The Trouble With How We Define Intelligence

What is intelligence? Robert Schuett explains why dominant definitions fail to capture intelligence as the political practice it has become (and in truth always has been).

For a practice as old and universal, yet so often portrayed as dark and sinister, it can seem counterintuitive to ask what intelligence actually is. It may appear even more awkward when such questions are raised by a former practitioner, particularly when they take the form of definitional and theoretical inquiry.

Wouldn’t I know what I was doing, why I was doing it, and within what limits? On a practical level, the answer is of course yes. But when it comes to theory—when measured against bodies of scholarship such as political theory, legal theory, or social theory—the answer must be more qualified, and at times comes close to a resounding no.

To be sure, there is a growing and increasingly rich body of scholarship engaging with the theoretical and philosophical dimensions of intelligence, including work on just intelligence theory as well as critical intelligence theory. Yet despite this progress, there remains a lack of clarity about what intelligence actually is.

This lack of clarity is not accidental. It reflects the dominance of a small number of definitions that are intuitively appealing and practically useful, but conceptually limiting. The problem is not that intelligence resists definition, but that the definitions we rely on narrow its meaning in ways that obscure its political and normative significance.

Three definitions, in particular, continue to structure contemporary thinking about intelligence. Each captures something important; none is sufficient.

One influential definition puts the matter starkly: “Put most simply, intelligence is information that gives policymakers an advantage over their adversaries.” Advanced by Amy Zegart in Spies, Lies, and Algorithms: The History and Future of American Intelligence (2022), the definition fits comfortably within familiar assumptions about power, competition, and survival, and mirrors how intelligence is often justified in policy discourse.

Yet this definition is also strikingly narrow. By equating intelligence with a special kind of information, it overlooks the historical and operational breadth of intelligence work. The historical record shows that intelligence has never been confined to information alone. It has encompassed sabotage, covert action, paramilitary operations, targeted killing, cyber operations, and prolonged campaigns such as the Global War on Terror. As studies of state-sponsored assassination demonstrate—most notably in Killing in the Name of the State, edited by Luca Trenta and Kiril Avramov (2026)—intelligence has often been less about knowing than about doing.

Reducing intelligence to information therefore misses its operational dimension. Intelligence agencies do not merely inform decision-makers; they frequently act on behalf of the state. Intelligence is not only epistemic or analytical but performative or operational, intervening in the world—often in secret, at other times violently—rather than merely interpreting it through tradecraft.

At the same time, defining intelligence in terms of adversarial advantage implicitly anchors it within a realism-by-default logic. Intelligence appears primarily as a tool for outperforming rivals under conditions of anarchy. Questions of legitimacy, restraint, and responsibility are subordinated to effectiveness. This makes it difficult to account for intelligence practices that are cooperative rather than competitive, or communicative rather than coercive—such as intelligence sharing among allies, the Five Eyes, intelligence diplomacy, or the strategic public release of intelligence.

In short, intelligence-as-advantage explains why intelligence matters in rivalry. It tells us far less about what intelligence is as a political practice—one that is necessarily concerned not only with power, but also with values, as even realists concede.

A second definition reacts against such narrowness by refusing reduction altogether: “Intelligence is as intelligence does.” Rather than defining intelligence in advance, this formulation—advanced by Mark Stout and Michael Warner (2018)—directs attention to practice. On this view, intelligence is understood not through prior definition (a priori), but through what intelligence agencies actually do.

There is much to commend in this approach. It recognises that intelligence is more than information, that it takes multiple forms across time and place, and that its meaning cannot be fixed independently of historical context. It also reflects a healthy scepticism toward abstract theorising detached from practice.

Yet this breadth comes at a cost. By refusing to define intelligence at all, this formulation becomes analytically unambitious. If intelligence is simply whatever a polity decides it is, then we are left with a circular account: intelligence is what intelligence agencies do because intelligence agencies do it.

From the perspective of open societies, this is potentially dangerous. If intelligence has no defining characteristics beyond institutional activity, then there are no principled grounds on which to distinguish legitimate intelligence work from abuse. Almost any practice can be justified as “intelligence” after the fact, so long as it carries official authorisation. In such conditions, the real risk is that conceptual openness easily collapses into political permissiveness.

In this sense, the intelligence-is-as-intelligence-does logic captures the empirical diversity of intelligence practice, but at the expense of both theoretical and normative clarity. It tells us what intelligence happens to be, but it leaves us none the wiser about what it can be—or what it ought to be.

A third definition attempts to restore precision: “National security intelligence is a secret state activity to understand, influence, or defend against a threat to gain an advantage.” Advanced by Andrew Macpherson and Glenn Hastedt (2023), this formulation has clear strengths. It moves beyond information alone, incorporates operational activity, and situates intelligence firmly within state authority.

This definition is analytically stronger than the previous two. It recognises that intelligence involves action as well as analysis, and that secrecy is not incidental but constitutive of intelligence work.

Yet it remains firmly tied to a realism-by-default logic. Intelligence is still framed primarily in terms of threats and advantage. Its purpose is survival and security, not order, legitimacy, or justice. The political environment appears as a Hobbesian landscape of existential dangers to be managed rather than a society of states governed—however imperfectly—by shared norms and expectations.

Moreover, by anchoring intelligence so firmly in secrecy and national security, this definition struggles to account for the increasingly social character of intelligence practice.

Intelligence today is routinely shared among allies, institutionalised in partnerships, embedded in international organisations, and at times mobilised publicly. These practices sit uneasily with an understanding of intelligence as purely secret state activity directed against threats.

What unites these three definitions is not what they include, but what they exclude. All three are state-centric, but society-blind. They treat intelligence primarily as a means to an end—advantage, activity, threat management—rather than as a political practice embedded in a wider social and normative context.

Each definition fails in a different way. The first is too thin, reducing intelligence to information. The second is too loose, collapsing normativity altogether. The third is too closed, locking intelligence into a narrow survival logic. What none of them can explain is how intelligence operates within a world of ongoing relationships, shared expectations, and contested norms. These expectations are rarely codified, often violated, and constantly contested, but they are nonetheless very real.

Definitions remain indispensable, but only insofar as they illuminate, rather than obscure, the kind of political practice intelligence has become, and in many respects always has been. Until intelligence is understood in these terms, scholarly and policy debates will continue to oscillate between instrumental justification and descriptive empiricism. From the perspective of a former practitioner, this is not simply a theoretical concern for intelligence theory or intelligence philosophy, but a practical one. For without shared definitions, there can be no shared limits—most notably among partners and allies.

The task, then, is not to abandon definition. The task is to rethink how intelligence is defined and why that choice matters for how intelligence is understood, judged, and constrained in international life.

Robert Schuett is co-founder and managing partner at STK Powerhouse, a global risk advisory firm. A former Defence civil servant, he also serves as Chairman of the Austrian Political Science Association and is a long-standing Honorary Fellow at Durham University.

Photo by Gantas Vaičiulėnas