Complex people don’t exist. Unstable systems do.

This work is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). Please attribute authorship to Vinesh Kumar (I Direct Independent Living) and link to the licence when you share or adapt this work. First published March 2026. Typeset in Garamond.

Foreword

by Dr Simon Duffy

Dr Simon Duffy is the founder of the Centre for Welfare Reform and Citizen Network. He has spent over thirty years working to advance citizenship and self-directed support for disabled people.

This paper by Vinesh Kumar is an important contribution to our understanding of what good support means and how support and care should be organised and regulated. Its roots are deep. It draws from a rich philosophical and ethical understanding of what is important and how the world really works. It also draws on the experience of providing thoughtfully designed support that has enabled people, often suffering decades of trauma, to escape institutional provision and to begin building lives of meaning in their own communities.

I will not try to summarise the paper, which speaks for itself. However, in introducing it, I want to draw attention to some simple truths and set the context.

When the big institutions were closed between 1988 and 2010, almost without exception, they were largely replaced by small institutions. Today, when people with learning disabilities or autism need additional help, this array of small-scale institutional provision remains the usual response to their needs. This is an inadequate and fragile system which frequently generates problems and further crises.

In addition, when the big hospitals were closed, the system chose to place a small percentage of people they deemed too complex for the community into small private hospitals and residential units. This system has become the new institutional structure that ends up supporting those that the small institutions fail. To a large degree, we have privatised the old institutional system and created a new profit-making system in its place.

There is an alternative, but it remains the road less travelled. In 1996, we created Inclusion Glasgow, an early effort to create a truly person-centred organisation where everyone was treated as an individual. We primarily supported people who today would end up in the new private institutions. At that time, not only did we support people to become citizens and lead much better and safer lives, we also did it at a fraction of the cost of the private institutions.

Unfortunately, such personalised support remains unusual in our current system. However, Vinesh Kumar and his wife, Subha Kumar, shocked by the abuse of the Winterbourne View Scandal, did something amazing and set up iDirect to show that this different way is possible. They now have over 10 years of successful practice. iDirect builds on the work of Inclusion Glasgow, but I particularly value, reading this paper, the maturity of the thinking that is exemplified by Vinesh and the team at iDirect.

This paper makes clear the personal and human dimension of good support. We can see that the system itself, its false assumption about quality, safety, and regulation, is actually creating risk, harm, and inefficiency. The paper calls those with power in the system to think much harder about what is really important. I believe the core concepts in the paper should become central to all our discussions about the future of care, not just in England, but in the world.

Why this work exists

This work comes from a pattern that will be familiar to anyone who has lived inside health and social care. We talk about the person but organise around the process. We talk about outcomes, but we live with delay and churn as normal. Escalation becomes routine. Responsibility gets pushed upwards until nobody can act, and the system then treats the failure as inevitable.

I have drawn on a decade of founding and building a care organisation in the South West of England. The ideas here have been shaped by thinkers and practitioners I list at the end, but most of what I describe was learned in the work. You see it when support meets real life: a plan that looks fine until someone’s week unravels, a small change that turns into drift, a family stretched thin, a colleague who spots something early, and a system that either lets them act or makes them escalate and wait. That is where the gap between what we say and what we do becomes obvious.

My belief is simple. People should be enabled to lead meaningful lives through consistent support and real autonomy in everyday decisions, not autonomy as a slogan or a box ticked after the fact. The test is whether the person can do what they wanted to do on a Tuesday, and whether the people around them have the authority to make that possible, so the person stays connected to their own life rather than being managed away from it.

A good human life is not just a matter of taste. People need certain conditions to live well. When systems remove those conditions, harm follows, even if nobody intended it. A service can look orderly and still make a person’s world smaller.

Unstable systems shrink real freedom. Risk gets managed by people who do not know the person. Decisions drift upwards and land late. The people closest to the person lose authority, and the person loses voice. That is why stability matters. It is the condition that makes freedom real rather than theoretical.

I have learned to be wary of confusing appearance with reality. In care, a system can claim protection while removing ordinary life. It can remove choice and call it safety. It can reduce a life and call it support. When judgement moves away from everyday life and into distant control, the system starts protecting itself first, and the person becomes a risk to be managed rather than a citizen to be supported.

When things get difficult, what do we actually do? Do we move closer to the person’s life or retreat into procedure? Do we keep responsibility near reality or push it upwards until nobody can act? These questions decide what happens next.

