The Free and Frank Series: The Relational Substrate

21/02/2026

This is the fifteenth instalment in a series exploring what free and frank advice means in Aotearoa’s constitutional arrangements. Over the past three posts, I have tested the advisory relationship against three traditions of political theory: consent, public reasoning, and non-domination. Each illuminated a different dimension of the space within which free and frank advice must operate, but none could fully account for the relational substrate beneath the institutional arrangements. This week, I turn to the fourth and final tradition: trust. By the end of this post, I hope to have shown that trust in the advisory relationship is not a soft addendum to the harder questions of authority and accountability, but the condition on which all of them depend, and that the reforms of the 1980s may have done more damage to that substrate than any amount of performance management could repair.

Three traditions ground the argument. Annette Baier (1986)’s (1986) work on the vulnerability inherent in trust relationships. Onora O’Neill’s (2002) account of trustworthiness as the proper object of institutional design. And Niklas Luhmann (1979)’s (1979) systems-theoretic distinction between trust in persons and confidence in institutions. I trace what each reveals about the advisory relationship, and why the accountability mechanisms designed to increase trust may paradoxically destroy the conditions that make it possible.

There is a small but telling detail in how the breakdown of free and frank advice tends to get described by those who have lived through it. Ministers do not usually say that officials failed to provide sufficient evidence, or that institutional arrangements were poorly designed, or that role boundaries became unclear. They say something closer to this: “I stopped believing they were being straight with me.” And officials, describing the same deterioration from their side, rarely say that political authority became arbitrary, or that the bargain was renegotiated unfavourably, or that the authorising environment grew too complex to navigate. They say something closer to this: “I no longer told him what the risks were. He did not want to hear it.”

Both formulations describe broken trust. And yet the three traditions we have examined so far (consent, public reasoning, and non-domination) each addressed the advisory relationship without fully accounting for its relational character. They mapped the authority, the reasoning, and the institutional constraints. What they could not reach was the substrate beneath all of those: the conditions that make free and frank advice possible in the first place. So in what follows, I want to argue first that trust is something more demanding than we usually assume, then that we have been asking the wrong question about it, and then that the reforms of the 1980s destroyed something specific and important: so that by the end, we can see why trust is not a supplement to the advisory relationship but the condition on which it rests.

What Trust Actually Is

The most searching philosophical account of trust in the contemporary literature belongs to Annette Baier, whose 1986 essay Trust and Antitrust reoriented the field by insisting that trust is not simply a cognitive judgement about probability, and not simply a contractual expectation about performance. Trust, on Baier’s account, is something more exposed and more morally significant than either of those formulations suggests. It is a reliance on another’s goodwill. And because goodwill cannot be contractually guaranteed or institutionally mandated, trust always involves vulnerability: the deliberate acceptance of a position in which the trusted party could harm the trusting party, and chooses not to.

This is worth sitting with, because it reframes what we are actually asking of officials and ministers in the advisory relationship. To trust someone to offer free and frank advice is not simply to expect that they will produce accurate analysis, though accuracy matters. It is to accept a position of exposure: to acknowledge that the trusted party knows things you do not know, sees risks you cannot see from where you stand, and could, if they chose, use that asymmetry against you. The minister who genuinely invites frank counsel is accepting vulnerability. So, in a different way, is the official who provides it, for frank advice that contradicts ministerial preference exposes its author to precisely the career consequences that the classical Westminster bargain was designed to prevent.

Baier’s framework suggests why this mutual vulnerability is not incidental to the advisory relationship but constitutive of it. The official who never says anything the minister does not wish to hear is not offering trust and not inviting it; they are offering a service, and a rather limited one. The official who speaks frankly, who accepts the exposure that frankness entails, is doing something categorically different: they are extending trust in Baier’s full sense, relying on the minister’s goodwill not to punish candour, and in so doing creating the conditions under which the minister can extend reciprocal trust in return.

This mutual vulnerability also explains something about why advisory relationships are so difficult to reconstruct once they have broken down. Trust that has been betrayed, for example, freeness and frankness that was punished, and candour that was weaponised, does not simply diminish trust, it poisons the well for the interactions that follow, because the very act of trusting again would require accepting a vulnerability that has already been exploited. Both parties know this. And so advice becomes more cautious, more hedged, more attentive to what will be well received. The advisory relationship continues; its essential Westminster character has quietly departed.

