Antisocial Care
« Back to Blog

Both kinds of error

The case for an Air Accidents Investigation Branch for adult social care

19 min read

Both kinds of error

The case for an Air Accidents Investigation Branch for adult social care

Adult safeguarding in England is failing in two opposite directions at once: vulnerable adults left to die, and families wrongly investigated for caring well. The two failures look like contradictions. They are related errors with overlapping structural causes. And the reform that would address them has, in March, found a political voice for the first time in twenty years.

On 6 July 2006, Steven Hoskin's body was found at the foot of the St Austell viaduct in Cornwall. He was thirty-nine, had a learning disability, and had been tortured for months by three young people who had moved into his flat, taken his benefits, and led him around on a dog leash. In the months before his death he had contacted statutory agencies more than forty times. The subsequent Serious Case Review identified more than forty missed opportunities to intervene.

On 1 January 2015, the Court of Protection ruled that Essex County Council had unlawfully detained a ninety-one-year-old Second World War veteran in a care home for nearly two years. P, as he was anonymised in the judgment, had lived in his home of fifty years with his cat until somebody at the council became concerned that one of the people he gave money to might be exploiting him. The proportionate response, the judge later observed, would have been to apply to the court to appoint a financial deputy. The intervention chosen instead — removing him from his life — was, in the judge's striking phrase, punishing the victim for the acts of the perpetrators.

These two cases belong to different conversations in English social care. The Hoskin tradition runs through the Care Act 2014, Safeguarding Adults Reviews, the Local Government Association's national thematic analyses; it is read by social workers and quoted in safeguarding training. The Essex tradition runs through the Court of Protection's case law, the deprivation-of-liberty literature, the Mental Capacity Act commentary; it is read by lawyers and quoted in capacity practice. The two literatures barely overlap. Each constituency reads its own cases. Each tends to suspect the other of either complacency or hysteria.

They are looking at related diagnoses with substantially overlapping causes. A regime that has lost its capacity for proportionate judgement will produce both kinds of error from the same staff in the same offices on the same day. Vulnerable adults whose obvious peril it cannot see; competent or recovering adults whose homes it disrupts in the name of protection. The relationship between the surface signs and the underlying reality becomes systematically distorted. The system mis-allocates protective attention.

The two errors are not separately caused. They emerge from a shared substrate of defensive practice culture, fragmented accountability, the absence of independent learning infrastructure, and the institutional incentive to escalate rather than to judge. They are not identical errors — under-reach and over-reach involve different operational mechanisms — but they are not independent either. They are the predictable joint output of a system whose architecture rewards procedural compliance over substantive judgement.

The cost is borne, in the first instance, by the people on the receiving end of either error. It is borne in the second by the public trust in a function that, in principle, almost everyone supports. Safeguarding is one of the few words in social-policy discourse that nobody is supposed to argue with. To be against safeguarding is to be against the protection of vulnerable people. But the regime that has accumulated under that word in England now produces, at scale and with full statutory authorisation, harm that adequate safeguarding would prevent — and harm of a different kind that adequate safeguarding would not commit. The word has decoupled from the practice.

What the data shows

The Local Government and Social Care Ombudsman, which adjudicates complaints against local authorities, upheld eighty per cent of the adult social care complaints it investigated in 2023–24. This figure is the post-jurisdictional-screening uphold rate among investigated complaints — that is, among cases that survived the LGO's threshold tests and were taken forward to substantive investigation. It is 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, and the figure sustains year after year. In most regulated industries, an eighty per cent fault-finding rate on cases taken forward to substantive investigation would prompt urgent intervention. In adult social care it is treated as a baseline statistic.

