When Murmurations Meet Targeting: A Critical Response to “Cognitive Warfare Without a Map”
The Small Wars Journal article raises important and timely questions about how we conceptualise cognitive warfare. Its murmuration metaphor is evocative and its critique of decision latency is well-grounded. But in seeking to challenge the kinetic targeting paradigm, it overshoots, discarding some of the most valuable intellectual tools we have — precision, problem-framing, and the discipline of focus.
Two specific claims in the article warrant serious challenge, because if accepted uncritically they could lead us toward a model of cognitive operations that is diffuse, ungovernable, and ultimately outmanoeuvred.
The Myth of the Indecisive Node
The article’s most striking assertion is that “there is no decisive node to strike; removing or influencing one account, platform, or individual usually doesn’t change the broader dynamics.” As a first principle of cognitive warfare, this is wrong — and empirically refutable.
Consider what we know of Cambridge Analytica’s methodology, even given the deliberately incomplete public record. The firm did not broadcast indiscriminately into the information ecosystem and wait for emergent behavioural patterns to shift. It did the opposite: it identified discrete, defined personality profiles using psychographic modelling built on OCEAN trait analysis (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism), derived through reverse engineering of Facebook activity. Working from the insight developed by Cambridge academic Michal Kosinski — that as few as 100 Facebook ‘likes’ are sufficient to predict a person’s core psychological traits, and 300 sufficient to model someone’s personality as accurately as their spouse — Cambridge Analytica constructed individual-level profiles across tens of millions of voters. It then crafted tailored messages calibrated to personality type: a neurotic individual received security-themed messaging; an agreeable one received family-values framing. This is not ecosystem-level intervention. This is precision targeting at the level of the individual human mind.
The commercial advertising world operates on the same logic. Programmatic advertising, retargeting, and lookalike audience modelling are not scattershot horizon-scanning activities — they are architecturally designed to identify, fix, and influence a discrete target audience. The entire $600 billion global digital advertising industry is predicated on the proposition that the right message, delivered to the right person at the right moment, produces disproportionate effect. This is not just plausible; it is the foundational business model of every major technology platform on earth.
The murmuration metaphor is instructive here in a way its authors do not intend. Yes, complex flock behaviour emerges from local interactions. But murmurations do change course decisively when a predator — a focused, discrete threat — enters the system. Predators in the cognitive ecosystem exist too. They are the influencers, the trusted voices, the narrative entrepreneurs whose credibility gives them outsized reach. Influence operations that identify and either co-opt or neutralise these nodes are not naive kinetic thinking applied to a non-kinetic domain — they are precision cognitive targeting. The question is not whether decisive nodes exist; it is whether we have the ISR architecture to identify them before our adversary does.
The Precision Problem in Sensing
The article’s second prescriptive pillar — continuous sensing rather than episodic snapshots — is sound, but its implications are incomplete and potentially dangerous if misread as a mandate for horizon-scanning at scale.
The authors are correct that the cognitive environment is in constant motion, and that episodic, snapshot-based intelligence processes cannot track a battlespace that propagates at network speed. Change detection — the foundational ISR technique of comparing two images across time to identify what has shifted — only works if your collection cycle is faster than the adversary’s operational tempo. As Col Jason Brown’s authoritative study on ISR strategy makes clear, the failure of U-2 change detection in Iraq was not a failure of the technique itself; it was a failure of focus. Collection managers, working without a coherent commander’s intent, distributed U-2 coverage across the entire theatre as a ‘peanut butter spread,’ ensuring that re-visitation timelines of four to five days made change detection useless against insurgents planting IEDs within hours. Continuous sensing without directed focus does not solve this problem — it compounds it.
