As one generation has given way to the next, and as the nature of work has changed, a public sector once centred on professional authority has given way to one increasingly preoccupied with the creation of and adherence to rules. The consequences are visible in the aggregate: project delays and cost overruns, growing administrative overhead, and the mitigation of acute risks at the cost of chronic failure. These new systems were not built through considered strategy. They accumulated piecemeal in response to changing local incentives and are instead the product of strategy's absence over many years.
Any attempt to rebuild a high-performance professional culture needs to be clear-eyed about the nature of the problem, and be willing to accept a potentially disruptive transition as professionals who’ve developed within the current system become apprentices in a new one.
I began my career at Watercare, Auckland’s water and wastewater utility company. I joined a small team of chartered professional engineers. Decisions were made throughout the day around a small table: a senior engineer, another working towards professional chartership, and an old-timer who’d joined as an apprentice and knew the location and type of every valve in the city. If a pipe failed, the team would swivel their chairs, pull out the maps, and discuss the problem. Perhaps more information would be needed, and they'd send me out to take photos and investigate. Either way, decisions were being made and actioned constantly: what to do, how it would be done, which contractor to hire, and who would take responsibility for seeing it completed.
The system was efficient and highly personal, but far from laissez-faire. Many decisions could only be made by a chartered engineer, who staked her professional reputation on the balance of risks always attendant to any engineering decision. Some projects crossed various thresholds and required higher management approval, and every decision made was ultimately subject to conflict of interest policies, Official Information Act requests, and auditing.
Variations on this system were the norm everywhere until relatively recently, but have rapidly given way to ‘modern’ approaches that distribute authority much more widely across teams, such as project management, procurement, design, legal, contract management. Of course, any large undertaking has always required multiple teams working in parallel, but the difference today is that these teams work independently, with project authority increasingly distributed across those specialist functions whose members answer vertically to their own functional hierarchies rather than horizontally to a person responsible for the undertaking as a whole. Understandably, their goals are to see their own KPIs met, and the risks they are accountable for managed, whether or not the costs of doing so exceed the wider benefits of doing so.
The Case of Wellington Water
Some public organisations like Wellington Water took the ‘fragmented assurance’ approach to extremes, from its organisational and governance structures to convoluted operational delivery processes. Even simple pipe renewal projects (i.e. replacing old/broken underground pipes) that previously would have been the work of a morning to design and farm out to a contractor to deliver, became a six-month ordeal as the project passed through a series of process gates, from external consultant design reviews to pre-procurement checks.
The results make for depressing reading. After establishment via amalgamation in 2014, and while still using the legacy operating models of the institutions it was amalgamated from, the organisation managed to deliver 88% of the capital programme and 99% of the operating programme during its first year of existence. From here, performance steadily degraded. By 2021, Wellington Water was delivering just 14.9% of the estimated 100km of annual pipe renewals required to maintain the network at a steady state. Accordingly, leaks became increasingly serious, rising from less than 15% in 20141 to 19% in 20192, 31% in 2022, and peaking at 44% in 20243, meaning nearly half of all drinking water produced at the time was lost to leaks, requiring water restrictions to manage the risk of acute shortages.
Leaks were far from the only problem at Wellington Water. In 2021, chronic operational problems at the Gear Island and Te Mārua treatment plants led to the organisation discontinuing fluoridation at these sites. Murky accountability meant senior leadership wasn’t informed of the issue for nearly seven months, while the governance board (along with the shareholding councils, and the general public) had to wait another three before being told4. The incident forced an independent investigation, which found:
“lack of clear ownership and escalation criteria”
“a culture of safety, rather than effectiveness”
“siloed and disconnected ways of working”
“complexity of the [organisational] model makes service delivery challenging”
Unfortunately, the scope of the investigation was limited to the fluoride dosing issue, and to the extent that its recommendations were implemented, they failed to address the broader performance problems. Continued failures forced another independent investigation in 2024 after the organisation was found to have omitted $51 million of overhead corporate costs from its capital delivery programme advice to shareholding councils5. The terms of reference now explicitly referenced the board’s loss of trust in management, and sought a broader review of the organisation’s systems and processes. The investigators “were struck by how much our recommendations align with those of the 2022 Inquiry”, finding the organisation ‘downplayed problems’, had ‘unclear structures and accountabilities’, and was slow to respond. Recommendations included ‘increasing focus on outcomes [over process]’, and advocating that the organisation be restructured to create ‘clear role accountabilities and responsibilities for roles holders’.