There is also a pattern I have seen too often. We treat anxiety, distress, and breakdowns in people receiving support as inherent features of who they are and miss the conditions that produce or intensify those patterns. Families and others in a person’s circle of trust often carry the load of instability for years before any service steps in. That history rarely appears clearly in assessments. Sometimes families are exhausted or frightened, and sometimes control can become part of the picture, so the system still has to keep the person’s voice real.

What stayed with me was not only the difficulty of these situations, but the improvement that often followed when conditions became stable. When relationships became regular and reliable, life often became calmer and predictable. The change could be remarkable. Nothing fundamental about the person had changed. The system around them had.

This is not another model to implement. It is a way of organising support, so judgement stays close to everyday life and ordinary life survives under pressure.

Citizenship as practice

From the outset, this work was guided by a concept of citizenship that is practical rather than sentimental. A person has a place and ties, and a life that is bigger than services. They are recognised as someone with value and expected to have something to offer in ways that make sense for them.

You see it in ordinary things. Who decides what time someone gets up? Whether they can go to the shop they like, or see the person they trust, without it becoming a request that has to travel up a chain. Whether a support plan makes those things easier or quietly makes them harder.

My thinking was shaped by the disability movement and the insistence that support should begin with personhood, relationships, and contribution rather than deficit. But we did not set out to adopt a model, and it was not another framework to display. What mattered was whether these ideas held up in the week, when plans met real life, and the work was inconvenient.

Citizenship, in this sense, was not a theory for us. It was practice. It meant staying close to people, listening properly and then acting on what we heard. It meant keeping a person’s voice real in decisions, while there was still time to make a difference.

When decision-makers barely know the person, choices get made on their behalf. Risk gets managed from afar by people who do not live with the consequences. Ordinary choices get replaced by procedures that protect professionals rather than enable lives. It can look tidy, and everyone can still say they followed the process while the person’s agency disappears quietly.

People may be kept safe, but they can also be separated from what makes life hold. Relationships thin out over time, routines get broken, and confidence disappears. At iDirect, organising support differently meant keeping judgement close to everyday life. It meant respecting existing relationships, including families and people in the person’s circle. It meant resisting systems that treated people as problems to be managed rather than citizens to be supported.

It also meant recognising something providers sometimes avoid saying out loud. Paid support has limits. Families, friends, neighbours, and community remain central long after services come and go. Citizenship cannot be proved with paperwork afterwards. You either organise around it every day, or you do not.

Lives usually stabilise when systems respect citizenship. When they undermine it, complexity follows quickly, regardless of how caring the intentions may be.

Judgement, morality, and institutions

At iDirect, we learned that complex support often fails not because people do not care, but because systems struggle to act when situations change quickly. Information is incomplete, and things shift under your feet. Decisions land late, and the consequences are rarely what anyone intended. Support can arrive after the moment has passed or fit the paperwork rather than the life the person is living.

In real situations, certainty is not available. Someone still has to act. In practice, morality is the question: what do we do next? Moral judgement, as we experienced it, meant taking responsibility for the next decision when no perfect option existed. We did not learn that from theory. We learned it when something went wrong and we had to decide what to do next, with incomplete information and real consequences.

What became clear through running iDirect is that morality does not sit only with individuals. It is shaped by how organisations are built. It shows up in where authority sits, how people are allowed to think, and how uncertainty is handled.

You see it in ordinary decisions that should be straightforward. A person is unsettled, their sleep is off, they have cancelled two activities, and a colleague who knows them well can see a pattern starting. Does that colleague have the authority to adjust support that week and act early, or do they have to escalate and wait while it slides into a bigger incident that later gets written up as behaviour?

When judgement drifts too far from everyday life, people freeze. Decisions get escalated by default. Support becomes cautious and procedural. People follow the process rather than using their minds because the system trains them to protect themselves. Autonomy depends on supported judgement. People need room to decide and learn, with a framework that helps them act rather than worry about blame.

Delay is not neutral. It changes what is possible and turns small decisions into bigger problems. In practice, it means a person waits longer for a decision that should have been made close to the reality, by people who actually know them.

People cannot live ordinary lives if every choice is escalated for approval or second-guessed by people who do not know them. Autonomy requires room to act, to choose, and sometimes to get things wrong. Our responsibility was not to eliminate risk entirely. It was to hold it thoughtfully and close to the person’s life, with the right people involved.

