or private: can each term be traced to the purpose of the service — or only to the survival of its balance sheet?
3.6 The blue flower and the cooperation that never was
Two final structural absences. First, the blue flower: the least well-resourced participant as the design constraint. Roughly a third of the average bill services financing rather than water; the customer who cannot pay subsidised, through decades, the returns of sovereign funds and the fees of the restructuring to come. No part of the architecture was designed from that customer; social provision was patchwork discretion, not design constraint. Second, cooperation: seventeen water companies operating the same physics in the same country as seventeen silos, with no mandated sharing of failure modes, operational data, or engineering learning. Insight 6 records that the initial resistance is always identical — ‘share with competitors? Impossible’ — and that results exceed expectation once sharing is mandated. In a sector of regional monopolies the competition objection was never even real: there was nothing to compete for, and everything to learn. The learning that each company’s failures could have given the others was simply forgone, decade after decade — the flame never passed.
4. Additional Lessons — Beyond the Comparator’s Four
L1 — Transfer of assets is survivable; transfer of understanding is not. Ownership can be exchanged and exchanged back. The understanding that leaves with outsourcing, restructuring, and disposal does not return with renationalisation; it must be rebuilt at full price. Any future transaction in essential services should be tested for what understanding it transfers, not only what assets.
L2 — Retained sovereign capability is the precondition of every other remedy. Stricter dividend rules, capital requirements, and integrated regulators all presuppose an overseer that understands what it oversees. The first investment of any reform is the state’s own capability to understand, challenge, and if necessary operate the system — not as ideology but as the minimum condition of genuine choice, including the choice not to nationalise.
L3 — Certify the failure level before the specification level. Design the regime from the question ‘what happens when it fails?’ — who acts, with what knowledge, under what pre-agreed terms — and derive the specifications from that. A licence condition that cannot be enforced at the point of failure is not a safeguard.
L4 — Price the institutional knowledge in every transaction. Mergers, disposals, and outsourcing in essential services should require an accounting of the operational knowledge transferred or destroyed, and provision for its retention — the lesson of Vertex, generalised.
L5 — Mandate cooperation where there is no market to protect. Regional monopolies have no competitive rationale for hoarding failure learning. Structural sharing of failure modes and engineering knowledge across the sector is the cheapest systemic improvement available and has never been tried.
L6 — Trace the thread of any rescue. Test each term of a recapitalisation or administration against the purpose of the service. Terms traceable only to balance-sheet survival are the old severance, renewed.
5. Candidate Insights for the Register
The following are offered as candidate deposits, phrased for the practitioner’s recognition and subject to it.
N1 — The dependency cascade nests. The cascade that hollows a state also runs inside the organisations it contracts with, level by level: each tier outsources its own understanding downward until no level of the system retains the capability the whole depends on. Auditing one level is insufficient; the cascade must be traced through the nest.
N2 — A specification that cannot be enforced at the point of failure was never a safeguard. Rules that bind only while the system is healthy are descriptions of health, not protections against failure. The test of any condition is whether it can be enforced at the moment enforcement matters.
N3 — Institutional knowledge is the asset on no balance sheet, and is therefore destroyed for free. Because accumulated operational understanding is unpriced, every transaction that destroys it books a gain. Systems that do not price knowledge in their transactions will liquidate it continuously and call the liquidation efficiency.
N4 — Drift can wear a mission statement. A severed golden thread does not announce itself; the stated purpose survives untouched in every document while the mechanics serve something else. The presence of purpose language is no evidence the thread holds — only tracing it is.
