The architecture of structural irrationality in adult safeguarding

A statutory regime built to protect vulnerable adults from harm appears, by the structural logic of its operating incentives, to be associated with significant harm of its own. The pattern is recognisable across England, and Bristol provides an unusually well-documented case study.

A statutory regime that began as a corrective to institutional indifference — to the failures exposed at Winterbourne View, to the deaths of Steven Hoskin and Fiona Pilkington, to the long history of vulnerable adults left to drown in the gaps between services — has, in the years since the Care Act 2014, developed structural features that warrant separate examination. It still names itself protection, and in many cases it does protect. The records examined here also disclose a substantial parallel pattern of harm produced by the regime's own operation — harm that the system has not yet developed the categories to name. This report examines the shape of that harm in adult safeguarding, the organisational mechanisms that appear to produce it, and the question of whether Bristol — a city whose name recurs in the academic, regulatory and journalistic record more often than would be expected for a city of its size and demographic profile — represents an exceptional case or a clearly visible instance of the national norm.

The framing chosen here uses a phrase this essay adopts as its organising shorthand: structural irrationality. It belongs to the same conceptual family as Mats Alvesson and André Spicer's functional stupidity,^[1] the academic term this essay relies on for the substantive analysis. The shorthand is useful because it names what the academic apparatus has to argue toward: that the outputs of certain institutional arrangements would, to any outside observer, be difficult to recognise as rational, regardless of whether anyone inside the institution finds them useful. Both terms name a phenomenon the literature now recognises: not the absence of intelligence in the people doing the work, but the organisational suppression of judgement, curiosity, and reflexive thought — a suppression that appears to produce poor outputs from cognitively capable individuals. In adult safeguarding, the documented outputs include carers wrongly accused of coercion, families separated by best-interests decisions never tested against the family's own knowledge, ombudsman complaint uphold rates that climb year on year, and a regulatory architecture in which the people most affected by decisions remain structurally absent from them.

None of this is an argument against safeguarding. The duty exists because the alternative is documented and lethal: Steven Hoskin tortured and driven to his death, Fiona Pilkington unable to get an adequate response to years of harassment, the residents of Winterbourne View abused on camera by the staff paid to care for them. A system that intervened too rarely produced those outcomes, and the instinct to escalate was the correct correction to a real and deadly under-reaction. Safeguarding also works, quietly and constantly, in cases that never reach an Ombudsman or a review — the financial abuse stopped, the neglect caught, the coercion named. The argument advanced here is narrower, and in a sense harder: that the same regime, driven by the same protective instinct, now also generates a second pattern of harm through the structure of its incentives, and that the system has not yet built the categories to see it. The failures of under-intervention are well understood and have produced more than a decade of reform. The failures of over-intervention are not yet counted at all. Nor is it an argument for non-cooperation: the critique is of how the system processes concerns once they reach it, not of whether concerns should be raised at all — a relative worried about someone's safety, or a professional who sees neglect, should still report it. This essay is about the uncounted side of the ledger.

Part One — The scale of the regime

Adult safeguarding in England is a statutory duty under section 42 of the Care Act 2014, discharged by 153 local authorities and overseen by the Care Quality Commission, the Local Government and Social Care Ombudsman, and locally constituted Safeguarding Adults Boards. The volume is large and growing.

A note on scope. "Adult safeguarding" is used in this essay in its wider administrative sense, not only the section 42 enquiry. It takes in the surrounding culture of risk escalation, Mental Capacity Act and best-interests practice, the Deprivation of Liberty machinery, Safeguarding Adults Review and audit systems, and the accountability incentives that run through all of them. These are formally distinct processes, operated under different parts of the legislation, but they are carried out by overlapping people under a shared set of pressures — and the argument advanced here is precisely that the pressures, rather than any single process, are what shape the outputs.

The growth in concerns and enquiries is sometimes presented as evidence of better detection. It is also, less comfortably, consistent with a regime expanding its catchment — drawing in cases that earlier definitions would not have classified as safeguarding at all. The 33.8% fall in "Other" enquiries (those outside section 42 thresholds) in 2024–25^[2] suggests local authorities have been reclassifying upward, into the statutory frame, where the procedural machinery is heavier and the threshold for closing the case is higher. The system, in other words, is not just receiving more concerns; it is processing them more formally.