Judgement carries blind spots. Even experienced colleagues miss things, and familiar patterns can become normal without being noticed. If an organisation does not make space for challenge, blind spots sit quietly until harm becomes visible.

So, we shaped iDirect, so judgement was visible rather than hidden. If we could not explain a decision to the person, their family, and the team, it was usually the wrong decision. Questions from frontline colleagues who knew the person well had to be welcomed, because the people closest to the work often see drift first.

We came to treat blind spots as an organisational risk, not an individual failing that deserves blame. Moral action depends on where an organisation places authority and responsibility in everyday work. We are still learning how to hold this well, but the pattern is clear. When judgement is supported and autonomy respected, situations stabilise and blind spots surface earlier. When judgement is pushed out, life becomes complicated quickly, and the person ends up paying for a system that is protecting itself.

Leadership, moral tone, and responsibility

If judgement determines what happens in real situations, leadership determines whether judgement survives inside an organisation. You can design structures on paper, but leadership decides what people do when reality stops being tidy.

At iDirect, leadership was not about issuing instructions or enforcing compliance. It was about staying close enough to the work to notice drift early and keep responsibility near the place where action is possible. People can feel the difference between leadership that is present and leadership that retreats into policy, especially when outcomes are unclear.

Leadership meant setting a moral tone through action. Decisions had to match our values, not just sound right. What leaders noticed, questioned, and allowed to pass shaped behaviour more than any policy, because it showed people what was really expected.

You see leadership most clearly when a colleague makes a call in good faith, and it does not land neatly. In that moment, do leaders stay with the truth of what happened or reach for blame and defensiveness? Do they support judgement and learning, or do they make an example that leaves everyone else cautious? That response spreads through a workforce faster than any training session.

When things became tense, we expected people to focus on what was right, not on protecting themselves. If self-protection becomes the default, judgement collapses. So, mistakes had to be met with steadiness, learning, and fairness.

Frontline colleagues had to feel trusted to act in real time and to know they would be supported when they made thoughtful decisions, even if things did not turn out perfectly. Trust, though, had to be paired with attention. Freedom was the presence of responsibility, not the absence of oversight.

Knowledge had to be shared widely. We worked deliberately to share reasoning, not just decisions. Leaders explained why choices were made, what they were worried about, and what they wanted colleagues to pay attention to next. That is how you stop the organisation becoming a set of private opinions, with the person carrying the consequences.

Learning had to travel across teams, not get trapped in pockets, so no one became indispensable and the organisation could hold its shape as people moved on. That mattered for the people we support, because stability should not depend on one key person or one manager. We also expect leaders to protect real agency for frontline colleagues. When they cannot do that, we move early. Delay creates drift, weakens judgement, and the cost lands on frontline colleagues and the people we support.

Silence and uniformity were warning signs. When everyone agrees too quickly, it often means someone has stopped speaking or has learnt it is safer not to. Leadership, therefore, had to make space for dissent. If a colleague questioned a direction, raised discomfort, or named uncertainty, we treated that as responsibility, not disloyalty. Doubts needed to surface early, while there was still time to correct course.

Psychological safety does not appear by accident. It had to be reinforced through everyday responses, with concerns treated seriously and learning prioritised over blame. In doing so, judgement sharpened because people stopped wasting energy on self-protection and started paying attention to what was happening around them.

This work remains ongoing. Leadership still involves tough decisions, and we still face blind spots and make mistakes. But one thing has stayed true: when leadership is stewardship rather than authority, people act with care, keep thinking, and tell the truth sooner. That steadiness is felt directly by those we support and allows the system to hold when it is tested.

Work, contribution, and dignity

Alongside questions of stability and systems, another thing kept coming back: how society values work and what kinds of work are treated as serious. Much of the work that holds lives together is poorly recognised, poorly rewarded, and spoken about as if it is secondary. That is not just unfair. It shapes what systems can hold because it shapes who stays, who leaves, and how people feel about the work they do.

A lot of social care work is treated as basic or interchangeable. It is not. Supporting someone with a profound learning disability or complex mental health in their own home is not a set of tasks. It is sustained attention, judgement, and responsibility. It is noticing what is changing and acting early, before small shifts become an escalation. If we talk about skill only as technical output, we miss the work that prevents harm.

This shows up in moments that never make it into a report. Someone can shift after contact with family or friends. What matters is whether the people who know them best can respond early and consistently, while things are still manageable.