Little was to change at Wellington Water, the report findings being philosophically opposed to the fragmented assurance management model that held sway. Tweaks were made, but the chronic performance issues remained, culminating in the release of 70 million litres of raw sewage into the sea off Wellington’s South Coast since February 2026.
“Abandoned beaches, public health warning signs and seagulls eating human waste are now features of the popular coastline” - Guardian UK Reporting
The Crown then became involved, finding systemic leadership failures and that “the event was not caused by a single decision or isolated failure, but by longstanding weaknesses in governance, accountability, asset management, risk management and infrastructure resilience”. Responsibility and authority had been split across too many teams, and organisations, leaving nobody with the incentive and ability to resolve chronic issues. Processes were followed, but problems accumulated with nobody addressing them until the system failed, failed again, and then failed three further times, before finally causing an international news story.
Following government-mandated ‘Local Water Done Well’ reforms, Wellington Water amalgamated into the new regional water-services organisation, Tiaki Wai, in July 2026. Tiaki Wai has a simplified operating model, bringing ownership and responsibilities that were previously split across multiple councils and organisations under one roof, though it remains to be seen whether the internal dysfunction resulting from a fundamentally broken management model will be resolved or simply reassert itself.
A Wider Pattern of Failure
Sadly, chronic failure resulting from the steady erosion of professional judgement, authority, and responsibility is evident across the public sector. Examples abound:
Oranga Tamariki (New Zealand’s Child Protection Service)
In 2020, following a series of high-profile failures, an investigation launched into Oranga Tamariki by the Ombudsman found that newborn ‘removal without notice’ had become routine rather than exceptional, and that many decisions were being “made without expert advice or independent scrutiny, and without wider whānau [family] involvement”. Superficially, this might seem cause for additional process, but a 2021 inquiry found the opposite: front-line staff were so overburdened by high caseloads, labourious process, and such a low organisational tolerance for risk that they held very little discretion in how cases were handled. Taken together, a pathological feedback loop had been created within the organisation where fear of failure led to the imposition of greater procedural controls. This in turn led to practitioners spending more of their time following process and filing paperwork, leaving less time and authority for staff to build understanding and wider family relationships. Delays became inevitable. Delays caused emergencies, and the emergency then triggered the highly coercive procedural response. The resulting scandal then created demand for yet more assurance and process to avoid recurrence, ultimately having the opposite effect. The inquiry report described an organisational emphasis on managing its own risks rather than meeting children’s and families’ needs and “trying to replace professional judgement with protocols, tools, and guidelines” [emphasis mine]. Recommendations included strengthening professional capability and authority, with oversight to be exercised by supervisors freed from front-line work of their own, rather than trusting in inflexible process.
Department of Corrections
Just after Christmas 2020, a riot broke out at Waikeria Prison which lasted six days and resulted in a fire which destroyed the facility known as ‘Top Jail’. The subsequent internal inquiry came to some familiar findings in 2022: the riot had no single cause, instead being the result of a cascading set of process failures. In particular, conditions in the prison were poorly managed, and front-line staff were given inadequate training, equipment, and authority to take timely action to the prisoner’s escalating behaviour.
The case is particularly interesting as the riot came after a series of earlier reports into Waikeria Prison in 2017 and 2019, which came to a similar set of findings and recommendations. These recommendations were accepted, yet were ‘not fully implemented’ by the time of the riot. The Chief Ombudsman, noticing this repeated failure to properly action recommendations across multiple prisons, launched an independent inquiry into the Department, finding that an elaborate machinery of managers, governance groups, reporting processes, and a centralised ‘action tracking database’ had developed for handling recommendations that had little or no relationship with ‘facts on the ground’. Actions were often marked complete on the basis that new processes had been developed, or would be developed, to manage the risk, regardless of whether they actually did so. Even when an action did reach a prison director, they’d often find the authority to deliver on it sat above them. Sometimes actions were assigned to positions that subsequently disappeared in restructures, while others had no named individual responsible.
