Loose Threads: The Other Allison

26/05/2026

E te whānau. A longer Loose Thread this week, prompted by a moment in Beijing that has sent half the commentariat scrambling for their Thucydides. Graham Allison is having his moment in the foreign policy sun. But the Allison I want to talk about is the one almost nobody remembers. This post starts with his trap, notes who was already using it, and then turns to an argument about government that I think we need to hear again, especially in Aotearoa, especially this week, as we head into another round of public service reform.

Thucydides and the trap

Earlier this month, at the Great Hall of the People in Beijing, Xi Jinping asked Donald Trump whether the United States and China could overcome the “so-called Thucydides Trap” and establish a new paradigm of major-country relations. The phrase was everywhere within hours.

The concept was popularised by the Harvard political scientist Graham Allison, first in a 2012 Financial Times article titled “Thucydides’s trap has been sprung in the Pacific,” then in a 2015 Atlantic essay, and finally in his 2017 book Destined for War (Allison, 2012, 2015, 2017). Drawing on Thucydides’ observation that “it was the rise of Athens and the fear that this instilled in Sparta that made war inevitable,” Allison’s research team at Harvard’s Belfer Centre identified sixteen historical cases of a rising power confronting an established one over the past five hundred years. Twelve ended in war. The theory does not claim that conflict is inevitable. It warns that structural tensions make it dangerously probable unless both parties take difficult action to avoid it.

Xi’s invocation of the trap in Beijing was not casual. As the Lowy Institute observed, the framing positions China as the historically literate, responsible power seeking to transcend conflict, and implicitly casts the United States as the historically illiterate party most likely to fall into it.

The trap in this context is doing political work alongside analytical work, which narrows the strategic imagination to a binary: confrontation or accommodation. But the binary is already outdated. As I argued in the False Binary post last October, Wong makes the case that China is not a rising power but a risen one: it has not and will not converge with Western norms, but economic elasticity means neither side can afford a complete rupture, which creates space for small and middle powers to build “plurilateral” coalitions rather than loose or binary camps. The trouble, as I explored in the Carney and the Edelman Trust Barometer post in January, is that the “plurilateral” alternative requires social cohesion, which, if we are to believe most of the empirics, is currently being eroded. This matters for all of us, but especially given the next round of public sector reform, for public institutions are the infrastructure through which societies build and sustain cohesion, and the way we design and run them is not a second-order question: it is the primary question. Which brings me to the Allison nobody remembers.

The script everyone else already knew

Before I get there, a small observation. The structural logic Allison described, the anxiety of an established power confronting a rising challenger, is not a script that was waiting for Beijing and Washington to discover it. Decolonial thinkers from Fanon (1961) to Tuck and Yang (2012) to Coulthard (2014) to local scholars and practitioners have long described what happens when established orders confront the political capacity of those they have tried to contain.

The fear that Thucydides identified in Sparta finds ready echoes in the responses of settler states to Indigenous political organisation, and in the discomfort that established global institutions display when the global South declines to remain a grateful audience. There is something gently amusing about watching two superpowers arrive at a script that a great many people in the Pacific, in Africa, in Latin America, and indeed in the whare rūnanga of Aotearoa could have handed them some time ago.

I raise the geopolitics because false binaries are the thread running through this post. The Thucydides Trap asks us to look at the world as if only two nations exist. Our domestic debate about government does the same thing, and it is just as misleading and unhelpful.

The other Allison

In 1983, Allison published a piece with a characteristically wry title: “Public and Private Management: Are They Fundamentally Alike in All Unimportant Respects?” (Allison, 1983). The title takes Wallace Sayre’s much-quoted “law,” that public and private management are fundamentally alike in all unimportant respects (Sayre, 1958), and turns it into a question.

Allison is disarmingly modest about what follows. It is not, he warns, a major research project. It is the reflections of a dean of a school of government who spends his time managing what Cohen, March and Olsen (1972) called an “organised anarchy” rather than thinking, much less writing. Allison offers the thinking as speculation and a clear target for others to shoot at. The modesty is charming. It is also misleading. What Allison produced remains, in my view, one of the most useful pieces of public administration writing of the past half-century.

Allison begins by acknowledging that, at a sufficiently high level of abstraction, all management looks the same. Strategy, staffing, organising, budgeting, managing external constituencies: any general management textbook will give you the list. He then draws on three overlapping accounts of the differences: from Dunlop’s (1979) impressionistic comparison of government and business, from Rainey, Backoff and Levine’s (1976) survey of the public administration literature, and from Neustadt’s (1979) work on the American presidency. All lines of enquiry converge. And every practitioner who has worked in both sectors, Allison notes, marshalling the thoughts of Shultz, Rumsfeld, Blumenthal, Ash, Hamilton and Romney as witnesses, judges public management different from private management, and often harder.

