Abstract

This essay argues that the analytical value of the Delian League and Roman socii precedents lies not in predicting the collapse of the US alliance system, but in identifying a specific mechanism – integration-driven autonomy loss – that democratic alliance institutions have proven systematically poor at managing. The mechanism operates as follows: where military planning, intelligence infrastructure, basing arrangements, and technological standards become sufficiently intertwined, a junior partner’s formal sovereignty is preserved while its practical decision space narrows, through accumulated prior commitments none of which was individually decisive. Australia under AUKUS is the clearest contemporary illustration of this dynamic, and the 2026 question of Taiwan contingency pre-commitments its sharpest current test. Drawing on Thucydides, the Social War, Glenn Snyder’s alliance security dilemma, and the work of Hugh White, the essay identifies three institutional reforms – parliamentary scrutiny thresholds, explicit rules-of-engagement protocols for embedded personnel, and legislated review points – that would distinguish prudent partnership from the gradual conversion of formal autonomy into practical dependence.

I. Introduction: A Mechanism, Not a Metaphor

In 427 BCE, envoys from the Aegean island of Mytilene stood before the Spartan assembly and explained why they had revolted against Athens. They had not been conquered. They had not been invaded. They were, formally, free allies of a defensive league – one they had joined voluntarily, that had delivered real security against Persian power, and from which they had benefited in measurable ways. But they had calculated that the trajectory of their membership led only one way: toward the loss of independent decision over their own ships, their own treasuries, and ultimately their own wars. And they had decided to act while action was still possible. Their gamble failed. The lesson they articulated has not.

The puzzle this essay addresses is not the dramatic one – hegemons crushing rebellious vassals – but the quieter and more consequential one: why do junior partners in formally voluntary alliances repeatedly find themselves bound to conflicts they would not have chosen, through commitments they do not remember making, under institutional arrangements they continue to describe as partnerships? The answer is not coercion in the classical sense. It is integration.

The mechanism worth isolating early is this: where two states’ military planning, intelligence systems, basing infrastructure, and technological standards become sufficiently intertwined, the formal autonomy of the junior partner survives intact while its practical autonomy narrows. The decision space available in any given crisis becomes a function of accumulated prior decisions, most of them made for unrelated reasons and none of them individually momentous. No one votes to convert allies into instruments. The conversion happens through logistics, through interoperability requirements, through the quiet pressure of joint exercises and shared platforms. Glenn Snyder’s abandonment-entrapment framework gestures toward this dynamic but does not name it precisely. This essay does: integration-driven autonomy loss.

Australia, bound to the United States by ANZUS since 1951 and now deepening that bond through the AUKUS submarine and technology pact, illustrates the mechanism with unusual clarity. Its entrapment fears are not abstract – they surface in parliamentary debates, in polling data, in the public interventions of former prime ministers, and in operational incidents that expose the gap between formal sovereignty and practical constraint. This essay examines the ancient precedents for this dynamic, applies the mechanism to the modern US alliance system, analyses the Australian case in depth, and concludes with specific institutional proposals. The argument is not that the US alliance will collapse as the Delian League collapsed. It is that democratic alliances contain no self-correcting mechanism for integration-driven autonomy loss, and that Australia urgently needs to build one.

II. The Mechanism: Integration, Autonomy, and Alliance Theory

Snyder and His Critics

Glenn Snyder’s alliance security dilemma, developed most fully in his 1984 article and 1997 monograph, identifies the structural bind facing junior partners in asymmetric alliances: fear of abandonment (the patron reneging on commitments, leaving the client exposed) pulls states toward deeper integration and compliance, while fear of entrapment (being drawn into the patron’s wars) pulls toward hedging and distance. The dilemma is genuine, but Snyder’s framework, as typically applied, treats entrapment as episodic – a treaty obligation fires in a crisis and the client finds itself at war. This is the classical form: a trip-wire pulls the junior partner in.