The Department’s approach has been inconsistent and fragmented. The challenge would appear to be less in identifying a process to respond to recommendations, and more in embedding appropriate lines of accountability…
The Ombudsman’s report also found a management culture that responded to often intense media scrutiny by prioritising the avoidance of acute risks, being reactive “playing whack-a-mole”, and ensuring procedural compliance to neutralise potential individual blame, even when this drew attention away from addressing persistently poor prison conditions.
Corrections shows a variation on the general theme. Unlike Wellington Water, which tended to avoid accountability, Corrections instead created extensive process specifically intended to ensure accountability and learning. Yet that machinery turned recommendations into administratively completed actions without anyone being accountable for whether the underlying problem was actually resolved, and the result was ultimately the same—chronic failure becoming a major crisis.
School Buildings
In 2024 Erica Stanford, the new Minister of Education, responding to the concerning results of an affordability review begun under the previous government, commissioned an independent inquiry into school property delivery. The report revealed that, of the 488 projects in the pipeline, only 153 were fully funded, with a looming budget shortfall of $2.8 billion (compared to a budget of $3.7billion), resulting from ‘an internal structure that is complex, confusing, and over-engineered’.
“accountabilities and processes for school property planning and delivery were spread across multiple business groups, with inadequate integration or coordination… at the same time, many critical business processes are excessively centralised and there are limited delegations in place to enable frontline staff to do their jobs”
Highlighted in the report is the 'Marlborough Schools Co-location Project’. Beginning in early 2015, the Ministry prepared a business case to bring together three area schools into one campus. Cabinet approved the project that October with a budget of $63.5 million. Over the following eight years the project cost would grow to over $400m until being indefinitely paused in 2023, after $25m had been spent in consulting fees and other sunk costs. The inquiry found that, without a single individual or team held accountable for the project outcome or given the authority to make key decisions, each of the myriad teams involved were incentivised to see that their concerns were addressed, and that their procedures were followed, all of which added to the project complexity and scope.
“Delivery managers were routinely required to secure as many as nine approvals and more than 100 signatures (including approvals and endorsements) before a project [could] enter construction.”
Procurement was highlighted as an area of particular concern, where procedure and process were prioritised “at the expense of outcomes, efficiency, and value for money”. Many staff viewed that procurement rules hindered the proactive management of contractors and consultants, with little ability to account for prior performance during tender processes. With much of the asset management process outsourced to third party consultants, the inquiry found that staff had little ability to control consultants pushing projects toward greater scope or higher specification, which of course they had every incentive to do, adding to their fees.
The inquiry concluded that the property system was “highly centralised, low trust, risk averse, and bureaucratic system unable to deliver simple and common-sense solutions, unable to secure timely approvals, and therefore unable to provide efficient or effective service to schools”. While individual processes often made sense from an internal perspective, the system as a whole was ultimately “set up to fail”.
Generalities
These are not identical failures. Obviously a single explanation cannot account for everything that goes wrong (or right) across the sector. Budgets are sometimes inadequate for the task. Public organisations are often given conflicting goals or constraints that make achieving their goals impossible6. Nevertheless, what’s clear from these examples and others7 is the recurrence of a particular organisational problem. In each case, relevant knowledge, formal responsibility, and the ability to act became disconnected. These organisations could describe how a problem was supposed to be addressed, but could not reliably establish who would do so. New rules and processes were created, but these were targeted at symptomatic past failures, rather than improving future system outcomes. Systems that had previously relied on trained professionals empowered to act became mired in fragmented processes designed to manage every risk where nobody was ultimately responsible for the overall outcome.
Why Is This Happening Now?
In some ways, these dynamics are a natural consequence of the scale and complexity of modern societies combined with the immense variety of tasks we now expect our bureaucracies to perform—often without these being clearly prioritised (except perhaps in retrospect, after public outcry and inquiries). Von Mises’ critique still holds water: without an overarching optimisation metric or ability to easily assess performance, assurance processes and fragmentation are virtually inevitable. Nevertheless, we can identify a number of reasons why they seem to have become particularly salient recently:
First, the modern media environment creates unusually strong incentives to avoid identifiable acute failures. News cycles are short, and easy to understand scandals spread quickly on social media. Consequently, the potential of an identifiable mistake may be far more costly for an individual manager personally, or even an organisation, than a series of cost overruns on the same project, even if the latter costs an order of magnitude more. ‘Explaining is losing’. Trying to explain how a high-profile failure was actually the result of theoretically optimal risk management, doubly so.
