From Australia, the images are difficult to comprehend. Burning streets. Smashed shopfronts. Police retreating under a barrage of projectiles. People cheering as their own cities suffer damage.
The immediate questions are understandable. Who are these people? What do they hope to achieve? How does destroying a school improve education?
But another question deserves attention: what happened before the first window broke?
By the time the evening news shows flames above a barricade, much of the opportunity to prevent disorder has passed. The critical decisions concern preparation, police deployment, communication with organisers and intervention when violence first appears.
Arresting offenders afterwards is necessary. Preventing the destruction is the more demanding test of government.
France’s October 17 demonstrations exposed serious weaknesses. Le Monde reports that intelligence services struggled to estimate how online mobilisation would translate into attendance. Police units became overstretched as confrontations developed at scattered locations. The interior minister denied underestimating the operation, blaming dispersed groups of rioters and looters. That explanation leaves an obvious question: why was the operation unable to cope with them? Le Monde
A country with France’s experience of public disorder should prepare for more than an orderly march along an agreed route. Contingency planning must cover larger crowds, simultaneous confrontations and attacks away from the main demonstration.
Officers need sufficient numbers, trained support and reserves available before a local confrontation becomes an invitation to wider disorder.
Police weakness can change a crowd’s calculations. Research into England’s 2011 riots found that perceptions of police vulnerability emboldened participants. Shared hostility towards police also helped people identify with one another and join the confrontation. Control and legitimacy both mattered. University of Sussex
That makes the first interventions especially important.
A competent operation must distinguish peaceful demonstrators from those throwing projectiles, attacking people or destroying property. It must protect the former while acting against the latter. Broad interventions that sweep peaceful participants into confrontation risk enlarging the hostile crowd. Allowing violent groups to act unchecked creates opportunities for others to join them.
The objective is clear: stop violence while retaining the cooperation of everyone willing to remain peaceful.
But policing is only part of the story.
Order also depends on people feeling that the streets, schools and businesses around them belong to a society of which they are part. A shop window represents somebody’s livelihood. A school serves somebody’s children. Public property is something neighbours must pay to replace.
The Nordic countries offer a useful comparison. OECD adult-skills survey results show much higher interpersonal trust in Denmark, Norway and Finland than in France. Sweden also employs specialist dialogue police who work with demonstration organisers before events. These are practices and social conditions worth studying. They do not provide complete immunity: Sweden experienced serious riots over Easter 2022. They do, however, direct attention towards the relationships on which public order rests.
A durable sense of belonging requires mutual obligations. Citizens must accept that grievance does not confer a licence to destroy. Institutions must demonstrate that peaceful participation receives a fair hearing. Police must enforce the law consistently and account for their own conduct.
Repeated disorder makes those obligations more urgent. Each episode should produce identifiable changes in preparation, command and practice. Otherwise, the next declaration that authorities were overwhelmed becomes harder to accept.
France needs a credible examination of what failed: attendance estimates, deployment, reserves, communication and the decisions that preceded escalation. Responsibility for arson, assault and looting remains with the offenders. Responsibility for an effective public-order operation rests with those who planned and commanded it.
No police force can prevent every act of violence. Governments can still be judged on whether they anticipated foreseeable dangers and contained them effectively.
For Australia, the lesson reaches beyond police equipment and officer numbers. We should examine the strength of our own shared belonging, and whether our institutions earn the cooperation they depend upon.
The burning barricade is the visible failure.
The work of preventing it begins much earlier—with capable policing, credible institutions and citizens who still see their neighbours’ losses as their own.
A business does not need artificial intelligence to replace every employee. It only needs AI to make fewer employees necessary.
That is the employment threat we need to take seriously.
As AI improves and becomes embedded in everyday business systems, organisations will use it to answer enquiries, process accounts, prepare documents, schedule work and manage routine administration. Where fewer people can deliver the same output, reducing labour costs will be an obvious attraction.
Some businesses will expand and redeploy employees. Others will cut staff. The savings can arrive quickly. Replacement livelihoods may take years.
Agriculture offers a warning.
Over roughly 150 years, mechanisation and economic change helped turn farming from a major source of employment into an occupation employing only a small percentage of workers in developed economies. In the United States, agriculture’s share of employment fell from 13 per cent in 1948 to 2 per cent in 2017 alone. USDA
Society had generations to adjust. Even then, the transition brought upheaval.
What happens if a comparable contraction in demand for office workers unfolds in ten years?
This is a scenario, not a prediction. But it is serious enough to demand a plan.
Banks, insurers, call centres, government departments and large corporate offices employ substantial numbers of people whose work involves processing information. If these employers reduce staffing together, displaced workers will find fewer opportunities to move into similar jobs elsewhere.
An organisation may solve its cost problem while society inherits an employment problem.
“Retraining” is the usual answer. Retraining for what?