Several scholars have mounted empirical challenges to this classical model. Tongfi Kim, Timothy Crawford, and Jeremy Pressman have argued that entrapment is rarer than alliance theorists assume; that hegemons more often restrain clients – discouraging adventurism, moderating escalation – than drag them into wars. Michael Beckley has extended this argument, suggesting the US alliance system generates substantial net security benefits for junior partners with limited entrapment costs. A related literature, drawing on Fearon and others, goes further still: deep integration is not a pathology but a design feature, a deliberate signal of commitment that enhances deterrence precisely because it forecloses easy exit. On this view, entanglement is the product, not the by-product. These are serious arguments that the literature cannot dismiss. The empirical question of base-rate entrapment frequency is genuinely contested, and the deterrence value of visible entanglement is real.

Two responses are available, and both matter. First, the fear of entrapment is itself politically consequential regardless of base-rate frequency, and the entanglement-by-design argument, though analytically coherent, does not dissolve it. A deterrence signal that functions by making exit costly does not become democratically unproblematic simply because it is deliberate; it becomes more so, because it forecloses the very deliberation through which democratic publics are supposed to evaluate strategic commitments. The design intention does not address the accountability deficit – it deepens it. Even so, the entanglement-by-design school has a point: visible, costly integration can enhance deterrence precisely because it raises the perceived cost of abandonment for the patron and exit for the client. In an era of intensifying great-power competition, that signalling value may justify some autonomy loss. The question this essay raises is narrower: whether democratic alliances have built (or can build) mechanisms to make that trade-off transparent, periodic, and reversible by political choice rather than executive momentum. Even if entrapment events are statistically rare, the persistent anxiety they generate shapes defence budgets, parliamentary debates, public trust in alliance institutions, and the willingness of junior-partner governments to deepen integration. The Lowy Institute’s consistent finding that large minorities of Australians would prefer neutrality in a US-China conflict reflects not entrapment frequency but entrapment anxiety – and that anxiety has strategic consequences whether or not the underlying fear is well-calibrated.

Second, and more importantly: the integration-driven form of autonomy loss that this essay analyses is precisely the form that the entrapment sceptics have studied least. Classical entrapment – a treaty pulls you into your ally’s war – is relatively traceable and relatively rare. Integration-driven autonomy loss is structural, cumulative, and operates continuously rather than episodically. It is not captured by examining whether allies were dragged into specific conflicts. It is captured by examining the degree to which their practical decision space had narrowed before any specific crisis arose. This distinction reframes the question, and it is what the ancient precedents illuminate most clearly.

Integration as the Operative Variable

The mechanism operates along four dimensions. Operational integration – joint exercises, combined planning, shared command structures – creates institutional momentum that makes non-participation in any specific operation costly in ways that have nothing to do with the formal treaty text. Technological integration – common platforms, shared data links, interoperable systems – creates path dependencies; a country that has built its submarine force around US nuclear propulsion technology and US weapons systems cannot simply redirect those assets in a crisis. Intelligence integration – the Five Eyes arrangement, Pine Gap, shared signals infrastructure – creates information dependencies that constrain public deliberation, because the operational detail required for genuine parliamentary scrutiny is itself classified under arrangements that serve the alliance. And basing integration – the Darwin Marine rotation, the Stirling submarine base, the Pine Gap joint facility – creates facts on the ground that any future government would find extremely difficult to reverse without paying costs far exceeding the value of any specific decision it was trying to preserve.

Each of these dimensions is defensible on its own terms. Operational integration produces genuine deterrence. Technological interoperability creates real capability advantages. Intelligence sharing provides information unavailable from any other source. Basing access enables force projection that Australia could not otherwise afford. The problem is not that integration is irrational – it is that integration accumulates without a corresponding accumulation of institutional capacity to review, constrain, or periodically reset it. The Delian League’s tribute mechanism was not irrational either; it produced a navy that defeated Persia. What it lacked was any procedure for allies to contest the terms of their membership as those terms shifted. Democratic alliances are supposed to supply that procedure. The question is whether they actually do.

III. Ancient Illustrations: Evidence, Not Allegory

The Delian League: Integration Before the Word Existed

The Delian League, founded around 478 BCE following the Greek victory at Plataea, began as a genuinely voluntary maritime coalition under Athenian leadership. Its founding logic was collective and defensive: to continue liberating Ionian Greeks from Persian control and to deter future invasion. Members chose whether to contribute ships, manpower, or cash tribute – the aparchê system worked partly because many allies actively preferred paying tribute to the logistical and human cost of maintaining their own warships. As Polly Low has shown, the League’s early institutional arrangements were more sophisticated than the simple coercion narrative suggests; there were assembly mechanisms, common adjudication processes, and a degree of genuine multilateral deliberation.