The Wrong Question

If Baier identifies what trust is and why it matters, Onora O’Neill identifies what we persistently get wrong when we try to repair it. Her argument, developed in her 2002 Reith Lectures, is straightforward: we keep asking whether people trust one another, when the more useful question is whether they are trustworthy. Trust extended to someone untrustworthy is not a virtue: it is in O’Neill’s reading a mistake. What institutions should be building is not trust itself but the conditions under which trustworthiness can be demonstrated, assessed, verified and re-verified.

That word, assessed, is where the argument bites. For O’Neill, trustworthiness is not simply a briefing paper, an intention or a disposition. It is something that others must be able to observe and evaluate, and evaluate again. Crucially, in O’Neill’s reading, it is not established in a single moment: it is built through the continuous, daily exchange of a working relationship, tested and confirmed or eroded over time. A minister learns whether an official can be trusted to be straight with them not from one briefing but from the accumulation of many. An official learns whether a minister can be trusted to receive difficult advice not from one conversation but from watching how that minister responds across a range of situations.

Applied to free and frank advice, this creates a specific and largely unacknowledged problem. The OIA’s protection of free and frank advice as a withholding ground means that this entire dynamic is hidden from outside scrutiny. Other ministers and other officials cannot see it. Parliament cannot see it. The public cannot assess it. What the Act protects as the condition for candour is also the condition that makes trustworthiness impossible to verify from the outside. We, the public, are asked to trust that the advisory relationship, in its most heated moment, is functioning as it should, without the means to observe the pattern of exchanges through which trustworthiness is either built or quietly abandoned.

The Public Service Act 2020’s prescription that officials should act in a free and frank manner is, at least, a behavioural prescription: a tentative movement toward O’Neill’s position. But prescribing a manner that remains protected from scrutiny does not solve the demonstrability problem. It simply restates the aspiration.

What the Reforms Destroyed

The sociologist Niklas Luhmann offers a third perspective that illuminates something important about the specific damage that the New Public Management reforms inflicted on advisory relationships in Aotearoa and across the Westminster world more broadly.

Luhmann’s account of trust distinguishes between personal trust, which is the trust that develops between individuals through repeated interaction and tested experience, and systems trust, which is the confidence that institutional arrangements will function reliably even when personal relationships are absent, underdeveloped, or strained.

Systems trust is what allows a new minister to accept a briefing from a chief executive they have never met before, on the reasonable expectation that the chief executive is operating within a framework of professional obligations that constrains their capacity to deceive. It is what allows officials to provide free and frank advice to a minister they have known for three weeks, because the institutional context creates shared expectations about how such advice will be received.

Personal trust is precious but fragile. It develops slowly, cannot easily be transferred, and does not survive significant betrayal. Systems trust is more resilient, more transferable, and more important for the functioning of complex institutions where personal relationships cannot always be present or adequate.

What the reforms of the 1980s and 1990s dismantled, in a Luhmannian reading, was not primarily personal trust, which is harder to destroy through institutional redesign alone. What they dismantled was systems trust: the confidence that institutional arrangements would protect officials who offered uncomfortable counsel, that professional standards would be maintained across the churn of fixed-term appointments, that the implicit bargain would be honoured even when the individuals holding ministerial and departmental positions had changed.

As Donald Savoie (2003) documented in Breaking the Bargain, the reforms did not replace this systems trust with alternative institutional arrangements that could perform the same function. They replaced it with contractual accountability that created incentives pointing in precisely the opposite direction.

The result, Savoie and others have argued is a policy advisory system that placed an unsustainable burden on personal trust to do work that could only ever be done, at scale and across a whole system, by systems trust.

Ministers and officials who happened to develop strong personal relationships could still achieve something approaching genuine frankness. Those who did not, or who had been assigned to each other by circumstance rather than by choice, were operating in a vacuum that no amount of guidance from the Department of Prime Minister and Cabinet or the Public Service Commission could fill.

Richard Norman (2003)’s Obedient Servants? describes the practical consequences of this vacuum with characteristically precise unease. When career security depends on ministerial satisfaction, officials need not be instructed to trim their advice. The system teaches the lesson through the accumulated experience of watching which judgements prove welcome and which disappear without trace. The damage in Norman’s reading is not necessarily visible in any single advisory interaction. It accumulates invisibly, in the growing distance between what officials think and what they write, until the gap between analysis and advice becomes too wide to bridge.

The Limits of the Frame

Trust theory carries its own limitations, and intellectual honesty requires acknowledging them.