Training takes time, costs money and needs to lead to an actual vacancy. A displaced administrator cannot simply become a specialist engineer. Nor can everyone move into management: there are fewer positions, and managerial work is also exposed to AI.
“New jobs will emerge” is another answer. They may—but not necessarily in the right places, requiring the right skills, or soon enough to pay next month’s mortgage.
Service industries may absorb some workers. But care, hospitality, retail and transport do not offer unlimited vacancies. These jobs have their own skill requirements, and an influx of applicants could increase competition for hours and put pressure on wages.
The damage may extend well beyond unemployment. People may accept lower incomes, insecure contracts or work that uses little of their experience. Young people may struggle to get started as the routine tasks through which beginners learn are automated.
We should be preparing several responses now.
Retraining and redeployment can help, but only when tied to realistic employment opportunities. A course completion certificate is not a livelihood.
Shorter working weeks could spread available work, but preserving incomes requires productivity gains to be shared. Employers and workers will not automatically agree on how.
Investment in care, infrastructure and other unmet needs could create useful employment. It requires funding, planning and time, and the jobs may not suit everyone displaced.
Stronger income support could protect families through the transition. It also needs a sustainable funding base, and financial support alone cannot replace the purpose and social connection people find in work.
There is no single, painless answer. That is precisely why planning cannot wait.
The ILO identifies clerical occupations as the most exposed to generative AI. Exposure does not tell us how many jobs will disappear, but it tells us where to pay attention. ILO
We should watch falling recruitment, shrinking graduate intakes and employees who leave without being replaced. By the time mass redundancies dominate the headlines, the adjustment may already be well underway.
AI could deliver extraordinary prosperity. It could also displace workers faster than our institutions can respond.
A transition that was manageable over generations could become a crisis if compressed into a decade.
Developed societies face the risk of an employment tsunami. We do not need certainty about its arrival to start preparing.
It is time to wake up.
The World Bank has a problem.
Its president, Ajay Banga, recently revealed that the Bank is discussing possible emergency financial assistance with between 30 and 40 countries.
The immediate causes are familiar: rising energy and fertiliser prices, economic disruption and increasingly expensive borrowing.
But behind these emergencies lies a much older problem.
According to the World Bank's latest International Debt Report, developing countries paid $741 billion more in principal and interest on external debt between 2022 and 2024 than they received in new financing.
Think about that.
Countries desperately needing capital to develop their economies are collectively sending hundreds of billions of dollars more to creditors than they receive in new loans.
The system designed to finance development has, for many borrowers, become an enormous drain on their resources.
And the proposed solution?
More financial assistance.
There is nothing inherently wrong with borrowing.
A developing country might borrow to build a port, improve its electricity network or establish a reliable water supply. If the investment generates economic growth, jobs and additional income, the country becomes better able to repay its debt.
That is productive borrowing.
But suppose the money finances government consumption, an uneconomic prestige project or infrastructure that never delivers the expected return.
The debt remains. The additional income does not materialise. Eventually another loan is required, or the original debt must be restructured.
The country is no closer to self-reliance.
Of course, education, public health and sanitation may deliver enormous long-term benefits without generating immediate financial returns.
But the principle remains: borrowing should improve a country's future capacity to prosper, not merely finance its present expenditure.
When an international institution lends hundreds of millions of dollars to a developing country, how confidently can it establish where that money ultimately goes?
The World Bank's 2025 report on debt transparency provides a disturbing answer.
Although more than 75% of low-income countries publish some debt information, only 25% disclose loan-level information about newly contracted debt.
The Bank also warns about hidden liabilities, off-budget borrowing and increasingly complex financing arrangements.
This does not mean that most development loans are stolen or wasted. But it does mean that citizens, lenders and international institutions may lack the information needed to judge how effectively governments are managing their obligations.
Governments borrow. Projects are announced. International institutions report the money committed or disbursed.
But who returns five or ten years later to establish whether the investment actually delivered the promised improvement in living standards?
Here lies a deeper problem, one regular readers will recognise.
The incentives of those making decisions are not necessarily aligned with the interests of the people they are supposed to help.
A government may benefit politically from spending borrowed money today, leaving future taxpayers to repay it.
A lender may be rewarded for expanding its loan portfolio, even when the long-term economic benefits are uncertain.
A development agency may measure success partly by the money it distributes and the projects it supports, rather than by how quickly recipients become independent of assistance.
And when a country gets into trouble, creditors may prefer restructuring the debt to acknowledging that the original lending decision was unsound.
None of this requires corruption or bad intentions.
It simply requires people responding rationally to the incentives confronting them.
Unfortunately, those ultimately paying the price are often the people least responsible for the decisions.
What makes a poor country prosperous?
Foreign assistance can help. But sustained prosperity requires something more fundamental.
