Cognitive Arbitrage: Complexity, Variety and Human Cognitive States are Related

by John Bicknell and Martin Jetton


Arbitrage is a process for finding and capitalizing on system disparities or imbalances. Cognitive arbitrage identifies and capitalizes on opportunistic moments within systems based upon changes in human cognitive states. Complexity, the variety of system activities, and human cognition are fundamentally related. This article explains this relationship, provides examples, and recaps several use cases which help set the conditions for Information Advantage (IA).

A brief introduction to complex systems science is needed first.

Introduction: Complex Systems

(Understanding of environment-system interactions is important. Interested readers might also appreciate reading about The Coin of the Realm, which discusses similar concepts and systems prediction. [1])

Long time listeners to the Cognitive Crucible podcast may recall that Prof. Yaneer Bar-Yam discussed the relationship between complexity, information, and entropy. In this discussion, he asserts, “…Information, complexity, and entropy are really kind of the same thing looking at them from slightly different perspectives.” Prof. Bar-Yam also mentions Ashby’s Law of Requisite Variety [2], which states that all systems must be able to respond to the variety of environmental stimuli or ultimately fail. What happens to the human brain during moments of excessive environmental variety?

Research by other respected cybernetics [3], complexity [4], and psychology [5]–[7] pioneers also reports linkages between system complexity, activity variety, and the capacity for humans to perform tasks or understand unfolding situations. Moreover, the ability to reduce entropy is one of the central hallmarks of any living organism [14]. The more complex systems get, the more likely humans deploy simplifying cognitive heuristics which some philosophers believe are similar to the physical science Principle of Least Action [3], [8], [9].

In short, one person can only know so much. There is a limited variety–or number of different things–that a person can respond to successfully. Highly complex tasks exceed an individual’s capacity to perform or understand [10]. As human system complexity increases, two things are true:

  1. People are contributing a greater variety of activities
  2. Simultaneously, greater activity variety challenges human cognition — meaning that people have a more difficult time orienting, prioritizing, deciding, and taking action.

Measures of complexity–or variety–therefore provide a lens into the cognitive state of humans who are either managing or participating within systems of interest. They tell a story of influence and where people are spending their time, resources, and energy. Human systems of interest include (but are most certainly not limited to):

  • Corporations
  • Social media and news
  • Urban environments
  • Orbital regimes
  • Combat operations
  • Tribes

But what do we mean by system “managers” and “participants?” Managers are people who at some level have a controlling interest in the system. Depending upon the system, however, the amount of management control varies. Participants, on the other hand, are people who experience systems. System managers are frequently participants; however, most participants are not managers.

  • Human system managers include:
    • Regional, state, provincial agency or tribal officials, secretaries, ministers
    • Critical infrastructure industry leaders, or global brands
    • News outlet owners, funders, or editors
    • Military commanders
    • City officials
    • Satellite operators
  • Human participants who are experiencing systems include:
    • Citizens, local nationals, or combatants
    • Critical infrastructure employees or customers
    • Consumers of social media or news feeds

Let’s consider two very different examples.

Example 1: Urban Crime Complexity

Imagine three different urban neighborhoods: A, B, and C. During a single month, Neighborhood A has 100 crimes with just two varieties: 50 Robberies and 50 Aggravated Assaults. Neighborhood B has 90 crimes reported during the same month: 30 Robberies, 30 Aggravated Assaults, and 30 Shootings. Finally, Neighborhood C has the fewest number of crimes reported with 80 but has four varieties–20 Robberies, 20 Aggravated Assaults, 20 Shootings, and 20 Rapes.

Who are the system managers in this scenario? They include the city government, law enforcement, first responders, and hospitals. Compared to Neighborhood A, the “managers” of Neighborhood C need a greater variety of response capabilities in order to manage the larger variety of crimes. Additionally, personnel with very different skill variety are needed to manage Rape or Shooting victims compared to those required to manage a Robbery victim. Would it not make sense that system managers responsible for Neighborhood C have a different cognitive state than those managing A?

Now what about the citizens who are experiencing these different systems? Similarly, the citizens experiencing Neighborhood C are having a different daily life that affects their cognitive states–especially when compared to Neighborhood A.

What cognitive inferences may be made for system “managers” and system “participants?” How might information operations practitioners use complexity measures and tailor influence strategies accordingly?

Example 2: Complexity in Global News

Imagine the global system of states–or actors. In this system, actors are interacting on the world stage with one another continually. They’re making trade deals; they’re breaking trade deals; they’re engaging in diplomatic discussions, combat operations, verbal and material threats, on and on. All of these events are being captured in news sources such as print, TV, web, and radio. The GDELT Project [11] hoovers up thousands of news outlets in about 100 different languages and labels the actors,  themes, and events within each document. These global events get coded as relatively cooperative or relatively conflicting. For more information about the GDELT project, check out this Cognitive Crucible interview with the creator, Kalev Leetaru.

In this global news system, the variety of possible activities consist of verbal cooperation, material cooperation, verbal conflict, and material conflict. In a recent government project, we discovered that 69% of global news tends to be cooperative, on average. This means that between countries, actors primarily send and receive cooperative verbal and material actions such as promises of trade deals, planning diplomatic meetings, or sending aid.

Systems which contain a large variety of events related to sending and receiving verbal and material cooperation and conflict are considered complex. Low complexity is measured when systems have fewer varieties of news events. Complexity measures provide a glimpse into the cognitive states of elite audiences, government leaders, news outlet managers, and even citizens.