The first is that it can individualise what are structural problems. If free and frank advice is failing because institutional arrangements make trustworthy behaviour professionally hazardous, then diagnosing the problem as a deficit of trust places responsibility on individual practitioners for failures that belong to the system. O’Neill’s reframe to trustworthiness helps, because it directs attention to conditions and behaviour rather than to attitudes. But even that reframe can shade into moral exhortation: asking individuals to absorb the costs of structural dysfunction rather than changing the structures that produce it.

The second limitation cuts deeper. Trust can be weaponised as a substitute for accountability rather than a complement to it. The claim that free and frank advice requires confidentiality and therefore protection from scrutiny is entirely correct as far as it goes. But as Nicola White’s scholarship on the Official Information Act documents, the official who invokes free and frank advice as a withholding ground is not always protecting their candour. They are sometimes protecting themselves and their agency. The difference is not always visible from the outside, which is precisely the problem. Invoking trust as a shield can make it harder, not easier, to assess whether the people in advisory roles are being trustworthy at all.

The Substrate, Not the Solution

So where does this leave us? I think what Baier, O’Neill, and Luhmann together suggest is this: trust is not what free and frank advice produces. It is the condition within which free and frank advice becomes possible. Remove it, and the advisory relationship does not disappear, for the briefings still arrive, the meetings still happen, the words are still carefully chosen. But something essential has gone. What remains is advice shaped by what will be well received rather than by what needs to be said, and the distance between those two things is exactly the distance between a functioning advisory relationship and one that has quietly failed.

That logic, I think, also explains why the definitional problem we have been circling throughout this series cannot be resolved by trust theory alone, any more than it could be resolved by consent, or public reasoning, or non-domination. Each tradition illuminates a different dimension of the space within which free and frank advice must operate. Together they show that the question what is free and frank advice, has political, epistemic, institutional, and relational dimensions simultaneously. The concept of free and frank advice might well resist definition not because no one has tried hard enough, but because any definition that captures one dimension tends to obscure another. The ghost, it turns out, needs all four walls of the room to haunt.

Trust, then, is not a single thing that operates uniformly across the advisory relationship. It is contested, multi-valent, and differently valued across different practitioner orientations. The accountability mechanisms designed to increase it may paradoxically destroy the conditions that make it possible. We are approaching the end of the theoretical landscape. Next week, we bring the political theory to a close.

Next in the Series

We are approaching the end of the theoretical landscape. Next week, we bring this political theory to a close by asking why the silence at the heart of free and frank advice is not a failure of constitutional design but possibly its most important feature.

Drawing on Bonnie Honig’s concept of remainder in democratic theory, we examine the deliberate silences in our Westminster arrangements: the things left undefined not accidentally but purposefully, because defining them would foreclose the democratic contest they are meant to enable. The argument will be that free and frank advice is not merely an undefined concept that has resisted definition: it is potentially a remainder; something that must stay unsettled if it is to keep doing its constitutional work.

From there, we turn to method and then, in March, to the findings. 142 practitioners. Six distinct postures. A structured disagreement along four fault lines. It turns out there is a possible definition of free and frank advice: but not everyone agrees. And the account of how that disagreement works, and why it endures, may itself be the answer to the question we started with: what is free and frank advice?

Footnote

There is a particular irony in the Official Information Act’s treatment of free and frank advice that O’Neill’s framework makes visible. The Act protects the withholding of advice that is free and frank. Yet, as this series has documented at length, neither Parliament nor the courts nor the Public Service Commission has ever defined what free and frank, let alone a free and frank manner actually means. We are therefore in the position of protecting something from scrutiny on the grounds that it possesses a quality that nobody has specified. The protection is real; yet the thing it protects remains broadly undefined. O’Neill would note that this makes trustworthiness not merely difficult to assess but structurally impossible to assess: you cannot evaluate whether advice was offered in a free and frank manner if there is no agreed account of what that manner requires.

References

Baier, A. (1986). Trust and antitrust. Ethics, 96(2), 231–260.

Luhmann, N. (1979). Trust and power. Wiley.

Norman, R. (2003). Obedient servants? Management freedoms and accountabilities in the New Zealand public sector. Victoria University Press.

O’Neill, O. (2002). A question of trust: The BBC Reith Lectures 2002. Cambridge University Press.

Savoie, D. J. (2003). Breaking the bargain: Public servants, ministers, and Parliament. University of Toronto Press.

White, N. (2007). Free and frank: Making the Official Information Act 1982 work better. Institute for Government and Policy Studies, Victoria University of Wellington.