6. Inversions
| Inversion U1 Conventional: Privatisation transferred risk to the private sector; the debate is about taking it back. WHAT IF PRIVATISATION NEVER TRANSFERRED THE RISK AT ALL — ONLY THE UNDERSTANDING — SO THAT THE STATE SPENT THIRTY-SIX YEARS HOLDING THE FAILURE STATE OF A SYSTEM IT WAS PROGRESSIVELY LOSING THE CAPABILITY TO COMPREHEND? The risk of an essential service can never leave the state: when the taps threaten to stop, the state must act. What left was the knowledge required to act well. The debate about transferring risk back is a debate about recovering something that never went anywhere; the thing that must actually be recovered is the understanding. |
| Inversion U2 Conventional: The central question is whether utilities should be publicly or privately owned. WHAT IF THE OWNERSHIP DEBATE IS THE DECOY — AND THE UNOWNED THING, UNDER EVERY OWNERSHIP MODEL TRIED, IS THE FAILURE STATE OF THE WHOLE SYSTEM, WHICH NO INSTITUTION PUBLIC OR PRIVATE HAS EVER BEEN CHARGED WITH UNDERSTANDING AND HOLDING? State authorities before 1989 and private companies after it shared one property: nobody owned the question ‘what happens when this fails, and who must know enough to act?’ Both eras answered it only in crisis. An ownership debate that does not create the owner of the failure state will reproduce the failure under whichever flag wins. |
| Inversion U3 Conventional: The regulator failed because it lacked sufficient powers; give it stronger powers. WHAT IF THE REGULATOR’S BINDING DEFICIT WAS NEVER POWERS BUT UNDERSTANDING — AND STRONGER POWERS GRANTED TO A BODY THAT CANNOT PENETRATE THE STRUCTURES IT OVERSEES SIMPLY RAISE THE STAKES OF ITS BLINDNESS? The offshore tiers existed to defeat comprehension, and succeeded. Insight 11’s law — you cannot regulate what you do not understand — implies the reform sequence: capability first, powers second. A new integrated regulator inherits the vine’s remits; whether it inherits the understanding is the question the legislation must answer. |
| Inversion U4 Conventional: Restructuring and disposal are how a business finds focus and efficiency. WHAT IF EVERY RESTRUCTURING WAS ALSO A LIQUIDATION OF THE SECTOR’S TRUE ASSET REGISTER — THE UNPRICED INSTITUTIONAL KNOWLEDGE — SO THAT THIRTY-SIX YEARS OF TRANSACTIONS BOOKED AS VALUE CREATION WERE, ON THE ASSET THAT MATTERED, CONTINUOUS VALUE DESTRUCTION? Vertex is the documented instance: the understanding was sold for £217.5 million and repurchased over a decade at a multiple of it, from regulated capital. No disposal document priced what left. A sector that does not price knowledge will keep selling it for the cost of the furniture. |
| Inversion U5 Conventional: Fix the current crisis: restructure the debt, stabilise the company, then reform from where we are. WHAT IF THE DESIGN MUST BEGIN AT THE OTHER END — STANDING IN 2040, ASKING WHAT MUST HAVE BEEN TRUE OF WATER AS A SYSTEM FOR IT TO BE SERVING ITS PURPOSE THEN — AND WORKING BACKWARDS, SO THAT TODAY’S RESCUE TERMS ARE DERIVED FROM THE DESTINATION RATHER THAN THE WRECKAGE? Insight 26: strategic foresight is standing in the future world and reading the clock backwards, not projecting forward from a present that will not exist when the strategy completes. A rescue designed forward from the balance sheet optimises the wreckage. A rescue designed backward from the purpose — safe water, living rivers, a system the state understands and the least-resourced customer can afford — may share some terms with the creditors’ proposal, but it would know why, and it would know which terms to refuse. |
7. What Recovery Would Require
Whatever resolution the coming months bring — creditor recapitalisation, special administration, or the reformed regulator — the Weavers’ recovery framework sets four conditions that distinguish recovery from renamed drift. First, failure-level certification: the regime is designed from the failure state outward, with a named owner of the whole system’s failure behaviour. Second, restored sovereign capability: the state rebuilds sufficient engineering, financial, and operational understanding to audit, challenge, and if necessary operate — the precondition of every other remedy and of genuine choice between them. Third, structural cooperation: mandated sharing of failure modes and engineering learning across the sector’s companies, where no competitive rationale for silos has ever existed. Fourth, golden-thread governance: every term of every rescue, price review, and licence traceable to the purpose of the service, with the trace published. None of these depends on resolving the ownership debate. All of them are absent from every ownership model yet tried.
| The Kokomo — held open If the state can no longer run what it must ultimately stand behind, and the companies no longer hold the knowledge their own predecessors built, where does the understanding of an essential system now live — and what would it take to gather it back before the next failure asks the question less politely? |
David Sutton CITP MBCS| davesutton19@gmail.com
Produced with Claude Fable 5 (Anthropic), with the Weavers Main v51 including Appendix 1 attached as working context. Comparator: a conversation with a leading AI assistant without access to the Weavers, July 2026, supplied by the author; its figures (debt totals, dividend totals, financing share of bills, Vertex sale price) are reported as its cited sources gave them and remain contested estimates. Firsthand evidence on NORWEB and Vertex contributed by the author. July 2026 situation from current public reporting: company funding position and creditor recapitalisation proposal, reported rescue terms and ministerial objection, the parliamentary letter on special administration, credit rating and licence position, and the Independent Water Commission’s 2025 recommendation of an integrated regulator. Human recognition and judgement throughout by the author. David Sutton. This report was produced in less than an hour using the Weavers System and Claude Fable 5.