In many market systems, output and speed are what get rewarded, and relationship-based work gets undervalued. Over time, that shapes how people doing that work are spoken about, as costs to be managed rather than contributors to be respected. Markets not only allocate money; they also allocate status, and status shapes what people take pride in, what they aim for, and what gets treated as a real career.

What a society honours is chosen, and it changes. In Renaissance Italy, prestige often sat with the fresco painter and civic builder, not the celebrity. The point is not nostalgia. It is that status follows what we decide is worth admiring, and that choice shapes the society we become.

There is a further risk when market logic is allowed to organise everything. Public life thins out, and decisions start to look like technical necessity rather than civic choice. When care is treated as a commodity and labour as a cost, the work that sustains dignity gets pushed down the pile, and fewer people feel called to it as a serious form of contribution.

Once that happens, it shows up everywhere else: pay, turnover, training, and how leaders speak about the work. People receiving support feel it too, even if a service looks compliant.

At iDirect, we knew we were doing essential work and that our colleagues who did it mattered. They mattered as people with judgement, temperament, experience, and gifts. The quality of support depends on whether those qualities are recognised, developed, and retained. Treating colleagues as people rather than numbers was a practical and ethical necessity.

Investment in colleagues through continuity, internal progression, learning, and trust was inseparable from the quality of support. When colleagues feel their work is meaningful and respected, they bring more of themselves to it. When they feel replaceable, they protect themselves. That is not a character flaw. It is a rational response to being treated as disposable.

This shaped how we designed roles and led teams. Promotion from within mattered, not as a badge, but as a way of retaining learning and honouring contribution. Experience was treated as an asset, not a cost. Constant replacement is not a neutral staffing decision. It breaks continuity, weakens judgement, and increases instability. Then we act surprised when lives become less steady, and incidents rise.

The wider point is this: if we keep undervaluing work that sustains human life and dignity, turnover will stay high, and systems will not hold under pressure. Respect for work is a condition for functioning systems.

Seeing work differently, especially work that is relationship-based and morally demanding, is not only an ethical stance. It is part of building systems that can endure, rather than constantly exhausting people and then replacing them as if the work is mechanical. It is also part of protecting dignity on both sides: for the colleague doing the work and for the person whose life depends on it.

Economics, prudence, and public responsibility

Working in this way forced me to see economics differently. This work sits inside a market economy, but it also sits inside a publicly funded, overstretched health and social care system. So, cost cannot be treated as an accounting line. It is a responsibility to use public money well. Instability is expensive, and the bill always appears somewhere, even when it is delayed or disguised.

Much of the system responds to crisis with emergency placements, one-off packages, and rushed decisions made under pressure. Some services look stable until pressure exposes how fragile they are. Either way, cost rise, and outcomes worsen. This is not because people are careless. It is because the system keeps paying for the same problems late, after judgement and continuity have already been squeezed out.

Delay sits at the centre of this. Decisions come slowly, and action comes late, until crisis is the only option. By then the spend looks unavoidable, but it was created upstream in small moments that the system did not deal with early.

What made the difference economically for us was predictability, and it came from stability. Relationships stayed in place long enough for people to learn properly, and that learning stayed with us rather than being lost through turnover. Judgement stayed close to the frontline, where reality was visible, rather than being escalated late to people who could only see a version of it through forms and risk language. When judgement is close, small adjustments get made early. When judgement is distant, the system waits until a situation is serious enough to justify escalation, and then it pays more.

A steadier system is less likely to swing from calm into crisis. Fewer crises meant fewer emergency decisions, and that reduced duplication and rushed handovers. It also reduced the back-and-forth that happens when risk is pushed between the provider and the placing authority, with each side trying to protect itself. That cycle is expensive, demoralising, and predictable.

A lot of so-called savings in social care are simply costs moved elsewhere until they return as a crisis. The difference for us was building stability earlier, so we did not keep paying for the same problems twice. It also meant the wider system did not have to rediscover the same risks at the worst possible moment. On the ground, it meant fewer shocks, fewer late escalations, and a quieter system where cost was easier to anticipate because the work was not constantly being reset by churn and delay.

We did not achieve this by reducing care or lowering expectations. We achieved it by investing earlier in the conditions that make stability more likely. That meant continuity, capability, and leadership presence that held the work steady. Learning built up because people stayed, and judgement improved because it was exercised repeatedly in context, not escalated under pressure to someone who did not know the person.

