Each Tuesday, I offer a reflection on what it actually takes to get things delivered. Not just announced, but properly embedded in systems that work. These reflections sit alongside the Cases for a Practical State series because, together, they’re testing the same idea: that public delivery only works when it’s built on relationships, trust, and a decent understanding of the ground it’s meant to land on.
I call this series Te Rā Whakamana because that’s what implementation is: it’s the day of giving effect. It’s not the tail end of the policy cycle. It’s the moment a system earns legitimacy or loses it altogether. That matters in any country, but especially here, where authority doesn’t sit in one place. It moves. It’s negotiated. It’s shared. And it often begins in places the formal system still treats as optional.
This week I’m picking up the thread of expertise. Who gets to claim it? Who gets to ignore it? And what happens when it fractures? I'm drawing on a 2023 piece by Brian Head, who makes a sharp argument: the biggest shift in modern policymaking isn’t just politicisation or populism or even distrust: it’s that expert advice itself is now structurally contested (Head, 2023). It doesn’t flow neatly from the centre. It comes from multiple places at once: communities, consultants, lobbyists, industry, Crown entities, iwi and hapū, local government, advocates and activists, people with lived experience: that’s not dysfunction, that’s the shape of the policy advisory system now.
But you wouldn’t know it from the way many systems behave. Too many still act as if expertise can be centralised and controlled. That’s the legacy of neoliberal thinking, where technical knowledge sits at the centre and gets pushed out to the edges through market contracts, KPIs, and delivery chains. It’s tidy in theory. However, it often breaks down in practice.
The argument I want to make has three parts. First, we have to stop treating contestable expertise or non-state expertise as a problem to be solved. It’s not going away. It’s the new normal. Second, neoliberal governance models were never designed to handle that kind of pluralism or technology, for that matter: they’re built for control, not negotiation; they work better if low-risk and uncontested political issues are involved. Third, if we want delivery to work, we need systems designed to work with contestation, not deny it. That means co-production, co-governance, relational accountability: approaches that start from the reality that no single actor holds all the truth (Ansell, Sørensen, & Torfing, 2021).
The mismatch is most evident at the point of delivery. If advice is contested at the front end, between officials, communities, iwi and hapū, industry, media, and ministers, it will carry that fracture into implementation. You’ll see it in who gets funded. Who gets trusted. Who ends up carrying the blame when things don’t work. And who gets left to interpret vague commitments with no power to make them real.
In Aotearoa, this isn’t just a policy design challenge: it’s also a constitutional one. Te Tiriti affirms a model of plural governance. Rangatiratanga demands it. Our third-party decentralised contract-state obliges it. However, our policy advisory system still operates as if pluralism is a nice-to-have add-on, rather than a foundational condition. Neoliberal frameworks, with their obsession with control, efficiency, and uniformity, are structurally incapable of recognising this.
And that brings us to trust. Head is clear on this: trust in expertise isn’t just about rigour or credentials: it’s about legitimacy (Head, 2023), and I would add social license. Legitimacy and social license are not established on The Terrace through cabinet papers. Both are earned through the long, slow work of co-production, testing, and relationship-building. If people aren’t part of the thinking, if they’re not in the room when advice is formed, if their realities aren’t reflected, then implementation will fail before it starts.
That’s why technocratic systems, for all their modelling and logic, often collapse at the point of contact. They’re good at designing incentives. They’re terrible at handling disagreement, complexity, intractability and inexplicitness. They assume implementation is neutral when, in fact, it’s where values collide. And when things go wrong, policy advisory systems often resort to blaming, rather than learning. They treat delivery as a technical glitch rather than a reckoning for a dumb political decision.
The literature backs this. Craft, Head, and Howlett (2024) argue that the current open and contested political era demands a full rethink of how advice is formed, sourced, and legitimised. Emamian and Bagheripour (2024) show that even in highly centralised states like Iran, expert authority must be negotiated with political and social legitimacy to be effective. Hicks, Kingsley, and Isett (2025) demonstrate that technically sound advice on emerging technologies still fails if it bypasses stakeholder voices. Across all these cases, the same theme recurs: advisory systems collapse when they deny the diversity of those who hold or contest expertise.
And if it’s ignored, failure isn’t just technical; it’s also political. Communities withdraw. Trust dries up. And policy that looks good on paper gets quietly resisted or worked around in practice. Ministers don’t get the outcomes. And, often, they don’t even get the conversation: the phone is always off the hook.
In our context, implementation that honours Te Tiriti needs to do more than consult. It needs to embed. That means investing in capability at the edges. It means recognising that expertise resides in whakapapa, story, and practice —not just in spreadsheets. And it means a policy advisory system that treats pluralism not as a threat to be managed but as a condition to be respected.
That’s not how neoliberalism in public governance works. It was built to measure, not to relate. It flattens context. It assumes away power. It treats pluralism as inefficiency. It believes advice is neutral. But governance doesn’t work that way. Not here. Not now. And not in the complex systems we’re all trying to improve.
So the question I want to leave us with is this: is the system we’re using designed to work with disagreement, or to pretend it doesn’t exist? Because if it’s the latter, then enduring delivery failure isn’t a surprise. The surprise is that no one is talking about how it’s now baked in.
References
Ansell, C., Sørensen, E., & Torfing, J. (2021). When governance theory meets democratic theory: The potential contribution of co-creation to democratic governance. Perspectives on Public Management and Governance, 4(4), 346–362.
Craft, J., Head, B. W., & Howlett, M. (2024). Expertise, policy advice, and policy advisory systems in an open, participatory, and populist era: New challenges to research and practice. Australian Journal of Public Administration, 83(2), 233–256.
Emamian, S. M. S., & Bagheripour, R. (2024). The Iranian policy advisory system: Contained politicisation and emerging technicisation. Australian Journal of Public Administration, 83(2), 233–256.
Head, B. W. (2023). Reconsidering expertise for public policymaking: The challenges of contestability. Australian Journal of Public Administration, 83(2), 156–172.
Hicks, D., Kingsley, G., & Isett, K. R. (2025). Steering the future: Expert knowledge and stakeholder voices in autonomous vehicle policy reports. Policy and Society. Advance online publication.
Deborah, this is a fantastic piece. As someone who works with complexity, it is such a relief to read the work of someone within the system who can articulate the way that complex systems (aka living systems) actually behave in the context of government. And why a technocratic mechanistic model of top-down (or centre-out) control is destined for non-delivery. Our current neoliberal architecture forgets that expertise is dispersed and multifaceted, codified and embodied/lived, not monolithic and monocultural and static. It also forgets that expertise does not necessarily respect the epistemic silos it has constructed. And it especially forgets that the “system” is actually humans whose trust or non-trust in the technocratic logic create feedback loops that give rise to emergent behaviour. Complexity cannot be controlled. It can only be worked with skilfully. A system that requires uniformity of “input” is a system that is not designed for the complexity of reality. I am so enjoying your writing. I come at this from a very different starting point than you, having worked in a university, but my experience of leadership, policy, projects and systems change in a large (colonial) institution parallels what you are writing about. Your rigorous scholarship and systematic argument here is very much admired!
This takes me back 20 years (!) to when you and I put the 'flexible and responsive' regime in place at Youth Development. We needed that initial change to gain the freedom to have conversations with our providers and communities, and we still needed another four years to learn how to participate in them, and convince our partners they were real.
But one other thing I recall, relevant to this piece, was how we had to carefully manage contract numbers vs contract management staff, maintain a balance of meaningful vs transactional contracts, so we didn't look over-staffed for the workload. Because relational models take more resource but get punished under transactional governance frameworks.