Disaggregate the data and the variation between councils is enormous. The London Borough of Haringey produces upheld decisions at the rate of 34 per 100,000 residents over two years; Lincolnshire County Council, at 3.8. Bristol City Council — the case study at the centre of this investigation — sits at 15.2 per 100,000, sixth of sixty-two unitary authorities in England, sixty per cent above the unitary peer-group mean, in London-Borough-equivalent territory while operating in a regional city. The comparative data brief that accompanies this essay sets out the methodology in detail and addresses the obvious confounders: complaint volume, demographic complexity, deprivation effects, reporting culture, and the statistical normalisation choices. Bristol is not the worst council in England. It is in the top decile of its peer group on the strongest available measure, and it is well-documented enough to function as the visible example of a pattern that recurs across most ASC-responsible councils.

Bristol also has the institutional trail. Three damning multi-agency reports in four years on its treatment of disabled adults — Bijan Ebrahimi (the first UK local authority finding of institutional racism, after the murder of an Iranian refugee who had begged the council for protection); Kamil Ahmad (a disabled asylum-seeker murdered by another tenant in a council-arranged placement); Sir Stephen Bubb's 2021 Building Rights report describing a decade of "inappropriate placements and ineffective and discriminatory support and safeguarding failures." A 2024 internal audit of Bristol's £1.8 million petty-cash accounts returned the lowest possible grade — "no assurance" — finding that individual social workers were authorising cash payments of up to £500 without oversight, with clients complaining of shortfalls in cash they had been due. The Care Quality Commission rated Bristol's adult social care "Good" in May 2025 while recording, in a sentence whose institutional drabness disguises its weight, that "hidden and unheard voices were frequently overrepresented where risk was the highest."

This is the empirical landscape against which any reform must be designed. The argument that follows is structural. It applies to Bristol, but it does not single out Bristol; Bristol is one of the more visible expressions of a national pattern.

The diagnosis

The most useful theoretical apparatus for understanding what has gone wrong was developed not in social work but in organisation studies. Mats Alvesson and André Spicer's 2012 paper, A Stupidity-Based Theory of Organizations, introduced the term functional stupidity for what they defined as "the inability and/or unwillingness to use cognitive and reflective capacities in anything other than narrow and circumspect ways." Functional stupidity is not the absence of intelligence in the people doing the work. It is the organisational suppression of judgement, curiosity, and reflexive thought — a suppression that produces measurably stupid outputs from non-stupid individuals. It is, in their account, actively cultivated by organisations because it serves their immediate need for compliance and coherence.

Map this onto an English Safeguarding Adults Board. Mandatory training modules with multiple-choice tests. Six principles to be memorised. A checklist of indicators of abuse — controlling behaviour, social isolation, financial dependency, refusal to engage with services, criticism of professionals — that cannot, in itself, distinguish a coercive controller from a devoted spouse, because the behavioural surface is similar. A defensive-practice culture in which workers are sacked, named in the press or prosecuted for missing real harms, but face no consequence for over-escalating false ones. A "disguised compliance" framework which treats criticism of professionals as evidence of the very thing being denied. The literature on this is substantial. Andy Whittaker's 2018 study in the British Journal of Social Work recorded social workers describing their own practice as "fear-based" — defensive procedures applied unreflectively, because the cost of challenging them is alienation from colleagues and managers. Eileen Munro's 2011 review of child protection had said it directly: the system had become so dominated by procedure that workers had neither the time nor the permission to think.

The functional-stupidity diagnosis does not, on its own, account for everything wrong with English adult safeguarding. The sector has also been operating under sustained austerity for fifteen years. Per-capita real-terms funding for adult social care fell by roughly 13% between 2010 and 2020 while demand rose, producing caseload pressures and workforce conditions that would have produced visible harm even in a well-designed regime. Resource collapse is part of the diagnosis. So is the documented workforce crisis — staff turnover above 30% in many councils, vacancy rates over 10%, the steady loss of experienced practitioners to early retirement and other sectors. The functional-stupidity argument is the analytical layer that explains how a system under those pressures produces the specific failure patterns observed; it does not deny that the pressures themselves are real and substantial. Both are necessary to the picture. Neither, on its own, is sufficient.