An uphold rate of 80% on investigated complaints is striking on its own terms. It is the post-jurisdictional-screening fault-finding rate among cases the Ombudsman has decided warrant substantive investigation — not directly comparable to general fault rates in regulated industries, where the denominator is calculated differently. But it is the most rigorous independent adjudication of council conduct available in adult social care. In most regulated industries, an 80% finding of fault on cases taken forward to substantive investigation would prompt urgent intervention. In adult social care it is treated as a baseline statistic.^[3]

Part Two — What the literature actually says

The academic and policy literature on adult safeguarding does not use the word "stupidity." It uses politer terms — defensive practice, proceduralism, fear-based practice, thresholds creep, professional helplessness, compliance culture, tickbox safeguarding — but the phenomenon described is recognisably the same: cognitively capable people in statutory roles producing decisions that, examined retrospectively, often fail tests of proportionality that the same workers would apply in their non-professional lives.

Defensive practice as the open secret

Whittaker's interview study of 70 social workers, published in the British Journal of Social Work, found that defensive practice was understood by practitioners themselves as fear-based behaviour driven by organisational anxiety, not by client risk.^[4] One participant described it bluntly:

"Defensive practice is about not challenging management decisions and following procedures in an unreflective and passive way … Avoiding challenging bad systems. Avoiding challenging because there are repercussions on you as an individual, because you'll be distanced and alienated from your colleagues." Participant 6, Whittaker (2018)

Another said they had been told directly by a senior colleague: "It is not always worth arguing with whatever you see. You know it's not right, but it's not worth it."^[4] The same study identifies two distinct modes of defensive practice — defending against the service user (avoiding contact, over-inflating risk to avoid blame) and defending against the organisation (not challenging managers, not raising concerns).^[5] Both produce decisions that look incompetent from outside and feel rational from inside.

The Munro inheritance

Eileen Munro's three-part review of child protection for the UK government (2010–11) became the canonical statement of the problem in safeguarding generally, even though its remit was children. Munro argued that the system had "become over-bureaucratised and focused on compliance" and that the cost of this had been the systematic erosion of professional judgement.^[6] The fix she proposed was a return to professional autonomy and reflective practice. Critics on the social-work left noted at the time that Munro's reforms placed the burden of repair back onto frontline workers without addressing the asymmetric punishment structure that had produced defensive practice in the first place.^[7] Fifteen years on, the procedural drift Munro diagnosed has, by most accounts, deepened rather than reversed — and the adult safeguarding regime under the Care Act has inherited similar institutional features without ever being subject to a Munro-style review of its own.

Professional curiosity, weaponised

"Professional curiosity" is the post-Munro term of art for the disposition workers are supposed to have: probing, questioning, refusing to accept the surface presentation of a family. It is invoked in every Safeguarding Adults Review where harm was missed.^[8] The Norfolk Safeguarding Adults Board, whose published work on this is among the most developed, describes professional curiosity as "having the capacity and communication skills to explore and understand what is happening with an individual or family … asking more, and using proactive questioning and challenge."^[8]

The asymmetry is in how the concept is operationalised. Professional curiosity is treated as a one-way virtue: practitioners are trained to be suspicious of the family's account, but the family's curiosity about the practitioner's reasoning is read as a different category — sometimes labelled disguised compliance. Leeds City Council's safeguarding briefing, drawing on the Safeguarding Network's framework, lists the markers of disguised compliance as "focusing on one particular issue; being critical of professionals; failing to engage with services."^[9] The structural problem with this is significant: anyone who criticises a professional's judgement can, within the framework, be classified as exhibiting evidence of the very thing the professional has decided is wrong. The framework is, in operation, difficult to falsify. Once the disguised-compliance frame is in play, the family's protests against it can be absorbed as further evidence supporting it.

A framework difficult to falsify in operation: the family's criticism of a wrong call can be absorbed as evidence supporting the call.