Imagine a relatively uncomplex system where actors are contributing a single activity variety: verbal cooperation. Now, imagine a system in which actors are also sending and receiving activities which are verbally and materially conflicting. Such a system is more complex; now, actors are contributing a larger variety of activities: verbal cooperation, verbal conflict, and material conflict. The complexity of these systems may be measured and compared at scale. National security organizations–and even commercial brands–may make cognitive inferences about the humans who are involved in these systems and conduct arbitrage.

Complexity and Information Advantage Use Cases

Strategic competition is an enduring condition to be managed, not a problem to be solved [12]. Free societies are in an open ended “infinite game” [13] which has no predetermined endpoint. Rather than winning, organizations must continually compete and keep the game going while pursuing goals. During his Cognitive Crucible episode on Information Advantage (IA), John Agnello asserts that operational units achieve IA when holding the initiative in terms of situational understanding, decision making, and relevant behavior–especially during the competition continuum phase of operations. Here are two use cases which leverage complexity insights:

  1. Key Leader Engagement. Army Techniques Publication ATP 3-13.5 defines Soldier and Leader Engagement (SLE) as a potent capability that commanders and staffs employ to create effects that can result in a decisive advantage over adversaries or enemies and opportunities with unified action partners [14]. SLE occurs at all levels and across the full range of military operations. Measured complexity serves as an input into KLE dossiers and helps maximize opportunity for favorable outcomes. For example, system complexity related to piracy and maritime matters in the Philippines is consistently high; how might this relatively complex (high variety) system affect the cognitive state of Philippine maritime officials? And how might this inform KLE meetings with Philippine maritime officials?
  2. Visualize the Information Environment (IE). Visualizing the IE is about understanding. This includes understanding the best timing for various outcomes, informing more effective approaches, reducing ambiguity, and improving engagement quality. Color-coded time series geospatial heat maps depict relative complexity, from which cognitive inferences may be made about human managers or participants within systems of interest. Visualizations may also be scaled from the geopolitical or strategic level all the way to tactical relevance. 


Complex systems are, well, complex. And, they do not share insights easily. Novel methodologies which generate actionable insights from complex systems are highly desirable, therefore. This is especially true since these solutions tend to apply to diverse problem sets, as discussed above. Complexity is an enduring feature of our national security landscape. 

Measuring changes in system complexity is the first step towards cognitive arbitrage. Complexity and the capacity for humans to perform tasks or understand unfolding situations are related. As human system complexity increases, two things are true. First, people are contributing a greater variety of activities to the system. Second, greater activity variety challenges people’s ability to respond successfully. Cognitive arbitrage is possible.

About the Authors

John Bicknell is the CEO and Founder of More Cowbell Unlimited. A national security thought leader and passionate analytics visionary, he has written extensively on national security matters related to information warfare, critical infrastructure defense, and space situational awareness. Before retiring from the United States Marine Corps in 2010 as a Lieutenant Colonel, John served worldwide, most notably in Afghanistan and at the Pentagon. He led enterprise-level process intensive human resources supply chain projects designed to discover inefficiencies, architect solutions, and re-purpose manpower savings. In his corporate career, he operationalized an Analytics Center of Excellence for a large EdTech firm, among other accomplishments. John is Vice President for the Information Professionals Association and host of The Cognitive Crucible podcast. His Master’s degree from the Naval Postgraduate School emphasizes econometrics and operations research.

Martin Jetton has worked in AI/ML aspects of cyber security, retail, transportation, human capital management, fraud detection, utilities operations and marketing optimization.  Martin has designed, implemented and led analytic solutions development across different industries, diverse processes and disparate data sources. He recognizes the value in measuring success, tying metrics and implementing solutions that last beyond his efforts. Examples include seasonal retail optimization of product availability, freight routing and mode selection to minimize operational costs, field personnel schedule to meet difficult to schedule changes, optimization of hourly hiring to account for seasonal demand and prediction of money laundering/fraud in bank transactions. Martin’s graduate education emphasizes operations research, systems science, and computer simulation.


[1] J. Bicknell and B. Russell, “The Coin of the Realm: Understanding and Predicting Relative System Behavior,” Information Professionals Association. 2023

[2] W. R. Ashby, “Requisite variety and its implications for the control of complex systems,” Cybernetica, vol. 1, no. 2, pp. 83–99, 1958.

[3] M. Lissack, “Requisite Variety, Cognition, and Scientific Change,” 2020.

[4] J. Boyd, “Destruction and Creation,” 1976.

[5] G. A. Miller, “The magical number seven, plus or minus two: Some limits on our capacity for processing information,” Psychol. Rev., vol. 63, no. 2, pp. 81–97, 1956, doi: 10.1037/h0043158.

[6] W. R. B. Gibson, “The Principle of Least Action as a Psychological Principle,” Mind, vol. 9, no. 36, pp. 469–495, 1900.

[7] Thinking, Fast and Slow, Kahneman, Daniel. 2011.

[8] M. Lissack, “Understanding Is a Design Problem: Cognizing from a Designerly Thinking Perspective. Part 1,” She Ji J. Des. Econ. Innov., vol. 5, no. 3, pp. 231–246, Sep. 2019.

[9] M. Lissack, “Understanding Is a Design Problem: Cognizing from a Designerly Thinking Perspective. Part 2,” She Ji J. Des. Econ. Innov., vol. 5, no. 4, pp. 327–342, Dec. 2019.

[10] Yaneer Bar-Yam, Why teams?, New England Complex Systems Institute, 2017.

[11] “The GDELT Project.”

[12] “Joint Concept for Competing.” Joint Chiefs of Staff, 2023.

[13] S. Sinek, The Infinite Game. Penguin Random House, 2020.

[14] “Army Techniques Publication (ATP) 3-13.5 (Soldier and Leader Engagement).” Army TRADOC, Dec. 21, 2021.