This is what Steven Hoskin's case and the Essex case have in common at the structural level. In each, no individual professional decided to fail. Every worker who encountered Steven discharged their immediate procedural duty as they understood it; the harm occurred in the gap between the immediate duties, where curiosity would have to live. Every worker involved in removing P from his home believed themselves to be discharging a safeguarding obligation; the harm occurred because no one was permitted to ask whether the chosen response actually addressed the identified risk, or whether its cost to him was acceptable. Both outcomes were produced by a system that had substituted categorical compliance for thought — in conditions where, additionally, the resource and workforce environment made the substitution easier to sustain.

A system that cannot judge will fail in both directions. The errors are related, not identical. They share substantial structural causes.

The asymmetry between aviation and adult social care

Most industries do not, in fact, learn. Education's improvement curve is famously flat. Medicine outside specific niches is barely better. The built environment learns generationally. Government administration learns approximately not at all. There are, however, industries that demonstrably do learn — and the difference is structural, not cognitive.

Software is the most visible case, but the most instructive non-software case is aviation. US commercial aviation went from one fatal accident per 200,000 flights in 1960 to one per 5 million by 2010. The mechanism is the Air Accidents Investigation Branch, and its equivalents in other jurisdictions — independent statutory investigators who, when an accident or near-miss occurs, deploy with statutory powers to take possession of evidence, compel testimony, and produce a published report identifying causes and issuing recommendations to the whole industry. Crucially, the testimony given to AAIB investigators is admissible only for safety-improvement purposes: it cannot be used in disciplinary, civil or criminal proceedings. This is what allows pilots and engineers to give honest evidence about their own conduct, which is what allows the investigation to find structural causes rather than scapegoats, which is what allows the recommendations to land sector-wide rather than locally.

A natural objection arises at this point. Aviation accidents are measurable, involve relatively stable causal chains, and admit of comparatively objective event reconstruction. Adult safeguarding involves ambiguous human relationships, contested values, capacity assessments, coercion, trauma, and incomplete information. The transferability of the AAIB model is not self-evident. This is a real critique and the supporting essay in this series engages with it at length. The short answer is that healthcare faces an analogous epistemic complexity — clinical decision-making under uncertainty, with contested values and incomplete information — and the AAIB model has been successfully transplanted into healthcare twice: into anaesthesia in the 1980s and 1990s (with mortality falling roughly twenty-fold over two decades) and into NHS healthcare more broadly via the Healthcare Safety Investigation Branch, which became the statutory Health Services Safety Investigations Body in October 2023. The HSSIB precedent is the load-bearing comparator for the adult safeguarding case, not aviation directly. HSSIB demonstrates that the AAIB structural features can cross the epistemic gap from technical-causal investigations to human-system investigations and still produce learning. The adult social care equivalent would be the third such transplant. The first two have worked.

Adult safeguarding has, formally, the Safeguarding Adults Review — a multi-agency review the Care Act 2014 requires when a vulnerable adult dies or experiences serious harm. But the Safeguarding Adults Board commissioning the review is constituted of the very bodies being reviewed; the author is selected, paid and briefed by that board; the report is signed off by the same body; the recommendations are addressed back to the same body, whose decision it is whether to implement them. There is no operational independence. There are no statutory powers of compulsion. There is no safe-space testimony regime. There is no family-initiated right of investigation. There is no statutory duty on other councils to read or act on findings. The form of an aviation-style learning system has been produced; the function it requires has not.

The political moment

On 5 March 2026, Baroness Louise Casey gave her first major speech as chair of the government's Independent Commission on Adult Social Care. In it she said successive UK governments had "abdicated their responsibility to protect vulnerable adults who are at risk of abuse" and called immediately for the establishment of a National Safeguarding Board to oversee Safeguarding Adults Reviews and identify national risks. She wrote separately to Wes Streeting, the Secretary of State for Health and Social Care, urging him to set the body up at once and to commission an urgent review of adult safeguarding statutory duties.

Streeting agreed to both.