Businesses must be able to invest, employ people, trade and retain enough of their earnings to make further investment worthwhile.
Governments need functioning courts, enforceable contracts, reliable infrastructure and predictable rules. Entrepreneurs need opportunities rather than political connections.
Without these foundations, even substantial foreign investment may deliver disappointing results.
South Korea, Singapore and Vietnam followed very different political and economic paths, but each demonstrates that long-term growth ultimately depends on productive capacity, not permanent foreign assistance.
Money can finance development. It cannot substitute for the conditions that make development possible.
Certainly not.
Humanitarian assistance is essential when people face famine, disease or disaster. Development grants can finance vital services, while carefully designed loans can help build productive infrastructure.
But these activities should be judged differently.
Humanitarian assistance should be measured by the lives it saves.
Infrastructure loans should be assessed by the economic and social benefits they deliver.
Development programmes should ultimately be judged by whether they reduce the need for further assistance.
Debt relief may be necessary when repayments become unsustainable. But forgiving debts without addressing the conditions that produced them risks restarting the same cycle.
And creditors must accept responsibility for imprudent lending. Financial discipline cannot be demanded only of borrowers.
Imagine an international development institution that measured success not primarily by how much money it distributed, but by how many countries no longer needed its assistance.
It would insist upon transparent borrowing, independently assessed projects and measurable long-term results.
It would help countries attract productive private investment, improve institutions and participate in international trade.
And it would be prepared to acknowledge when its programmes had failed.
The objective would not be to create better-managed debtors.
It would be to create prosperous, self-reliant countries.
The World Bank and other international institutions have helped finance genuine improvements in health, education and infrastructure. Those achievements should not be dismissed.
Nor can every debt crisis be blamed on irresponsible spending. Wars, pandemics, natural disasters and global interest-rate shocks can overwhelm even well-managed economies.
But after decades of development lending, why do so many countries remain dependent upon further borrowing and assistance?
Perhaps we have concentrated too heavily on how much money the world provides, and too little on whether that money produces lasting independence.
The ultimate measure of successful foreign assistance should not be how much we give, but how little assistance the recipient eventually needs.
This article continues a theme explored in two earlier Grappy's Soapbox posts: how well-intentioned policies can produce unintended and sometimes harmful consequences when the incentives of decision-makers differ from the interests of those they are meant to serve. See Perverse Incentives: How Good Intentions Can Produce Bad Outcomes and Perverse Incentives — Part 2.
When those making the decisions don't bear the consequences, whose interests are really being served?
Some time ago, I wrote about perverse incentives — situations in which individuals or organisations, acting quite rationally in their own interests, produce outcomes that are contrary to the interests of society as a whole.
It is one of the most important, and least appreciated, explanations for why institutions so often fail to deliver what they were created to achieve.
A recent article in TrialSite News, Was the COVID-19 Pandemic a Trojan Horse?, by Dr Ronald N. Kostoff, provides a fascinating case study.
The article asks a provocative question: did the extraordinary measures introduced during the pandemic serve purposes beyond protecting public health?
My interest is in a somewhat different question.
What happens when the people making decisions benefit from those decisions, while the costs are borne by someone else?
Consider a public health official confronting a new and potentially dangerous virus.
If the official recommends severe restrictions and those restrictions later prove excessive, the consequences are largely borne by businesses, workers, schoolchildren and families.
But if the official recommends restraint and the disease spreads disastrously, the professional and political consequences could be devastating.
Which course is the safer one for the official?
The answer is obvious. And therein lies the problem.
The decision that minimises personal or institutional risk is not necessarily the decision that produces the best outcome for society.
This does not require dishonesty, corruption or even bad intentions. It requires only ordinary human behaviour responding to a particular set of incentives.
During COVID-19, this asymmetry was particularly pronounced. The consequences of doing too little were immediate and visible. Many of the consequences of doing too much — educational losses, business failures, delayed medical treatment, psychological distress and social isolation — emerged more slowly.
The incentives favoured caution, even when caution imposed enormous costs on others.
Kostoff's article examines the extraordinary redistribution of economic activity, institutional power and financial resources during the pandemic.
Consider some of the outcomes.
Large businesses versus small businesses. While many small shops and restaurants were forced to close, large retailers and online businesses were often able to continue operating. Some emerged with stronger market positions, while smaller competitors disappeared.
Pharmaceutical companies. Vaccine manufacturers benefited from substantial government funding, enormous purchasing commitments, accelerated regulatory processes and significant liability protections. Some of these arrangements may have been justified by the emergency, but they also created powerful commercial incentives.
Technology companies. Lockdowns accelerated the shift to online shopping, remote working, digital education and electronic communication. Companies already dominating these markets benefited enormously.
Government agencies. Emergency powers allowed executives and administrators to make decisions that would ordinarily have required extensive consultation, parliamentary scrutiny or judicial review. Once acquired, such powers are not always surrendered enthusiastically.