The same structural mechanism — legitimate criticism from people an institution serves being reframed by the institution as a character defect of the critic — was identified by Sir Martin Moore-Bick in the September 2024 final report of the Grenfell Tower Inquiry, in a different sector but with recognisably the same shape.^[28] The inquiry found that residents of Grenfell Tower who had raised consistent and substantive fire-safety concerns over years were, in the Kensington and Chelsea Tenant Management Organisation's processing, recharacterised as "militant troublemakers." The residents experienced the institution as "an uncaring and bullying overlord that belittled and marginalised them." Moore-Bick described the resulting relationship as "a toxic atmosphere fuelled by mistrust on both sides." The classification was structural rather than individual — propagating through how the TMO trained, supervised, and rewarded its staff — and the operational consequence was that the residents' warnings ceased, in the institution's processing, to function as information about fire safety. They became, instead, evidence about the residents. Moore-Bick's framing is, in safeguarding's adjacent regulatory sector, the most authoritative recent public-inquiry statement that the asymmetric-framing mechanism described above operates at scale in state-adjacent institutions and produces catastrophic safety failures. The mechanism does not name itself; it has to be identified from its outputs.

The carer-harm definition gap

A 2025 study in the British Journal of Social Work by Donnelly and colleagues, looking at the experience of family carers in Ireland (whose safeguarding architecture parallels the UK's), identifies a specific definitional failure: the dominant academic definition of "carer harm" presumes intent and coercive behaviour on the part of the cared-for person.^[10] In practice, carers report harm that is unintentional, relational, contextual — and falls outside the categories the safeguarding system can recognise. The result is "professional helplessness and a reluctance on the part of services to engage with carer harm."^[10] The system cannot easily see what does not match its categories, and what it cannot see, it cannot redress.

The Mental Capacity Act and DoLS

The Deprivation of Liberty Safeguards regime, intended to provide procedural protection against unlawful deprivation of liberty in care settings, has been the subject of consistent regulatory criticism from the CQC.^[11] The replacement scheme — Liberty Protection Safeguards — was originally legislated in 2019, postponed because of the pandemic, and then in April 2023 postponed indefinitely "beyond the life of this Parliament." The CQC's response, repeated in successive State of Care reports, has been to warn that "the current system is not effectively protecting the rights of many people who use health and social care services" and that "many of the issues we raise risk infringing people's rights or even contributing to abuses of individuals' rights."^[11] A regulator describing the safeguard regime itself as a possible site of rights abuse is the formal version of the lay observation that something has gone wrong with safeguarding's relationship to its own purpose.

Part Three — The Alvesson–Spicer frame

To name the mechanism precisely, the most useful theoretical apparatus is Alvesson and Spicer's 2012 paper in the Journal of Management Studies, "A Stupidity-Based Theory of Organizations,"^[1] expanded into book form in 2016 as The Stupidity Paradox.^[12] Their concept of functional stupidity is defined as:

"the inability and/or unwillingness to use cognitive and reflective capacities in anything other than narrow and circumspect ways … an absence of reflexivity, a refusal to use intellectual capacities in other than myopic ways, and avoidance of justifications." Alvesson & Spicer, 2012

Three features of the concept matter for safeguarding. First, functional stupidity is not a property of individuals; it is an organisationally supported state. The same person who, in their personal life, would solve a problem with three minutes' thought, may, in their professional role, follow a procedure that produces an obviously poor outcome. Second, it is functional — that is, it appears to serve the organisation's coherence, hierarchy, and self-protection, which is part of why organisations cultivate it. Third, it is reinforced through what Alvesson and Spicer call "stupidity management": vision statements, mandatory training, the language of compliance, and the discouragement of dissenting thought.^[12]

Map this onto an English Safeguarding Adults Board. Vision statements: "Making Safeguarding Personal." Mandatory training: annual safeguarding modules with multiple-choice tests. Compliance language: section 42, threshold criteria, the six principles. Discouragement of dissent: the disguised-compliance framework, the professional-curiosity asymmetry, the well-documented reluctance of frontline workers to challenge managers (per Whittaker). Many of the conditions Alvesson and Spicer identify as conducive to functional stupidity are recognisably present in adult safeguarding institutions. It would be more surprising if the outputs were consistently rational.