This is, in the abstract, the moment the proposal advanced in this investigation has been waiting for. A Health Secretary on the record agreeing to create a national safeguarding body. The Association of Directors of Adult Social Services has publicly welcomed it. The Social Care Institute for Excellence has welcomed it. The political consensus exists. The body will be created within the next twelve to twenty-four months.

The question is what it will be. A "National Safeguarding Board" framed as Casey has framed it could be one of two things. It could be the AAIB model transplanted to adult social care — operationally independent, with statutory powers of compulsion, safe-space testimony rules, family-initiated investigation rights, system-not-individual focus, sector-wide propagation. Or it could be a soft coordination body — an aggregation function for SAR findings, with no powers, no testimony protection, no family-initiated trigger, dependent on local SAB cooperation. The first would, on the evidence of analogous transplants into healthcare and aviation, materially reduce both kinds of safeguarding failure within a decade. The second would change very little.

The two options will look outwardly similar in the political announcement. They are radically different in their practical effect. The fight over which is built is happening now, inside the Casey Commission's Phase 1 drafting, inside the Department of Health and Social Care, and in the small number of submissions that will shape the policy detail. The Phase 1 report is due in 2026.

What this investigation has done

The companion pieces published with this essay attempt, between them, to lay out the case for the hard version of the National Safeguarding Board. The first essay sets out the corporate-stupidity diagnosis in detail, drawing on the academic literature on defensive practice, on Munro and Whittaker and Alvesson and Spicer, and using Bristol City Council as a recurring example. The second is a quantitative anchor: a two-year comparative analysis of all 151 English councils with adult social services responsibilities, ranking each on the Local Government and Social Care Ombudsman's data, with Bristol's position contextualised. The third is the structural argument for the AAIB transplant, with the six features any such body needs to function. The fourth and fifth are case studies — one composite, drawing on Steven Hoskin and the Essex case from the public record; one specific, drawing on documentary material from a Bristol family currently subject to a Section 42 enquiry that on the documentary record has no substantive basis. The sixth examines the asymmetric resource burden by which most safeguarding disputes are decided not by adjudication but by attrition — the legal-aid collapse, the administrative-burden architecture, and the structural selection of carers as the population least able to absorb the cost of contesting wrong decisions. The seventh traces the design priors of the system to the specific class that built it. The eighth and ninth are the workforce-side counterparts: the structural selection of frontline staff for compliance with the operating culture, and the post-traumatic growth cohort whose alternative trajectory the institution has structurally excluded.

Read together, they form a position. Adult safeguarding in England fails in two related directions because it has lost its capacity for proportionate judgement, has been operating under resource and workforce pressures that compound the failure, and lacks the independent learning infrastructure that comparable safety-critical sectors have built. The reform that addresses both kinds of failure is the transplant of the Air Accidents Investigation Branch model — already operating in aviation, marine accidents, rail accidents, and now NHS healthcare — into adult social care, alongside the workforce, legal-aid and design-layer reforms the supporting pieces identify. The political opportunity exists now and probably will not exist again for another decade.

If the National Safeguarding Board is built with the structural features that have made the AAIB model work in every other sector to which it has been transplanted, the regime that produced Steven Hoskin's death and P's two-year unlawful detention will, slowly and unevenly, begin to learn. If it is built without those features, the next twenty years of adult social care in England will look very much like the last twenty.

Both possibilities are live. The window in which the choice can be influenced is approximately the next twelve months. After that, whichever version has been built will be the version we have.

References and source URLs

Steven Hoskin (Cornwall, 2006)

Essex County Council v RF [2015] EWCOP 1

The Casey Commission (March 2026)

Local Government and Social Care Ombudsman data

Bijan Ebrahimi (Bristol, 2013)

Air Accidents Investigation Branch and HSSIB

Functional stupidity (Alvesson and Spicer 2012)

Munro Review and Whittaker on defensive practice

Adult Safeguarding Review · Volume I · May 2026

The series