Public spending. Trillions of dollars were distributed through emergency programmes, often with reduced scrutiny. The United States Government Accountability Office subsequently documented extensive fraud and weaknesses in oversight.
None of these outcomes proves that the beneficiaries engineered the policies that benefited them.
But that is not the point.
Once a policy creates winners and losers, the winners acquire a powerful incentive to preserve it — even when its original justification has weakened.
Kostoff goes further than this.
In his concluding remarks, he advances the hypothesis that the pandemic was used deliberately to bring about political change in the United States, with economic disruption and changes to voting procedures serving that objective.
That is a serious allegation, and one that requires considerably stronger evidence than the identification of those who benefited.
Indeed, an important part of the analysis presented in the article makes precisely this distinction.
It separates four questions:
What actions were taken?
What consequences followed?
Who benefited?
Is there evidence that those benefits motivated the original decisions?
The first three questions can often be answered with reasonable confidence. The fourth is considerably more difficult.
And here is where I think the concept of perverse incentives offers a more compelling explanation.
We do not need to believe that someone planned the entire sequence of events.
A genuine emergency can create extraordinary opportunities. Individuals, businesses, bureaucracies and political institutions then respond to those opportunities in ways that advance their own interests.
The resulting pattern may look remarkably coordinated, even when no central coordination exists.
It is the cumulative consequence of many people pursuing their own objectives.
In a competitive marketplace, poor decisions generally carry consequences for those making them.
A business that repeatedly misjudges its customers risks losing those customers. An investor who makes poor investments loses money.
But government and large institutions frequently operate under a different set of rules.
The people making decisions are often insulated from their consequences.
A bureaucrat who imposes an unnecessary restriction does not personally compensate the business owner who loses everything.
A policymaker whose decisions damage children's education does not necessarily bear the long-term costs.
An institution that expands its authority during an emergency may have every reason to defend that expansion long after the emergency has passed.
This is not an argument against government, public health or collective action.
It is an argument for recognising that good intentions do not eliminate bad incentives.
Indeed, the more noble the declared objective, the easier it can become to overlook the incentives operating beneath it.
The pandemic presented governments with genuine and difficult choices, particularly when knowledge of the virus was limited. Some interventions saved lives; others imposed substantial collateral costs. Their effectiveness and proportionality varied by policy, place and time.
But whatever one's assessment of the individual measures, the experience should have taught us something about institutional behaviour.
Emergency powers require independent scrutiny.
Temporary measures need clear conditions for their termination.
The people advising governments should be required to consider not merely the benefits of their recommendations but also their economic, social and human costs.
And where possible, the consequences of decisions should be visible to — and influence the accountability of — those making them.
Most importantly, we should resist the temptation to judge a policy solely by the intentions of its proponents.
The relevant question is not simply, "What are they trying to achieve?"
It is also, "What incentives are driving their behaviour, and who bears the cost if they get it wrong?"
Regular readers of Grappy's Soapbox will recognise this as a theme I have explored before.
In Perverse Incentives — How Good Intentions Go Bad and its follow-up, Perverse Incentives — Part 2, I explored how individuals and institutions, acting quite rationally in pursuit of their own interests, can produce outcomes that are contrary to the interests of society as a whole.
The COVID experience provides another, particularly dramatic illustration of the same underlying problem.
The essential question is not whether the decision-makers believed they were doing the right thing. Many undoubtedly did.
It is whether the incentives governing their decisions encouraged outcomes that were also right for the wider community.
When personal or institutional success becomes disconnected from the public good, even well-intentioned decisions can lead us badly astray.
Perverse incentives are not confined to pandemics.
They can be found in climate and energy policy, financial regulation, education, welfare, corporate management and virtually every field in which decision-makers are rewarded for achieving narrow objectives while the broader costs fall on others.
They explain why bureaucracies expand, why temporary programmes become permanent, why regulations sometimes protect established businesses against competition, and why policies that demonstrably fail can nevertheless survive.
We often look for villains when institutions produce bad outcomes.
Sometimes villains exist. Corruption and deliberate exploitation are real.
But often the explanation is more mundane — and therefore more troubling.
Ordinary people, pursuing entirely understandable personal and institutional interests, can collectively produce extraordinary harm.
The greatest danger may not be a conspiracy of bad people. It may be a system in which perfectly ordinary people are rewarded for making decisions that are bad for everyone else.
And unless we understand that, we are destined to repeat the same mistakes.
For readers interested in the detailed analysis, Dr Ronald N. Kostoff's article, “Was the COVID-19 Pandemic a Trojan Horse?”, published by TrialSite News on 6 October 2026, examines the economic, political and institutional consequences of the pandemic response. It is an opinion article, and some of its more far-reaching hypotheses should be distinguished from its documented observations.