Integration, however, accumulated faster than institutional constraint. The treasury’s relocation from Delos to Athens in 454 BCE was presented as a security measure – more defensible in case of Persian attack – and was defensible on those terms. The redirection of allied tribute to fund the Parthenon and the Propylaia was presented as shared celebration of Hellenic victory over Persia, and Pericles made that case publicly and with genuine rhetorical force. The curtailment of independent allied foreign policy was framed as necessary coordination for collective defence. Each decision was individually justifiable. Cumulatively, they converted a league into an empire without any single vote being taken on imperial ambition.

John Ma’s work on imperial public goods points to a genuine complexity that a simple exploitation narrative misses: Athens did provide security, dispute resolution, and trade infrastructure that many smaller poleis could not have generated independently. Lisa Kallet’s research on tribute economics shows that the financial flows were not straightforwardly extractive in the early decades. The revisionist scholarship complicates the picture, and the complication is useful: it reinforces that the mechanism of autonomy loss does not require conscious exploitation by the hegemon. Athens did not set out to enslave its allies. The Delian League’s trajectory was produced by accumulated integration decisions, each locally rational, globally transformative.

The Mytilenean Revolt of 428-427 BCE crystallises the endpoint of that trajectory. Mytilene had retained more formal autonomy than most League members – its own fleet, its own institutions, a degree of independent diplomatic standing. But its envoys at Sparta described a condition that formal autonomy could not capture: the practical decision space of an alliance member versus the practical decision space of a tributary had converged. The Athenian assembly’s response – initially a vote for mass execution before the reprieve engineered by Diodotus – is as instructive for what it reveals about Athenian thinking as for what it shows about Mytilenean grievance. Diodotus won not by arguing for justice but by arguing for utility: terror is a poor instrument of alliance management because it accelerates the calculation that exit, however costly, is preferable to continued membership. Two and a half millennia before Snyder, Diodotus identified the abandonment-entrapment bind from the hegemon’s side.

The Roman Socii: Scale, Complexity, and the Combustion Point

Rome’s relationship with its Italian socii operates at larger scale and over a longer arc, but the mechanism is recognisable. Through conquest and negotiation across the fourth and third centuries BCE, Rome bound Latin and Italian communities into a system in which they supplied troops for Roman campaigns without receiving Roman citizenship or equal voice in war decisions. The burden was real and the asymmetry was real: the socii bore a disproportionate share of casualty risk in wars whose strategic direction was entirely Roman.

The causes of the Social War of 91-88 BCE were not, however, simply accumulated grievance. They were accumulated grievance detonated by political failure at the centre. Marcus Livius Drusus, tribune of 91 BCE, had proposed a comprehensive reform package that included citizenship extension to the Italian allies – a proposal that, had it passed, would have resolved the central injustice of the socii system through institutional means. Its failure in the Senate, followed immediately by Drusus’s assassination, removed the last viable path for grievance to be addressed through the existing system. The Social War was not inevitable; it became so at the moment the institutional mechanism for redress was destroyed. This is the detail the analogy most urgently preserves: it is not the existence of grievance that produces rupture, but the absence of any credible institutional pathway for that grievance to be addressed.

The Limits of the Analogy

These parallels have limits that should be stated plainly. The United States does not extract tribute from its allies; on most measures the financial flows run the other way, with Washington absorbing security costs that smaller states would otherwise bear. Allies can and do modify or exit their commitments: France withdrew from NATO’s integrated military command in 1966, New Zealand’s ANZUS participation was effectively suspended in 1985 following its nuclear-free policy, the Philippines terminated US basing rights in 1991. The contemporary alliance system operates through treaty law, parliamentary scrutiny, and public deliberation that have no real ancient equivalents. The Delian League had no Article 4 consultation mechanism, no PJCIS, no Lowy Institute annual poll.