Second, as I’ve discussed previously, many of the dynamics that underpin the old system are increasingly inapplicable in the modern world. Thinking back to the Watercare example, how often now could a team rely on the expertise and deep knowledge of an individual who’d spent their entire career working within that one system? Modernity—smaller families, greater mobility, the internet—makes moving jobs and towns easier than ever before in history. The result is shorter tenures within any one organisation, particularly among those mid-career in decision-making roles8.
There are many economic benefits to higher labour mobility. People find it easier to find roles they’re suited to, while transferring workers bring knowledge, ideas, and skills to their new organisation improving productivity. However, it does undermine not only the deep institutional knowledge that longer tenures can bring, but also, more subtly, the set of reputational incentives: both for an individual to prioritise longer-term outcomes over short-term concerns, and also for contractors and consultants, who must do the same to win further work from a long-standing procuring manager. Absent these reputational incentives, each contract becomes an opportunity to extract maximum value while remaining technically compliant, which institutions have responded to by strengthening procurement and contract management processes, often locking out suppliers without large compliance departments who may be happy to trade on their reputation.
Third, modern management philosophy and hiring practices must share a portion of the blame. The model currently in vogue assumes that work is far more modular and easily compartmentalised than it really is. By decomposing tasks previously exercised by a single role or integrated team across multiple functions and processes, we’ve inevitably missed critical functions (often unrecorded in any job description) that enabled the system to work effectively9.
The combination of these factors has unintentionally created a feedback loop, accelerating the transition to pathological organisations and poor outcomes:
Shorter tenures, outsourcing and interrupted succession
↓
Less embedded expertise and less confidence in discretion
↓
More gates, documentation, specialist functions and external assurance
↓
Less end-to-end authority, apprenticeship and feedback for internal professionals
↓
Internal roles become coordination and contract-administration roles
↓
Capable experts leave, fewer successors develop, and outsourcing becomes more necessary
No individual step requires anyone to intend the eventual result. A manager concerned about inconsistent decisions introduces another review process where projects must get third-party approval before proceeding. A specialist team is established to manage a risk that existing staff lack the time or knowledge to handle. Each seems like a reasonable response to an immediate problem, but if these changes reduce the opportunities for internal professionals to exercise judgment and learn from the results, the organisation becomes progressively less capable of operating and achieving its high-level objectives, with results like we’ve seen in the examples above.
Where do we go from here?
The public sector has enormous power over people’s lives, and often operates with a legally enforced monopoly. As such, it cannot and should not operate just like a private business. Transparency, consistent and lawful treatment, and protection against corruption are part of what public institutions must provide. The mechanisms that provide those safeguards can’t simply be dismissed as mere obstacles placed in the way of the real work, but the past shows these need not conflict with the fundamentals of high-performance:
A good procedure embodies professional judgment, rather than replacing it. Engineering standards can save engineers from having to solve already solved problems. A well-designed escalation threshold tells a junior colleague when they need help. An effective emergency plan translates specialist knowledge into actions that someone without that expertise can take immediately. Procedures can provide people with a basis for acting, but can’t substitute for responsible people (as we saw with Oranga Tamariki), nor action itself (Corrections).
Responsibility and authority must be delegated together. Giving somebody responsibility for an outcome is meaningless if they cannot make the decisions necessary to achieve it. This includes access to appropriate resources, the ability to change an approach that is not working, and the ability to elevate decisions outside their delegation to a higher level of responsibility. This requires a structure of nested responsibility with a single individual responsible for every bounded operational undertaking. Any other approach is unworkable.
Expertise must be embedded. External consultants can be extremely useful, but a team must have sufficient formal professional knowledge, practical familiarity with the system, and understanding of adjacent work to be able to determine requirements, manage context, and know when outside assistance is needed. The advent of reasoning AI only makes this more critical. Even advanced models capable of solving Erdős problems and helping cure cancer will produce only superficially helpful outputs if their input contexts aren’t properly understood and communicated by their users.
