Every system operating under finite resources – seemingly every system now – needs to compress its environment: simplify, filter, decide what to pay attention to and what to let go. The fix isn’t to track everything perfectly or to try to see everything – it would be impossible to do so. Instead, the solution is to deliberately focus on the areas where errors carry the greatest consequences.

Australia’s national security ecosystem faces a threat environment that has expanded not merely in volume but in complexity. Threats are interacting, overlapping and blurring together in ways that no single agency or analytical framework can hold in focus at once. The instinctive response is to invest more – more data, more analysts, more collection, more coordination. This is not wrong, but it rests on an assumption that needs to be challenged: that more resources automatically means better awareness. At some point, adding more information increases noise and reduces clarity.

The deeper problem is not a shortage of resources; it is a problem of design. Australia should design its security architecture around diverse, independently maintained analytical frameworks focused on the distinctions that carry the highest consequences, rather than pursuing the comprehensive awareness that complexity will always outpace. That means being explicit about two questions: where can we not afford to be wrong, and where is good enough actually good enough?

This isn’t just a pragmatic hunch; it’s supported by research into how organisms perceive complex environments. A system that tries to track everything faces costs that grow in lockstep with complexity. It will always be outpaced. A system that accepts deliberate approximation, reserving its sharpest focus for the distinctions that matter most, can scale. Such distinctions include, for example, whether a cyber intrusion is state-backed or criminal, or whether a foreign engagement is diplomacy or covert interference. Below the threshold of consequence, approximation is safe. Above it, every error matters.

Hamas’s attack against Israel on 7 October 2023 is an example of what can happen when this design fails. Israel’s intelligence community operated under a settled view of Hamas: deterred, pragmatic, invested in stability. Hamas, as Muhanad Seloom has documented, spent years studying how Israeli intelligence saw the world, then fed that system exactly what it expected to see: diplomatic restraint; managed provocations followed by calm; and compartmentalised planning that denied signals intelligence any contradictory intercepts. Israeli analysts were not incompetent; they were operating within a framework deliberately manipulated by an adversary who understood it better than they did. When junior analysts raised alarms, the institutional system filtered them out. The framework held until it catastrophically did not.

This is the failure mode that matters: not missing data, but a system that has become rigid, predictable and legible to its adversaries. A single analytical framework, no matter how sophisticated, is structurally vulnerable. What one framework filters out, another may catch. What looks benign through one lens may look suspicious through another.

Australia’s ecosystem already has this diversity. Intelligence agencies, law enforcement, cyber specialists, academic researchers, policy experts, industry partners and allied services all see the world differently. That is not a coordination problem to be solved by standardisation; it is a strategic asset to be deliberately cultivated. The risk runs in the opposite direction: well-meaning drives toward unified threat pictures can gradually erode the variety that makes the system hard to fool. Coordination and diversity are both necessary. The design challenge is achieving one without sacrificing the other.

What does a system designed around targeted fidelity rather than comprehensive awareness actually look like? We offer five recommendations.

—Name the things you cannot afford to get wrong. The distinctions that carry the worst consequences if confused should be explicitly identified and regularly revisited. This is a continuous practice rather than a settled judgment. The Israeli experience shows what happens when it calcifies into consensus.

—Cultivate diverse frameworks deliberately. The value of having multiple agencies and partners looking at the same environment through different lenses is that their blind spots are different. That diversity deserves the same deliberate investment as interoperability and standardisation.

—Build contestability that actually works. Structured dissent mechanisms are only useful if they can challenge the dominant framework itself, not just conclusions within it. A devil’s advocate who operates within the same professional culture they are meant to challenge risks becoming a ritual.

—Treat boundary-spanners as critical infrastructure. Boundary-spanners are people who have multi-sector sensemaking functions. This includes cleared academics, embedded industry partners, and cross-agency liaison officers. They are not simply assets that are nice to have. They are the points at which different ways of seeing the world get compared and calibrated.

—Resist the high-fidelity trap. The right question is not how to gather more information but how to aim finite attention with precision. Some threats are best watched through automated systems. Others need community engagement. Others need formal modelling. Matching method to domain is itself a capability.

The lesson of 7 October is not that Israel needed more intelligence; it had enough. What it needed was a system designed to know where it could not afford to be wrong, maintain genuinely independent ways of seeing, and challenge its own assumptions before an adversary could exploit them. That is the standard Australia’s security architecture should be designed to meet.