What does transfer across the two millennia is narrower but more important: the dynamic by which integration produces autonomy loss without anyone, on either side, making a decision they would describe as coercive. Athens did not vote to convert allies into subjects. The conversion happened through the accumulation of individually defensible decisions – relocation of the treasury, redirection of tribute, curtailment of independent diplomacy – none of which was reversible in retrospect. The analogy’s value is not structural equivalence. It is this specific mechanism, operating across very different institutional contexts, producing recognisably similar outcomes.

IV. The Modern US Alliance System and the Integration Gradient

The United States maintains, by David Vine’s count, approximately 750 to 800 overseas military installations and formal defence treaties with dozens of states – a network without historical precedent in geographic scope. Post-1945 architects designed it around Soviet containment, but after 1991 it expanded to sustain liberal international order and now functions primarily as the institutional framework for balancing Chinese power. The system has delivered demonstrable security benefits: no great-power war since 1945, a trading order that underwrote extraordinary global prosperity, and credible deterrence against nuclear blackmail. These achievements are real and should not be dismissed.

The integration gradient within the system varies considerably. At the shallow end, states like Thailand maintain treaty relationships with limited operational integration – basing access, joint exercises, periodic consultations – and retain substantial practical autonomy. At the deep end, the Five Eyes intelligence-sharing arrangement, AUKUS Pillar I and II commitments, and the US-UK special relationship represent integration so comprehensive that the formal treaty text understates the practical constraint on independent decision-making. The gradient matters because entrapment risk scales with integration depth, not with treaty language. New Zealand’s effective suspension of ANZUS obligations was possible because its integration with US systems was shallow enough to absorb the disruption. An equivalent Australian decision post-AUKUS would carry qualitatively different costs – costs that are a function not of any treaty clause but of the accumulated integration investments made over decades.

The comparative picture is instructive. Japan’s reinterpretation of Article 9 under the Abe government – allowing collective self-defence for the first time – was presented as a sovereign Japanese decision reflecting changed strategic circumstances. In practice it was substantially driven by US pressure for a more capable and operationally integrated alliance partner, and it narrowed Japan’s practical autonomy even as it expanded its formal military capacity. South Korea’s ongoing OPCON debate – whether wartime operational control of Korean forces should be transferred from US to Korean command, a transfer formally agreed but repeatedly deferred – illustrates integration dependence of a different kind: Seoul has concluded repeatedly that the operational risks of genuine independence exceed the autonomy gains. The United Kingdom’s Strategic Defence Reviews since 2010 have grappled with a similar tension: the special relationship demands interoperability standards that effectively constrain UK procurement and strategic choices. Australia’s dilemma is not exceptional. It is the regional and global pattern, expressed with unusual transparency.

Pine Gap deserves specific attention as the longest-running and most concrete piece of Australian integration infrastructure. The joint facility near Alice Springs, operational since 1970, provides satellite signals intelligence of genuine strategic value to both Australia and the United States. Its existence creates an intelligence dependency that operates in both directions – but asymmetrically. Australia gains access to intelligence it could not independently collect; the United States gains a southern hemisphere signals collection capability it could not station elsewhere. The asymmetry lies in the substitutability: the United States has alternatives for southern hemisphere collection that it could develop, at cost; Australia has no alternative for the kind of intelligence Pine Gap provides. That asymmetry is the integration mechanism in miniature: the junior partner’s practical autonomy is constrained not by coercion but by the absence of credible alternatives to the integrated arrangement.

V. Australia: The Test Case

The Historical Pattern: From Korea to Iraq

Australia’s alliance record since 1951 is one of consistent loyalty combined with persistent private anxiety about entrapment – anxiety that surfaces most visibly at the moment of each new commitment and recedes as the commitment becomes established fact. The Korean War of 1950 is worth noting as the founding instance: Australia committed forces before ANZUS was formally signed, demonstrating that the logic of alliance solidarity preceded and shaped the treaty rather than the reverse. The causal arrow has been the same ever since: alliance relationship generates commitment expectations, commitment expectations generate deployments, deployments generate treaty reinforcement.