The shorthand structural irrationality captures something slightly different from, and complementary to, the academic term. "Functional stupidity" emphasises that the unreason has a function for the organisation — academically defensible, but tonally generous to the institution. "Structural irrationality" foregrounds the other half: that the behaviour is built into the architecture of incentives and procedures rather than into the people, and that its outputs would be hard for any outside observer to call rational. For investigative purposes both have their place: structural irrationality names the observable phenomenon; functional stupidity is the analytical apparatus that explains why it persists. The body of this essay relies on the technically precise term where the argument is doing analytical work, returning to the plainer framing at the structural-summary moments.

Part Four — Bristol: does it stand out?

The question asked here is empirical: in the records of the Local Government and Social Care Ombudsman, the Care Quality Commission, the Safeguarding Adults Reviews library, the local press, and the academic literature, does Bristol City Council appear more often than would be expected for a city of its size and demographic profile? A qualitative survey of regulatory findings, SARs, audit reports and Ombudsman data suggests Bristol is unusually prominent in the safeguarding record. The evidence is multi-stranded, and the comparative statistical anchor is provided in the data brief that accompanies this essay (Bristol is in the top decile of unitary authorities for upheld complaints per capita), which ranks all 151 ASC-responsible councils in England on the Ombudsman's published data over two combined years and places Bristol 6th of 62 unitaries on upheld decisions per 100,000 residents. The following sections set out the qualitative evidence alongside the statistical picture.

Ombudsman uphold rates

The most precise comparator is the Local Government and Social Care Ombudsman's own published statistics, which adjust for council type and population.

A 6-percentage-point gap in the uphold rate, sustained across an investigative dataset, is a meaningful signal. The Ombudsman's own framing of Bristol's record is also unusually pointed: the regulator has formally noted that the council has, on more than one occasion, failed to provide evidence of completed remedies — meaning the council told the Ombudsman it had implemented recommendations and was found not to have done so.^[13]

On method. The comparative figures here, and the top-decile ranking, are drawn from the Ombudsman's published decision data for the two most recent complete reporting years combined, expressed as upheld decisions per 100,000 residents using ONS mid-year population estimates. The comparator group is England's other adult-social-care-responsible unitary authorities — a like-for-like set that controls for the structural differences between unitary and county/district arrangements — and the same calculation run across all 151 ASC-responsible councils is what places Bristol in the top decile nationally. The full denominators, the comparator list, and the per-council workings are set out in the companion data brief, so the ranking can be checked or contested directly. One caveat travels with any per-capita complaint measure: it counts complaints that reached the Ombudsman and were upheld, not the underlying rate of poor practice, which is unobservable. A higher figure is therefore consistent with both worse practice and a more complaint-literate population — a tension the closing sections of this part address directly.

Safeguarding Adults Reviews

Bristol Safeguarding Adults Board (now the Keeping Bristol Safe Partnership) has produced or been a partner in a notably high-profile sequence of Safeguarding Adults Reviews and Serious Case Reviews over the last decade:

• Winterbourne View (2012) — abuse by staff at a private assessment and treatment unit in South Gloucestershire, exposed by BBC Panorama; nationally galvanising.^[14]
• Bijan Ebrahimi (2013) — disabled Iranian refugee murdered by a neighbour after years of reporting harassment to Bristol City Council and Avon and Somerset Police. The Safer Bristol Partnership Multi-Agency Learning Review found "evidence of both discriminatory behaviour and institutional racism" on the part of both Bristol City Council and the police — described by Bhatt Murphy Solicitors (who represented the family) and by Disability News Service as the first finding of institutional racism of its kind against a UK local authority, and, after Macpherson, the second against a UK police force.^[15]
• Melissa (2014) — 18-year-old murdered shortly after transition from children's to adults' services; the SAR identified failings in transition planning, risk assessment, and out-of-area placements.^[16]
• Kamil Ahmad (2016) — disabled asylum-seeker murdered by another tenant in a Bristol mental-health-related housing scheme; SAR identified multi-agency failings.^[17]
• Sir Stephen Bubb's Building Rights report (2021) — the third damning multi-agency report into the city's treatment of disabled adults in four years, describing a decade of "inappropriate placements and ineffective and discriminatory support and safeguarding failures."^[18]

For a city of 483,000 people, that is a heavy sequence. Disability News Service noted in 2021 that activists were "horrified" by what had by then become a recognisable pattern of repeating, themed failure across what should have been distinct services.^[18]

The 2024 imprest audit

In July 2024 Bristol City Council's internal auditors returned the lowest possible audit grade — no assurance — on the council's £1.8m of petty-cash (imprest) accounts. The audit found:^[19]

• Individual social workers requesting, authorising and issuing cash payments of up to £500 with no oversight.
• Inappropriate use of imprest accounts to circumvent council payment systems.
• Inadequate arrangements to ensure clients received correct cash payments, and limited follow-up of client complaints of shortfalls in cash received.
• No follow-up checks on the validity of emergency payments.
• No central record of what was being paid out.