Allison notes that not all the differences carry equal weight. Some are practical: shorter time horizons driven by electoral cycles, tenures measured in months, and no agreed performance metric equivalent to profit. But the differences that matter most, Allison argues, concern accountability, equity, and the prevention of arbitrary power. He argues that the government must show its workings. Its processes are exposed to public scrutiny by design. Legislative oversight and judicial review constrain executive freedom in ways that have no private-sector equivalent. The public manager must balance equity across constituencies, not merely optimise for efficiency. These obligations are not incidental burdens. They, Allison argues, are the actual point.

Underlying all of them is what Allison calls a fundamental constitutional difference: in business, general management is concentrated in a single individual with authority commensurate with responsibility; in democratic government, those same functions are deliberately dispersed among competing institutions. The aim, as Justice Brandeis put it, “was not to promote efficiency but to preclude the exercise of arbitrary power” (cited in Allison, 1983, p. 81).

What is often missed is what Allison does with this analysis. He does not conclude that government is different and therefore cannot improve. My read is that he concludes the opposite. The demand for performance from government, he writes, is “both real and right.” Government’s performance does fall short. But this is the most important finding: direct transfer of private management practices will not close the gap.

Drawing on Chandler’s (1977) history of the managerial revolution in American business, Allison observes that the great productivity gains in the private sector came not from specific techniques but from the articulation of a distinct management function: a self-consciousness among managers about what their role required.

Allison wanted the same for public administration, but not by importing private-sector thinking. In fact, my read of his paper is that his point was precisely this. Because government must do its work in public, because it must show its workings, demonstrate fairness, and operate with a transparency and accountability that private firms are not required to match, public management needs its own self-conscious practice, its own philosophy, so to speak, built from its own problems and its own constitutional foundations. Improvement of “perhaps an order of magnitude” is possible (Allison, 1983, p. 88), but only if the field starts from the problems faced by practising managers, who are given opportunities to learn from other competent practitioners, and something in the system that builds its own case literature. This is a practical, constructive, and hopeful argument. In my reading, it says: understand the constitutional architecture, then do the hard work of building a public management practice that operates within it. His argument is not do what the private sector does.

Why this matters here

That argument matters in Aotearoa because it keeps being collapsed, from both directions, and because a third collapse is now emerging that neither direction anticipated.

A reflex on the political left tends to assume that public administration delivery systems are better than private-sector, Te Ao Māori, or community ones. That assumption collapses because it is not supported by the evidence, and it should at least be unsettled by the Royal Commission of Inquiry into Historical Abuse in State Care (2024), which documented what can happen when the state delivers services to vulnerable people without adequate accountability, transparency, or community oversight. Yet in my experience, the new left does little to make big government accountable in the ways Allison’s constitutional framework demands: showing workings, demonstrating where taxes go, and giving citizens genuine ownership over how public money is spent. The preference is for the state to deliver, and for the rest of us to trust that it will.

The political right collapses the distinction from the other direction. It insists that the government should be run “like a business,” treating the constitutional architecture that makes public administration different, the transparency, the equity, and the dispersal of authority as inefficiencies to be removed rather than as democratic safeguards. And here the irony is acute. Allison drew on Chandler (1977) precisely to show that productivity gains came from building management capability. A government that slashes its workforce by fourteen per cent while demanding higher performance is not emulating Chandler’s managerial revolution. It is doing the opposite: dismantling the very capability that Allison, drawing on the business history, the right claims to admire, argued was the prerequisite for improvement.

Between these two extremes, the institutional response to Allison’s challenge was, in principle, the right one. The Australia and New Zealand School of Government (ANZSOG) set out to do precisely what Allison called for: to build a self-conscious public management practice, to professionalise the field, to develop its own case literature and training doctrine. One of the frameworks ANZSOG adopted to do that work was Mark Moore’s (1995) “public value.” Moore’s approach encouraged public managers to think beyond narrow compliance, to identify and create value for the public, and to operate as strategic actors within their authorising environments. It became part of the dominant vocabulary of the Wellington and Canberra bureaucracies.

But as Rhodes and Wanna argued in what became one of the most cited articles in Australian public administration, Moore’s framework was designed for the American government, where prescribed public management roles carry more authority than Westminster systems allow.

Exported into our system, it encouraged unelected officials to define and co-produce the “public interest,” a function that in Westminster belongs to elected Ministers. Public servants advise and implement; Ministers carry the democratic mandate. When public managers act as entrepreneurial value-creators, they risk becoming what Rhodes and Wanna famously termed “Platonic Guardians”, which, from where I stand, was a fierce warning about rescuing responsible government from unelected arbiters of the public interest. In my humble opinion, ANZSOG got the question right. But it chose the wrong answer. The result was a professionalisation of public management thinking in Aotearoa that, from time to time, undermines the very constitutional safeguards Allison (1983) said mattered most.