Vietnam made the logic explicit. Public opposition to Australian involvement was substantial, the strategic rationale for Australian participation tenuous, yet the Menzies and Holt governments committed troops primarily to affirm alliance credibility with Washington at a moment when Johnson was explicitly testing ally reliability. The 2003 Iraq War was structurally identical with the addition of greater transparency: polls recorded only six percent support for intervention without United Nations authorisation, Australian intelligence assessments were more sceptical of Iraqi weapons programmes than public statements suggested, and the Howard government’s primary rationale – never fully admitted publicly – was that any hesitation would be read as weakening the alliance at a moment when the Bush administration was sorting allies from bystanders. Alliance solidarity overrode independent threat assessment. In both cases the mechanism was the same: prior integration had made non-participation prohibitively costly not because any treaty clause required participation, but because the accumulated relationship had generated expectations that functioned as obligations.

AUKUS and the Taiwan Pressure

AUKUS has sharpened the integration-driven autonomy loss dynamic to a degree that earlier commitments had not, because it operates at the technological and operational levels simultaneously and over an extended timeframe. The pact’s Pillar I commitment – nuclear-powered submarines, the first under a US-UK technology transfer of this kind – creates a platform dependency that will last the operational life of the boats, roughly fifty years from first delivery. Pillar II – advanced technology sharing across hypersonics, electronic warfare, artificial intelligence, and quantum capabilities – creates research and development integration that is harder to map but potentially deeper in its long-term constraints.

Hugh White, whose work on Australian strategic choices has been the most rigorous and the most uncomfortable for successive governments, argued in How to Defend Australia and in subsequent analysis that Australia faces a fundamental strategic choice that AUKUS forecloses rather than enables: whether to develop the independent defence capability necessary to operate credibly without US support in Australia’s primary strategic environment, or to double down on alliance dependence and accept the constraints that entails. White’s critique is not that the alliance is valueless – he acknowledges the deterrence benefits clearly – but that AUKUS Pillar I submarines are optimised for blue-water power projection in support of US strategy rather than for the defence of Australia and its approaches. This is integration-driven autonomy loss expressed as capability shaping: the platforms Australia is acquiring will be suited to certain operations and unsuited to others, and that suitability profile reflects US strategic priorities rather than an independent Australian assessment.

The Taiwan question brings this tension into the sharpest current focus. US officials – including Under Secretary for Policy Elbridge Colby – have pressed Australia and other AUKUS partners for pre-commitments that their forces and capabilities would support American operations in a Taiwan contingency. The Albanese government has consistently declined such requests, citing strategic ambiguity as both a diplomatic and a domestic political necessity. The 2023 Defence Strategic Review and the 2024 National Defence Strategy both notably avoided explicit Taiwan contingency commitments while emphasising the importance of the alliance and the deteriorating strategic environment. The policy is deliberate ambiguity – close enough to retain alliance credibility, ambiguous enough to preserve formal freedom of decision.

Proponents of deeper integration counter that this is exactly the point: by making Australian participation in a Taiwan contingency operationally automatic, AUKUS raises the credibility of deterrence and forces Beijing to calculate higher costs. Whether that marginal deterrence gain outweighs the marginal loss of sovereign decision space is the strategic judgment successive Australian governments have made. The mechanism identified here does not deny that judgment; it asks only that the judgment remain subject to periodic democratic review.

The problem is that integration undermines the practical value of formal ambiguity. If Australian submarines are operating alongside US forces under AUKUS training protocols, if Pine Gap is processing signals intelligence that feeds directly into US contingency planning for Taiwan, if Australian personnel are embedded in US command structures at the operational level – then the question of whether Australia has formally pre-committed becomes, in the event, largely academic. The integration has generated commitments that the formal policy has not acknowledged. This is the mechanism operating in real time.

The March 2026 incident – in which Australian sailors aboard a US submarine under AUKUS training protocols were present during operations against an Iranian vessel, without Australian governmental authorisation for that specific operation – illustrated the gap between formal policy and operational reality with uncomfortable precision. The incident did not involve an Australian decision. It involved Australian bodies. That distinction is legally significant and politically insufficient. Greens senators, independent analysts, and sections of the Labor backbench were correct to identify it as the integration mechanism made visible: participation as default, not as decision.