The audit identified seven high-priority and two medium-priority findings. The report described the safeguarding-relevant subset of imprest spending as "statutory payments to those in need." From a structural-irrationality standpoint, this presents as a clean case: an organisation handling vulnerable-adult cash payments at scale, with limited functioning controls, in a context where clients had already complained that they were receiving less than they were due. Whether any of that money was misused remains formally unknown, because the audit trail to test the question does not exist.

The CQC's "Good" rating, in context

In May 2025 the CQC published its first local-authority assessment of Bristol's adult social care under the new Care Quality Commission powers (extended to councils by the Health and Care Act 2022). Bristol was rated Good.^[20] Within the supporting text, however, the assessors recorded that 88% of staff had completed mandatory safeguarding training (above the national average), that average waiting time to enquiry decision had reduced from nine to six days, that 506 people were still waiting for a safeguarding response, and — most pointedly — that "hidden and unheard voices were frequently overrepresented where risk was the highest."^[21]

That last phrase is bureaucratic, but it is also one of the more telling sentences in the entire CQC dossier on Bristol. It says, in formal language, that the people most at risk of harm are also the people least likely to be heard by the system that is supposed to protect them. This is not a Bristol-specific finding — the CQC's own State of Care reports make similar observations nationally^[11] — but it appears in Bristol's assessment as a named characteristic, not a passing concern.

The December 2025 audit

In December 2025 the Bristol Post reported, citing council audit findings, that "too few quality checks" were being carried out on the roughly 500 care firms now contracted to provide the majority of adult social care in the city — a number that has "rapidly risen" in recent years.^[22] Councillors described the audit findings as "quite worrying." Bristol now plans to spend £268.7 million on adult social care in 2025–26.

The financial context

Grant Thornton, Bristol City Council's external auditor, has warned for two consecutive years (2023 and 2024) that the council "may not be financially sustainable" because of social care and SEND overspends. The reported overspend of £3 million in 2022–23 was found to mask a forecast gross service overspend of £57.9 million.^[23] Financial precarity at this scale is not in itself a safeguarding failure, but it is the constant background pressure against which every individual safeguarding decision is made — and it is associated with a documented organisational pattern of incentive distortion: closing cases faster, raising thresholds in places that aren't measured, lowering them in places that are.

For a city of 483,000, an upheld-complaints rate 49% above the unitary peer average, three damning multi-agency reports in four years, and a "no assurance" finding on its statutory cash payments is a pattern recurring across multiple regulatory and review domains over a decade.

What the pattern means

Bristol's profile across these data sources is consistent with two interpretations, which are not mutually exclusive. One: Bristol is genuinely worse than its peer councils on adult safeguarding — its complaint uphold rate, its sequence of SARs and the institutional-racism finding, its audit failures, and its repeated independent-review findings all point that way. Two: Bristol is no worse than average, but its problems are unusually well-documented because of an unusually active local press (the Bristol Cable, Bristol247, the Bristol Post's local democracy reporting), an engaged disability-rights community (Bristol Disability Equality Forum, SARI), and a city culture of institutional self-examination that has produced more Sir-Stephen-Bubb-style reports than other comparably sized cities have commissioned.

The two interpretations diverge in their implication. If Bristol is genuinely worse, the question is what makes it so. If Bristol is merely more legible, the question becomes how bad the unseen councils are — because the documentation suggests something serious is happening here, and the safeguarding statistics nationally suggest nothing about Bristol is fundamentally unique. The most likely answer, on the evidence, is that Bristol is both somewhat worse than average and better-documented than average. The combination makes it useful as a case study, but it would be a mistake to treat it as exceptional.