This is precisely the observation that J. Richard Hackman drew from his red-versus-blue team simulations, as cited by Brown. Hackman found that the blue team — the defenders — consistently lost not because they lacked resources, but because they lacked focus. Faced with an adversary with a clear purpose, the blue team flooded the simulation controllers with broad, undirected questions and consequently drowned in the data that came back. The cognitive failure was not a shortage of sensing; it was an absence of targeting logic — of knowing what to look for, where, and why. Hackman concluded that the blue team needed to reorient from a defensive, reactive posture to an offensive, problem-centric one: to determine what it would do if it possessed the red team’s capabilities and resources. As Brown summarises Hackman’s prescription, “just that simple cognitive change can reorient members toward the specific information that has the greatest potential analytic payoff”.
This is not an argument against continuous sensing. It is an argument that continuous sensing must be targeted. Broad, undirected situational awareness is not a warfighting capability — it is a vulnerability. An adversary who understands your collection posture and sees that you are watching everything will find the seams between your sensors and manoeuvre through them. ISR — whether in the physical or the cognitive domain — must be prioritised, focused, and linked to defined intelligence problem sets (IPS). The cognitive domain demands no less rigour than the physical one; if anything, it demands more, because the signal-to-noise ratio is inherently worse and the volume of manipulable data is orders of magnitude higher.
However…
None of this is to dismiss the Small Wars Journal article’s core insight. The authors are correct that the joint targeting cycle — designed around 96-hour air tasking order horizons — is structurally misaligned with an information environment where adversaries can weaponise a narrative in under ten seconds. They are correct that decision latency is generated less by uncertainty than by governance architecture: the sequential staffing, legal review, and approval chains that consume time faster than the cognitive environment remains stable. And they are correct that pre-set authorities and delegated risk tolerances are necessary to enable action at relevant speed — a principle well-understood in direct action targeting but systematically under-applied in information operations.
Where they go wrong is in using these valid observations to argue for an ecosystem approach that deprioritises targeting logic entirely. The NATO Chief Scientist’s 2026 report on cognitive warfare is more nuanced on this point, recognising that the target set does indeed expand — from discrete platforms to cognitive and social systems, trust networks, identity narratives, and institutional legitimacy — but that micro-segmentation of populations for psychographic and behavioural targeting remains a core offensive capability, now amplified by AI. The answer is not to choose between ecosystem awareness and precision targeting. It is to integrate both within a coherent architecture.
Toward a Problem-Centric Cognitive ISR Framework
What the cognitive domain requires is not the abandonment of targeting logic but its evolution — away from a requirements-centric production model and toward what Brown calls a problem-centric, strategy-oriented ISR approach. That means:
Framing intelligence problems, not requirements. Commanders must define the cognitive behaviour and intent they need to understand — not simply task collection against named accounts or platforms. Just as the Marine unit in Helmand shifted its ISR from route scanning to network understanding, cognitive ISR must shift from monitoring content to mapping interaction dynamics within a defined problem space.
Continuous sensing, but targeted continuous sensing. Change detection in the cognitive domain requires revisitation rates faster than the adversary’s narrative cycle. That demands prioritisation — dedicating persistent collection capacity to the specific nodes, networks, and narratives identified as decisive, rather than horizon-scanning across the entire information ecosystem.
Precision psychographic targeting as a capability, not a liability. The Cambridge Analytica episode is morally complex and legally contested, but its methodology is analytically instructive. Identifying the discrete personality profiles most susceptible to specific narratives, and tailoring cognitive effects to those profiles, is not a departure from good targeting practice — it is its highest expression. Cognitive ISR architecture must develop this capability within an appropriate legal and ethical framework.
Pre-set authorities calibrated to velocity thresholds. The authors are right that approval timelines are the proximate cause of cognitive effect misalignment. But the solution is not to abandon deliberation — it is to front-load it, building pre-approved response options tied to measurable indicators of cognitive movement: velocity, coherence, saturation, and drift.
The murmuration is a powerful image. But the lesson military planners should take from it is not that all nodes are equal and no intervention is decisive. The lesson is that the interaction rules governing collective behaviour can be altered — and that altering them at the right node, at the right moment, produces non-linear effects that a purely kinetic targeting model could never anticipate. That is a case for smarter targeting, not for abandoning it.