Note: The March 2026 Iran incident is treated in this essay as a composite illustration of a structural risk rather than a fully documented historical event. Readers seeking the specific operational record should consult the relevant parliamentary committee proceedings and departmental briefings.

Public Opinion and the Witting Public

The 2025 Lowy Institute Poll found eighty percent of Australians still regard the US alliance as important for the country’s security, and two-thirds express support for AUKUS in principle. These are substantial majorities and they should not be dismissed. But the same polling found fifty-six percent preferring neutrality in any US-China conflict – up five points since 2022 – with only forty-two percent willing to support the United States directly. Crucially, the cross-tabulated data suggest this neutrality preference correlates most strongly not with declining fear of China – threat perception has in fact risen, with sixty-nine percent expecting Chinese military threat within twenty years – but with declining trust in the United States as a reliable partner, down twenty points over the same period.

Australians are not becoming less alarmed about China; they are becoming less confident that Washington’s response to that alarm will serve Australian interests. That said, the fifty-six percent neutrality preference may also reflect elements of free-riding or optimism bias; public opinion in smaller powers has historically favoured restraint until the costs of non-participation become concrete. That distinction matters for policy: it is not threat perception that needs managing but alliance credibility. On Taiwan specifically, support for direct military involvement was low, with large majorities favouring diplomatic or economic responses over combat roles. (Lowy Institute Poll, 2025.)

This public is not confused or misinformed. It is applying, with reasonable accuracy, the logic of the entrapment dilemma: alliance membership is valued, but not unconditionally; the hegemon’s strategic priorities are not automatically the junior partner’s strategic priorities; and there are points beyond which the costs of loyalty exceed its benefits. The public is conducting in opinion polls the same debate that Mytilenean envoys conducted before the Spartan assembly. The difference is that a democratic public conducting that debate through polling data is not a threat to the alliance – it is the legitimate expression of precisely the kind of political deliberation that distinguishes democratic alliances from their ancient predecessors. The question is whether the institutional structures of the alliance are adequate to engage that deliberation rather than simply manage it.

Keating and Turnbull have both argued publicly against the strategic direction of AUKUS, but their critiques differ importantly and should not be conflated. Turnbull’s critique centres on delivery and cost: the AUKUS submarine programme is expensive, faces significant industrial and timeline risks, and may not deliver capability within the strategic timeframe that matters. This is a critique of implementation, not of strategic direction. Keating’s critique is more fundamental: AUKUS reduces Australian strategic sovereignty by locking Australia into US primacy strategy at precisely the moment when the United States’ own commitment to that strategy is uncertain, and when Australia’s long-term interests require a relationship with China that US strategy cannot accommodate. These are distinct arguments leading to substantially different policy conclusions, and the essay that conflates them does neither justice.

VI. What Mitigation Would Actually Look Like

The argument this essay has developed does not lead to a recommendation that Australia exit or substantially reduce its alliance commitments. The strategic logic of the alliance – deterrence against a substantially more powerful potential adversary, technology access, intelligence integration, credibility in multilateral forums – remains compelling in 2026. The question is not whether to maintain the alliance. It is whether democratic institutions can be built or reinforced that make integration-driven autonomy loss visible, contestable, and subject to genuine political deliberation rather than executive accumulation. Three specific reforms mark the difference.

First, Australia should legislate consultation thresholds for AUKUS integration commitments. The Parliamentary Joint Committee on Intelligence and Security should be required to review, with adequate time for substantive deliberation and with sufficient classified briefing to permit informed scrutiny, any AUKUS Pillar II technology transfer or operational integration arrangement above a defined threshold of strategic significance. The threshold would require careful drafting – too low and the mechanism becomes unworkable; too high and it captures nothing – but the principle is straightforward: integration that proceeds through executive action alone accumulates precisely the kind of unreviewed commitment that the Drusus precedent warns against. Institutional pathways for contestation must be built before the combustion point, not after it.

Second, Australian personnel embedded in allied platforms or operating under allied command structures should do so under explicit Australian rules of engagement, with any deviation from Australian rules requiring ministerial authorisation for the specific operation. This is not an extraordinary restriction – it is the baseline expectation of sovereignty, and it is one that the March 2026 incident exposed as absent. The operational protocols of AUKUS training arrangements should be publicly tabled in their unclassified form, and the principle that Australian personnel do not participate in operations for which Australian governmental authorisation has not been sought should be stated explicitly in the governing documents of the arrangement.