Part Five — How structural irrationality gets built

Synthesising the literature with the Bristol record, the institutional mechanism by which adult safeguarding tends to produce poor outcomes from cognitively capable people can be specified with reasonable precision. Six pressures, all well-documented, appear to compound each other.

Asymmetric punishment. A worker who fails to escalate a real concern, and a vulnerable adult dies, will be named, prosecuted in extreme cases, and used as a cautionary example in training. A worker who escalates a non-concern and disrupts a family's life will, in most cases, face no formal consequence; the family has no straightforward route to bring a complaint that finds the worker, only the council, at fault, and the Ombudsman's remit is restricted to maladministration by the institution rather than misconduct by named individuals. This rationally incentivises over-escalation.^[4]

Threshold creep and audit pressure. Local Authorities are measured on the number of section 42 enquiries they complete and the proportion that "remove or reduce" risk (a national average of around 90%).^[24] The metric rewards finishing cases more than getting the threshold right. The 33.8% national fall in "Other" enquiries in 2024–25, with a corresponding rise in section 42s, is consistent with the metric's footprint.^[2]

Training-induced checklist thinking. Safeguarding training, especially the mandatory annual variant, is built around indicators of abuse: controlling behaviour, financial dependency, social isolation, refusal of professional involvement, criticism of professionals. Many loving long marriages exhibit some of these indicators on the surface. Indicator-based safeguarding frameworks, used as the primary screening tool rather than as one input alongside interviews, multi-agency evidence, chronology and capacity assessment, can struggle to distinguish coercive control from intense but legitimate caring relationships, because the behavioural surface is similar. Workers are not failing to think; the training tends to crowd out thought, and the institutional pressure tends to discourage challenge to the framework.^[8]

Inversion of expertise. The family member has years of intimate knowledge of the cared-for person. The social worker has, at most, a few hours of contact and a file. "Professional judgement" is institutionally weighted above "lived knowledge." This produces outcomes that look implausible from outside and feel professional from inside.

Unfalsifiable framings. The disguised-compliance framework treats criticism of professionals as evidence of the very thing being denied.^[9] Mental Capacity Act assessments, where they are used to override family knowledge, are largely insulated from family challenge unless the family has the legal resources to take the case to the Court of Protection. The Court of Protection's adversarial cost structure makes that route effectively unavailable to most families. (The asymmetric burden this creates is the subject of a separate piece in this series.)

Mandatory reporting culture. "Safeguarding is everybody's business" is one of the system's most repeated mantras. Its inverse, in practice, is that anyone can trigger a safeguarding investigation against anyone else, and the investigation runs even where the referral is groundless or malicious. This is documented in research-grade evidence — the Action for M.E. Families Facing False Accusations survey and the Cerebra/University of East Anglia report on Fabricated or Induced Illness allegations document the same pattern in clinical-services-adjacent safeguarding.^[25] The FACT (Falsely Accused Carers and Teachers) helpline carries the accumulated case work of the wider problem.^[26] Community testimony in family forums (Mumsnet threads, separated-fathers' networks) corroborates the pattern at the level of lived experience — though such forums are anecdotal rather than research-grade, they sit alongside, not as substitutes for, the academic and helpline-derived evidence.

Hannah Arendt's frame — that institutional harm can arise from routinised thoughtlessness rather than individual malice^[27] — is the philosophical complement to Alvesson and Spicer's organisational one. The two analyses converge on the same picture: a system whose architecture appears to suppress the cognitive operations its own stated purpose requires. The adult safeguarding regime in England is, on present evidence, an institutional arrangement to which this diagnosis applies. Bristol is one of the places where the diagnosis is most visible in the documentary record.

The evidence reviewed here suggests that training alone is unlikely to correct the problem. Bristol staff are trained at above the national average and the outputs remain the outputs. The corrective is structural: symmetric accountability, falsifiable framings, family expertise weighted equally with professional judgement, an Ombudsman remit broad enough to find against named workers as well as institutions, and a Liberty Protection Safeguards regime that is actually implemented. The independent investigation body proposed in the supporting essays of this series — modelled on the Air Accidents Investigation Branch — would address the regime's failure to learn from its own outputs. None of these reforms is in any current programme of government. Until they are, on the structural argument advanced here, the regime is likely to continue producing, at scale and with full statutory authorisation, harm of the kind it was designed to prevent.