The Algorithmic Guardian

I am not coming at this from either camp, and I do not have a tidy ideological home on this question. I am pro rangatiratanga and pro devolution. I am not anti kāwanatanga. Like most people outside the system, my position is somewhat heterodox: I do not think the state is the best vehicle for all delivery, nor do I think the market is, and I believe community and local organisations are often better placed to serve their own people. What I care about is that whoever delivers public services does so accountably, transparently, and without the exercise of arbitrary power. I expect kāwanatanga to do its work in public, to show its workings, and to demonstrate fairness, precisely because so much of the state’s delivery work is with the most vulnerable people in the country, and that work, above all other work, needs to be open to scrutiny.

Which is why the third collapse concerns me most. The public service is still operating in the shadow of public value. Moore’s framework has not been replaced. It remains the training doctrine, the institutional vocabulary, the way public managers understand their role. And now, into that environment, comes AI. The government has announced a wider public service overhaul that includes reducing the core workforce by nearly 9,000 roles alongside a direction to increase the use of AI and other digital tools. AI deployment is now being framed as a basic expectation across public entities. I wrote about the risks of AI in the public sector in the AI Bubble, Hype and Ethics post last year, and those concerns have not diminished. Indeed, I put them in an article in The Conversation recently, and am about to start a series on the local issues when AI meets public administration.

If human public servants exercising Moore’s public value already risked becoming Platonic Guardians, then an AI system optimising for public value outcomes makes that problem permanent and unaccountable. At least the human guardian can be challenged, overruled, and held to account in a select committee, by other offices of parliament, or by the courts. The algorithm cannot. It optimises. It does not deliberate. It cannot show its workings in any democratically meaningful sense. It cannot be asked why it weighed one equity against another. It cannot maintain accountability to a Westminster parliament. The Algorithmic Guardian is not a new problem. It is the Platonic Guardian problem, automated at scale, with human judgement removed and constitutional accountability severed.

And Allison’s original point comes back with force. The constitutional architecture exists to preclude arbitrary power. A human guardian, however misguided, at least operates within a system designed to check ambition with ambition. An algorithmic guardian operates entirely outside that system.

Whether the democratic architecture of governance is subverted by an over-enthusiastic bureaucrat playing the Platonic Guardian, or by an optimised piece of algorithmic code playing the Algorithmic Guardian, the casualty is the same: the primacy of democratic politics, the Westminster chain of accountability, and the constitutional duty to show your workings.

Allison’s argument asks us to be honest about what government is before we set about making it more efficient. Public administration is not private management conducted under more tiresome conditions. It is a constitutionally distinct practice, shaped by the deliberate fragmentation of power, the plural sources of legitimacy, and the irreducible complexity of serving a public that is not a market. The argument is forty years old. It has not been answered. The other Allison deserves a hearing.

Friends, this thinking is a work in progress. Push-back is welcome. I am excited about getting the “Under the Hood” series underway next week.

Ngā mihi nui.

References

Allison, G. T. (1983). Public and private management: Are they fundamentally alike in all unimportant respects? In J. Perry & K. Kramer (Eds.), Public management: Public and private perspectives (pp. 72–92). Mayfield Publishing. (Earlier version published as Harvard Kennedy School Discussion Paper, 1980)

Allison, G. T. (2012, August 21). Thucydides’s trap has been sprung in the Pacific. Financial Times.

Allison, G. T. (2015, September 24). The Thucydides Trap: Are the U.S. and China headed for war? The Atlantic.

Allison, G. T. (2017). Destined for war: Can America and China escape Thucydides’s trap? Houghton Mifflin Harcourt.

Chandler, A. D. (1977). The visible hand: The managerial revolution in American business. Harvard University Press.

Cohen, M. D., March, J. G., & Olsen, J. P. (1972). A garbage can model of organizational choice. Administrative Science Quarterly, 17(1), 1–25.

Coulthard, G. S. (2014). Red skin, white masks: Rejecting the colonial politics of recognition. University of Minnesota Press.

Dunlop, J. T. (1979). Public management. Unpublished manuscript, cited in Allison (1983).

Fanon, F. (1963). The wretched of the earth (C. Farrington, Trans.). Grove Press. (Original work published 1961)

Moore, M. H. (1995). Creating public value: Strategic management in government. Harvard University Press.

Neustadt, R. E. (1979). Report prepared for the National Academy of Public Administration’s Panel on Presidential Management. Cited in Allison (1983).

Rainey, H. G., Backoff, R. W., & Levine, C. H. (1976). Comparing public and private organizations. Public Administration Review, 36(2), 233–244.

Rhodes, R. A. W., & Wanna, J. (2007). The limits to public value, or rescuing responsible government from the Platonic Guardians. Australian Journal of Public Administration, 66(4), 406–421.

Royal Commission of Inquiry into Abuse in Care. (2024). Whanaketia: Through pain and trauma, from darkness to light. New Zealand Government.

Sayre, W. S. (1958). Premises of public administration: Past and emerging. Public Administration Review, 18(2), 102–105.

Tuck, E., & Yang, K. W. (2012). Decolonization is not a metaphor. Decolonization: Indigeneity, Education & Society, 1(1), 1–40.