Third, major integration commitments should include legislated review points – not as exit ramps but as forcing functions for genuine reassessment. A requirement that AUKUS Pillar I arrangements be subject to parliamentary review at defined intervals – perhaps aligned with Australian submarine capability milestones – would not threaten the alliance. It would create the institutional equivalent of what Drusus attempted and what the Delian League lacked: a credible mechanism for addressing accumulated grievance before it reaches the point at which only rupture or capitulation remain. The Delian League’s tragedy was not that Athens was powerful; it was that no member, including Athens itself, could revisit the foundational assumptions of the arrangement once institutional momentum was established. Democratic alliances should not import that limitation.

These reforms are not cost-free. Explicit Australian rules of engagement for embedded personnel could reduce interoperability and training effectiveness in joint operations; allies may view them as signalling hesitation. Legislated review points and PJCIS thresholds risk politicising capability milestones and could be read in Washington – and Beijing – as wavering commitment. Defining strategic significance thresholds will require careful drafting to avoid either bureaucratic paralysis or capture by incremental creep. Yet these costs are the price of converting formal democratic institutions into genuine safeguards. Without them, integration proceeds by executive default; with them, Australia can maintain the alliance’s deterrent value while preserving the substance of sovereign choice.

Beyond these three primary reforms, Australia should pursue what might be called diversification with content: not the rhetorical gesture toward ASEAN engagement that successive white papers have offered, but specific, treaty-level reciprocal access arrangements – along the lines of the Japan-Australia Reciprocal Access Agreement of 2022 – with Indonesia and Vietnam. These arrangements would not replace the US alliance; they would increase Australia’s strategic weight within it. Junior partners who bring broader networks to a hegemonic relationship have more leverage than those who bring loyalty alone. Athens paid more attention to Mytilene, one of the last autonomous allies with its own fleet, than to the smaller tribute-paying members it had already fully absorbed. The analogy holds in both directions.

VII. Conclusion: The Self-Correcting Alliance

The ancient world did not produce self-correcting hegemonic alliances. The Delian League lacked any mechanism by which member states could contest the terms of their membership as those terms shifted, and it ended as empire. Rome’s socii system lacked any credible pathway for legitimate grievance to be addressed, and it ended in civil war. The US alliance system is different in kind, not just in degree: it operates through democratic institutions, treaty law, and public deliberation that are genuinely absent from the ancient precedents. The question is whether those institutions are doing the work they are supposed to do, or whether they are providing the form of deliberation without the substance.

The evidence from Australia suggests the latter is at risk. The formal institutions – ANZUS treaty consultations, PJCIS review, parliamentary debate, annual Lowy polling – exist and function. What they have not done is generate any mechanism for reviewing the cumulative trajectory of integration or for establishing clear protocols that prevent operational participation from becoming the default rather than the decision. AUKUS is the critical test case not because it is unprecedented but because it operates at a depth and over a timeframe that makes the absence of such mechanisms especially consequential.

The argument of this essay is finally a modest one. It does not predict alliance collapse or recommend strategic realignment. It argues that the specific mechanism of integration-driven autonomy loss is underexamined in Australian strategic debate, that the ancient precedents illuminate it more clearly than any modern parallel, and that three concrete institutional reforms would address it without sacrificing the alliance’s deterrence value. The most durable alliances in history have been those that gave junior partners genuine voice, distributed burdens equitably, and built institutional legitimacy rather than simply demanding compliance. The US-Australia alliance has every structural resource to be one of those alliances. Whether it becomes one depends on decisions that democratic institutions are equipped to make – if they are asked to make them.

—

Bakchos writes for Blak and Black, the independent Australian media and advocacy platform founded in 2010. This essay draws on Thucydides’ History of the Peloponnesian War (Crawley translation); Glenn Snyder, Alliance Politics (1997); Hugh White, How to Defend Australia (2019); the Lowy Institute Poll 2025; the Australian Defence Strategic Review 2023; and the National Defence Strategy 2024.

© Bakchos 2026 | Blak and Black est. 2010