This is where commissioning matters. Commissioners cannot buy stability by guessing who will be good, and they cannot buy it by writing longer specifications. But they can fund and protect the conditions that make stability more likely and stop purchasing arrangements that quietly drive instability.

They can fund continuity and expect leadership presence that is real, not remote. They can insist that frontline colleagues are listened to, because judgement lives there. They can protect space for discretion rather than squeezing it out with process. They can also avoid procurement that rewards bid writing and scale over learning, judgement, and the ability to hold a person’s life steady.

They can keep the conversation anchored in what is happening now, not only what was promised in a tender. Over the next couple of years, the temptation will be to treat faster technological tools as a substitute for stability, when the real cost control still comes from continuity and early judgement.

That is why the section later in this paper on signals, attention, and judgement matters. It offers a way of noticing drift early, while there is still time to act, and it keeps attention on the conditions that predict whether you will end up paying for crisis later.

When those conditions are underfunded, the system does not save money. It buys volatility and pays for crisis later, often in ways that get described as inevitable cost pressure rather than preventable instability. In a system under financial strain, being serious about stability is a responsible use of taxpayers’ money. It moves money away from crisis and into the quieter work that lasts. This is not a moral luxury. It is prudence.

One risk in the wider system is that the marketplace does not always reward long-term stability-building. Procurement can drift towards the familiar, where scale and bid capacity matter more than judgement, continuity, and learning on the ground. Over time, that narrows the field. You get fewer new entrants, fewer practitioner-founders, and fewer organisations willing to grow slowly and stay close to place.

This is not an argument against scale. It is an argument against models that cannot protect continuity and judgement when pressure rises. If we want systems that endure, we need a marketplace that allows serious organisations to form and survive, not only the biggest ones. Publicly funded care should not become a vehicle for value extraction. The issue is not whether investors exist, but the structures that pull money out while leaving risk and instability behind, including complex group arrangements, excessive related-party charges, leverage, and short-term targets that drive churn.

If the funding is public, the discipline should be public too, with fair tax, transparency, and reinvestment in the conditions that make stability possible. The same logic shaped our growth. Stability is what makes the economics hold, so we chose slow, deliberate expansion rooted in the South West of England. Scaling without protecting the conditions that produce stability would have been economically reckless as well as ethically unsound.

Stability does not travel well unless it is actively designed for and defended. We are still learning and still making mistakes, and new pressures keep testing these assumptions. But the pattern remains consistent. When systems are calmer, costs become more predictable. When systems fragment, costs rise in ways that are hard to see until it is too late.

Technology, AI, and responsibility

Technology, including AI, helped our operations by improving communication and removing avoidable friction, but it was never a replacement for human judgement in daily work. Used well, it reduces admin, helps information travel, and makes the basics less clunky. That matters because in social care, small delays and small misunderstandings accumulate quickly.

AI can be useful as decision support when plans are long, language is dense, and teams are under pressure. It can turn documents into plain prompts and consistent summaries. It can help someone prepare for change and communicate when stress makes words harder and help teams keep track of what helps and what makes things worse. Done properly, it can reduce missed signs and prevent avoidable escalation, giving people more time for real support.

But care still depends on relationship and responsibility. AI is good at processing text and spotting patterns. It can suggest and summarise, but it cannot carry responsibility for what happens next. That boundary has to be clear.

These tools are already being sold as more than admin support. Some are starting to look like day-to-day companions, helping a person plan, rehearse changes, and find words when stress rises. That may become genuinely useful for some people, especially when used carefully alongside human support. But it must not shift accountability away from people.

In social care, much of the work is interpretive. It depends on trust built over time, familiarity, and noticing small shifts that rarely appear in formal data. Meaning is not only in what is said. It is in what is implied, what is avoided, what is unusual for that person, and what is happening around them. A tool can process information, but it does not live inside the consequences.

Technology can support the work by improving continuity of information and reducing avoidable delay, but it cannot replace the judgement that comes from being close to someone’s life.

The real danger is not simply that machines will output recommendations. It is that organisations will start treating those outputs as decisions, and responsibility will quietly move into systems that cannot be held to account. These tools do not have morality. They reflect what we build into them and what we allow them to become in practice. That is why values and boundaries have to be explicit and why problems have to be named early.

There is also an irony here. As AI takes over more of the work that looks objective and productive – writing, summarising, planning, reporting, optimising rotas, producing policies, generating risk language, the things institutions often reward – the work that remains hard to replace is the work that sits inside relationships: staying steady with someone; holding trust; making a judgement call with partial information; and taking responsibility when the answer is not in the data. In care, that is the work that keeps lives ordinary.