Institutions must rebuild and sustain their capability. Professional capability can’t be built solely through recruitment, training courses or revised job descriptions. Junior staff need to be making progressively more difficult decisions of their own, with experienced colleagues available to supervise and challenge them. They also need to stay connected to the consequences of their actions long enough for it to form useful feedback. A team cannot build a sense of shared responsibility and accountability to each other overnight, and an institution working to rebuild its capability must be willing to accept some small-scale failures which may previously have been avoided by assurance as a natural part of learning.
Governance bodies must accept the risk trade-offs, and hold senior leaders accountable to outcomes rather than task completion. Ultimately, governance bodies must decide whether to continue requiring increasingly elaborate assurance and risk-management processes while tolerating declining performance, or to hold management accountable for outcomes, appoint capable senior leaders and give them the authority and backing needed to deliver them. Without such expectations and support, any changes made by management will not last.
“Couple weeks from now, you're gonna be in some district somewhere with 11 or 12 uniforms looking to you for everything… All of them will take their cue from you. You show loyalty, they learn loyalty. You show them it's about the work, it'll be about the work. You show them some other kinda game, then that's the game they'll play.” -Cedric Daniels, The Wire
The figures presented here actually understate the worsening trend, as the 2014 water loss figure includes firefighting, as well as Council use in parks, swimming pools and so on, whereas the later figures are estimates of leaks only.
Here, the 19% reported leak figure is actually just the midpoint of a 6-31% range with 95% confidence, a staggeringly wide interval demonstrating how the organisation’s understanding of its own assets had become.
Even then, the report to the board lacked context and implied the fluoridation had only ceased recently. It wasn’t until the independent inquiry was underway that the board learned how they’d been kept in the dark.
As with the fluoridation failure, the overhead accounting error was known for months by staff, but went unreported during that time thanks to a lack of clear responsibility—no one individual was responsible for the error or was accountable for reporting it.
As one example, even with the most efficient organisational arrangements, public housing construction initiatives will continue to fail if planning regulations make the construction of additional dwellings all but impossible, as I’ve discussed with Taiwan.
In addition to the numerous examples sent by readers, other high-profile examples include:
-The Novopay debacle, where the inquiry found a key failure was ‘the lack of a programme director with overall accountability’
-EQC’s failures to process Canterbury Earthquake claims effectively as a result of moving from a case manager system with one individual responsible to a fragmented assurance scheme which duplicated processes and overmanaged risk (EQC moved back to case management following a public inquiry).
-The 2015 Productivity Commission’s findings that social services had been ‘hampered by unclear responsibilities and fragmented accountabilities”.
I use US data here as the Bureau of Labour Statistics provides by far the most comprehensive data for long-term trends out of any OECD nation—Statistics New Zealand only began collecting tenure data from 2016. I’ve limited the chart to men to enable a cleaner comparison over time, since the continued entry of women into the workforce over the earlier portions of the time series otherwise create the impression that individual tenure for those mid-career was considerably lower at that time than it was in reality.
There’s an interesting parallel here with predictions about AI and the future of work that illustrates how pervasive this sort of understanding has become. There’s a famous prediction from ‘the Godfather of AI’ Geoffrey Hinton in 2016 that demand for radiologists would collapse as “it is just completely obvious that within 5 years deep learning will do better than radiologists”. In fact, as Nvidia CEO Jensen Huang has become a fan of pointing out, radiologists are now in higher demand than ever. Hinton’s prediction was wrong, but only because his understanding of a radiologist’s work was inadequate. AI is now superhuman at reading scans and other image-recognition tasks, but the prediction missed the work bundled into the occupation of “radiologist” which is not captured by those tasks: integrating clinical context, deciding which questions matter, communicating with clinicians and patients, and resolving ambiguities.
It’s the same story with software engineers, despite AI capabilities having advanced more here than in any other field. Reality has a surprising amount of detail, and lowering the cost of producing code has only made all the other tasks of an engineer—like working with customers to understand requirements, managing context, and exercising taste—all the more valuable. Any organisation that takes an overly mechanistic view of its people and its systems will pay the price for it.