Public policy is built around intervention. A problem becomes visible, government identifies a cause, resources are applied and results are measured. That sequence is sensible where cause and outcome are reasonably direct. Many social problems, however, are produced by systems in which causes become consequences, consequences alter behaviour, and that behaviour feeds back into the original problem.
Youth homelessness illustrates the difficulty. A young person leaves an unsafe or unstable home and loses secure accommodation. Housing instability disrupts schooling or employment, lower participation makes income harder to sustain, reduced income narrows housing options, repeated moves weaken social networks and mental health may deteriorate. The original housing problem becomes entangled with consequences that feed back into it. Housing instability is simultaneously producing and being produced by other pressures.
IEP’s Halo framework describes this through feedback loops. Reinforcing loops amplify movement in one direction, while balancing loops resist change and push a system towards an established state. Once a reinforcing loop develops, asking what originally caused the problem may become less useful than identifying what is now keeping it going. Emergency accommodation may be essential, but accommodation alone may not change the trajectory if disrupted education, poor health, weak income and social isolation continue feeding instability.
Systems also carry history. Halo describes this as path dependency: the way a society, institution or community responds to a new intervention depends partly on the route by which it reached its present condition. Two communities can receive the same consultation process and interpret it differently because one has repeatedly seen engagement lead to visible change while another has repeatedly been consulted without meaningful influence. The formal process may be identical, but the system receiving it is not. This is also why changing written rules does not necessarily change behaviour. A department can introduce a whistleblower policy while employees remain convinced that speaking up will damage their careers; a government can mandate consultation while communities continue to believe the decisions have already been made. Institutions operate through both formal rules and the accumulated expectations people have learned from experience.
Systems also resist change through what Halo calls homeostasis. Existing structures, incentives and habits create pressures that tend to return the system towards a familiar state. A government may announce major reform, yet legislation, budgets, procurement systems, professional cultures, accountability arrangements and institutional memory remain largely unchanged. Staff know which risks are punished and which behaviours advance careers. Programs have existing contracts and constituencies. Measures of performance have accumulated around the old way of working. Reform enters a system with momentum, and its eventual shape may become more modest than its original ambition without anyone consciously sabotaging it.
Australia’s experience with Closing the Gap provides a substantial demonstration of the problem. The National Agreement covers outcomes spanning health, education, employment, housing, justice, child development, safety and wellbeing, yet it also recognises that progress cannot be separated from the way governments themselves operate. Its four Priority Reforms focus on shared decision-making, strengthening the Aboriginal and Torres Strait Islander community-controlled sector, transforming government organisations and sharing access to data. The Productivity Commission’s July 2026 report shows how difficult the challenge remains: of 19 national targets, one had been met, three were on track, five were improving but not on track, and four were worsening.
The significance lies not simply in the disappointing numbers. The agreement itself represents a shift in diagnosis. The problem is no longer framed only as a collection of service deficits requiring better individual programs. Decision-making authority, institutional behaviour, data access and the relationship between communities and government are treated as part of the system producing the outcomes. This does not prove that better coordination or community control will automatically produce better results, nor does it remove the need for adequate health, housing, education and employment services. It recognises that programs operate inside an institutional environment. A technically sound health initiative can underperform where trust is weak; a well-funded education program can struggle where families experience institutions as distant or unresponsive.
Systems thinking also complicates the assumption that the leanest system is necessarily the strongest. Halo distinguishes efficiency from redundancy. Efficient systems minimise unused resources and spare capacity, while resilient systems often retain alternative pathways or backup capability that may appear underused during normal conditions. A hospital operating permanently near maximum capacity can look efficient until demand surges. A regional town with one doctor, one bank and one major employer may function perfectly well until one disappears. Community organisations, local journalism, neighbourhood networks and trusted intermediaries can play a similar role socially. Their value may be difficult to see while institutions are functioning normally, yet they provide alternative routes for information, coordination and support when pressure arrives.
Government structure introduces another complication because agencies are generally held accountable for what happens within their own portfolios. A hospital can reduce length of stay while increasing pressure on community care. A policing initiative can reduce visible offending in one location while displacing it elsewhere. A tighter eligibility rule can reduce spending in one program while increasing emergency accommodation, health demand or administrative appeals. The original program may have achieved exactly what its performance measure required while the wider system has barely improved. Systems thinking therefore changes the unit of analysis. It asks not only whether the intervention worked, but what happened next.
That question matters for peace because immediate security and long-term resilience can move in different directions. A policy can suppress visible conflict while leaving grievance unresolved, or increase compliance through coercion while reducing legitimacy. The purpose of systems thinking is not to make government hesitant. It is to identify which relationships are doing the most work: which feedback loop is sustaining deterioration, which incentive keeps restoring old behaviour, which apparent success simply moves pressure elsewhere, and which capacity has been removed in the name of efficiency but is essential for resilience.
When increasing expenditure, changing programs and reorganising services repeatedly produce the same broad outcome, the system may be telling us that the problem sits somewhere else.