At iDirect, technology remains a tool – useful, limited, and always subordinate to human judgement and moral responsibility. We keep it under review, test it in small ways, and correct course when it starts pulling attention away from the reality of the person’s life.

Signals, attention, and judgement

Over time, we found that traditional performance measures, such as key performance indicators (KPIs) and outcome metrics, often arrived too late to help. They tell you what happened after the fact. They miss what is happening as life unfolds, the small shifts, the pressure building, the quiet delays, and the way decisions get pushed upwards until nothing moves.

We were often recording outcomes after situations had escalated or costs had risen. By then the system is already reacting and explaining. That is what pushed us to develop what we now describe as signals, rather than relying on KPIs alone.

Signals are a way of noticing drift early, while there is still time to act. They are not a scorecard, and they are not a compliance tool. They sit alongside formal measures as prompts for attention, focused on the conditions that make good support more likely, judgement close to the work, continuity, leadership presence, and whether learning is staying in the system.

We did not develop signals to manage colleague behaviour or add another layer of reporting. We built them because systems rarely fail all at once. They usually weaken in small steps: decision-making slows, people start working around problems, and someone stops speaking plainly. By the time outcomes show a clear decline, the drift has often been there for weeks or months.

Signals change the conversation because they force you back to practical questions: Who knows this person well? Where is judgement sitting day to day? What happens when something goes wrong? When someone is concerned, do they speak plainly, and what happens when they do? The answers tell you whether the system is holding or slipping.

Signals can be qualitative as well as quantitative. They rely on observation, conversation, and honesty. Used badly, they become another way of policing people and managing appearances. Used properly, they help teams and commissioners name what is changing early and respond before escalation becomes the only option.

We are still formally developing this approach, but the practice is already embedded. It has helped us intervene earlier, hold relationships steadier, and notice instability beginning to form before it becomes obvious elsewhere. The point is not to replace formal oversight or outcomes. It is to add an earlier layer of judgement, so the system can correct course while it still can.

We have included two short examples of how signals work in practice in the Appendix section.

Closing reflection

I offer this work as unfinished, because it is. We still make mistakes, and new pressures keep testing us. There is no point where you get to say the work is done and relax. What exists instead is a discipline: paying attention, correcting course, and taking responsibility when things fail to hold.

Some things have stayed true under pressure. When the conditions around a person change, even imperfectly, life often becomes calmer and more ordinary again. The shift can be obvious, not because the person’s life has improved, but because the system has stopped making it harder than it needs to be.

In the end, the measure is practical freedom. Can the person shape their own day in ways that make sense for them? Can they keep relationships and remain part of their own life? Can they recover when things go wrong without everything turning into an incident and an escalation? We have to stay honest about that freedom and not confuse neat paperwork with a life that is truly holding.

I have learned that good support is less about control and more about judgement. It is knowing when not to intervene and holding the conditions so people can keep living rather than being managed. The moment support becomes control, the person starts to disappear.

That means sharing responsibility rather than passing it upwards. It means staying steady when situations are difficult and acting early when something is drifting, before crisis forces your hand. It means having the nerve to respond while there is still time to keep things ordinary.

Leadership, in this sense, is not heroic. It is consistent, present, and serious about doing what is right when there is no applause for it. It is protecting judgement in the organisation so that responsibility stays close to the person’s life.

The same responsibility applies at an organisational and civic level. Providers operating within public systems are custodians of trust and public money, not just holders of contracts. Responsibility rarely disappears. It reappears later at greater cost, often as crisis spending that everyone claims was unavoidable.

That is why stability matters. It is the condition that makes freedom real, and it is what makes the economics hold. Crisis spending is what happens after judgement and continuity have already been pushed out.

iDirect has not solved these problems, but we have learned one pattern again and again: instability causes complexity more often than people do. When systems hold well, it becomes easier for people to live ordinary lives.

That is the responsibility, and that remains the work.

References and influences

These people and resources shaped how this work developed in practice. I list them here for anyone who wants to see the roots of the ideas.

• Amartya Sen – justice, freedom, and enabling people to live lives they have reason to value.
• Philippa Foot – ‘natural goodness’: the idea that a good human life is not a matter of taste, and people need certain conditions to live well.
• Václav Havel – ‘living in truth’: the moral weight of ordinary choices and refusing to confuse appearance with reality.
• Franco Basaglia – institutions can ‘negate’ the person while claiming to protect them.
• Simon Duffy – citizenship, disability rights, and support rooted in ordinary life.
• John O’Brien – personhood, contribution, and the role of relationships and community.
• Karl Popper – learning through error, correction, and institutional openness.
• David Deutsch – moral knowledge as provisional and improvable through reason.
• Donella Meadows – how systems behave, where instability forms, and where leverage lies.
• Zeynep Ton – economic discipline through investment in people and continuity.
• G. Venkataswamy – Aravind Eye Care System: purpose-driven care built on discipline, scale, and compassion.
• Ivan Illich – limits of institutions and unintended harm in health systems.
• Michael Sandel – the moral value of work and contribution to society.
• Karl Polanyi – when markets organise everything, human and civic needs are pushed aside.
• Peter Drucker – management as a moral and social practice.
• Nassim Nicholas Taleb – fragility and hidden risks in systems.
• The King’s Fund – analysis of system fragmentation in health and social care.
• Joseph Schumpeter – change, renewal, and disciplined disruption.

Author note

Vinesh Kumar is the founder of iDirect, an independent living organisation based in the South West of England. The reflections in this paper arise from building and correcting iDirect over a decade, within publicly funded health and social care systems.

Appendix: Eight signals of system stability

This appendix lists eight signals that help commissioners notice whether a system is likely to hold under pressure or whether instability is starting to form. They sit alongside KPIs and regulatory oversight as early prompts for attention, not as targets or a scoring tool.

They are designed to prompt a few simple questions early:

• What is changing? What has shifted recently in the person’s life or the support around them.
• What is holding or drifting? Are the basics staying steady or quietly weakening over time.
• What support is needed now to prevent escalation? What would help the system hold before it tips into crisis.

Used together, the signals keep the discussion grounded in everyday reality, making drift visible sooner while there is still time to act.

1. Continuity around the person

When relationships are stable, lives tend to stabilise; when they churn, complexity rises early, often before anything shows up in the data.

2. Judgement stays close to everyday life

Excessive escalation is usually a sign that people do not feel trusted or safe to use judgement.

3. Leadership presence where the work happens

Tone travels faster than policy, and leadership behaviour sets the limits of judgement.

4. Psychological safety and speaking up

Most systems fail quietly before they fail publicly; if people cannot speak, blind spots become normal until an incident forces attention.

5. Learning stays in the system

When people leave, learning leaves with them; stability allows knowledge to accumulate and judgement to strengthen.

6. Colleagues treated as trusted professionals

Support quality depends on people’s judgement and capability, not the process alone.

7. Families and circles of trust treated as partners

Families and trusted others often hold vital knowledge long before services become involved.

8. Working relationships across organisations

Fragmentation shows up at organisational boundaries first, and instability grows fastest there.

Two examples of signals in practice

These are short examples based on real patterns in our work, with details changed to protect privacy.

Example 1: Drift noticed early

Someone we support was mostly steady, then the week started to slip. Sleep dipped, two activities were cancelled, and the person’s tone changed. The paperwork would not have flagged it yet, but the people who know them well could see it building.

The signal was not the sleep dip. The signal was continuity and judgement close to everyday life. The same colleagues were there, they had context, and they had the authority to act. They adjusted the week early, took off what was not necessary, and leaned into the one relationship that reliably steadies the person. Leadership backed it quickly instead of sending it up a chain.

Within days, the person settled, and the drift stopped becoming a story. We avoided the usual sequence: an incident, a rushed response, and restrictions added afterwards to look safe. The cost was small, and it was paid early: attention, a modest change in staffing focus, and permission to act in real time.

Example 2: A concern raised early

In another setting, on paper things looked fine. The risk was not a crisis. The risk was that a slightly controlling way of working had become normal, justified as safety. A frontline colleague named it early, plainly, as discomfort.

The signal here was psychological safety, backed by leadership presence. What mattered was not that someone raised it, but that they could, and that it was treated as protection rather than trouble. We stayed with the reality, asked what had changed and where the pressure was, and checked what the person’s life actually felt like day to day.

We changed course quickly. We reduced unnecessary controls, clarified who had day-to-day authority, and put the person’s preferences back at the centre. The person became more engaged and tension dropped. That correction happened because someone could speak early and the organisation was willing to listen, learn, and act before crisis forced our hand.