BECO: Behavioral Economics of Cyberspace Operations

Examples of BECO Biases

Insider Biases and Mitigation

The BECO Black Swans are insiders (aI) that disclose information of vital importance or enable adversaries to penetrate the friendly defenses. Insiders may operate in various building blocks of the BECO framework and are particularly dangerous inStrategy and Operations. Like with all Black Swans (Taleb, 2010), their actions and motivations are analyzed a posteriori, and CO proceeds to fight the last war. The most dangerous insiders are turn-coat defenders. Their downfall is gradual, from minor infractions to full-blown national security violations. This pattern resembles coherent arbitrariness (Ariely, Loewenstein, & Prelec, 2000), a human propensity for mentally anchoring on arbitrarily selected initial conditions (“arbitrariness”) and then making judgments systematically related to the initial selection (“coherence”).

In Ariely’s experiments, students were either paid or received payment for listening to his poetry recital, depending on whether a session was offered as an entertainment or a chore (2009). The students’ initial decision to pay for Ariely’s poetry was as arbitrary as that of Tom Sawyer’s friends agreeing to pay for whitewashing his aunt’s fence, but after that decision had been made, the amounts paid were coherently proportional to the duration of the experience. To fight the formation of undesirable patterns, Ariely urges decision makers to question their repeated behaviors and pay particular attention to the initial decisions in what is going to become a long stream of decisions. Likewise, in BECO it is important to identify and prevent decisions that may turn defenders into insiders and break harmful patterns as soon as they form.

Insider actions may also be forestalled by using a powerful psychological mechanism of cognitive dissonance described by Leon Festinger (1962) as “that if a person knows various things that are not psychologically consistent with one another, he will, in a variety of ways, try to make them more consistent” (p. 93). Ariely provides an example of how “doctors reason that if they are telling others about a drug, it must be good—and so their own beliefs change to correspond to their speech, and they start prescribing accordingly” (2012, p. 81). Furthermore, actions create preferences, because “decisions can be highly sensitive to situational factors, even when such factors are unrelated to the actual utility of that course of action” and individuals “rely not only on stable hedonic utilities but also on their memories of utility for their own past behaviors” (Ariely & Norton, 2008, p. 13). InBECO, cognitive dissonance can be used to enhance the loyalty of the cyber workforce by asking defenders to perform patriotic duties beyond their normal responsibilities and thereby develop a positive mindset. Negative attitudes of defenders must be curtailed before they deepen and lead to adversarial insider actions.

Hawkish Biases

Hawkish biases influence military strategy towards aggressive “hawkish” attitudes and downplay conciliatory, or “dovish,” attitudes beyond common considerations of prudence. They affect attackers on the friendly and adversary sides in the BECOstrategy block. Mitigation of these biases requires actions by attackers, defenders and Red Teams as discussed in this section. The name “hawkish biases” was introduced by Kahneman and Renshon (2009) who reviewed an extensive list of cognitive biases in the context of military and diplomatic actions and found that all of them were strongly directional towards aggression. The set of positive illusions includes “unrealistically positive views of one’s abilities and character, the illusion of control, and unrealistic optimism” (p. 4). Unrealistically positive views lead people to consider themselves better decision makers and negotiators than they are. People experience the illusion of control when they exaggerate the impact of their actions on the outcomes, and under stress, they prefer strategies that they think would give them more control. Unrealistic optimism causes people to overestimate the odds of positive for them events, have more confidence in their predictions than the circumstances warrant, and discount the abilities and skills of their peer group. Political science studies and simulated conflicts have demonstrated that the positive-illusion biases cause leaders to have unrealistically positive views of the balance of military power, and many wars start because leaders on each side believe that they will win. Furthermore, during a conflict, negotiations stall because each side thinks that it has a stronger hand and, therefore, it is less likely to make concessions. Positive illusions take place in BECO when each side overstates its attack capabilities against the other side’s defenses. The attackers’ illusion of control may be exploited by the opponent setting up deception systems.

The Fundamental Attribution Error (FAE) is a bias of explaining behaviors of others by exaggerating their intentions and discounting their circumstances. This bias persists even when people are aware of it. In conflicts, “beliefs in the hostile intentions of adversaries tend to be self-perpetuating—and … self-fulfilling” (p. 8), whereas the true reasons for hostile actions could be in response to the opponent’s domestic politics or to one’s own aggression. With the FAE, the hawkish behavior is prompted by attributing the opponent’s moderate behavior to their situation and their hostile behavior to their disposition. The FAE affectsBECO when each side assumes that the other side is preparing cyber attacks. Moreover, in cyberspace, the FAE may be exaggerated even further, because many exploits are invisible until they are launched. The players must be aware of the FAE and try to distinguish genuine attacks from random events before responding in kind. Considering the difficulty of attribution in cyberspace and the need for an almost instantaneous response, defenders must have effective diagnostic capabilities. When there is a possibility that an adversary may misperceive an attack, direct communications between decision makers are particularly important.

Loss aversion is a manifestation of people’s greater sensitivity to losses than gains. Related biases are the endowment effect of overvaluing the items people already own in comparison to identical items that do not belong to them, and the status-quo bias of the preference for the existing situation even if a change would be more beneficial. Loss aversion negatively affects negotiations, because each side considers its concessions as greater losses than they are gains for the other side. In BECO, the endowment effect causes cyberactors to overestimate the merit of their strategies, processes, and technologies. Recommendations for significant changes must not only be justified logically but also address commanders’ biases, and analyses of alternatives must be conducted by independent parties. Confirmation bias causes commanders to overvalue evidence supporting their beliefs that some types of cyber attacks are more likely, some adversaries are more dangerous, and some defenses are more effective. The BECO countermeasures should include independent reviews and stress tests by Red Teams.

Risk seeking in losses causes people facing a sure loss to take greater risks. In conflicts, the side anticipating a significant loss is prone to engage in a disastrous campaign that has a small chance of winning; instead of ignoring sunk costs, leaders escalate commitment . An agency problem compounds these effects, because leaders (“agents”) are punished for losses and rewarded for gains even in situations where their constituency (“principals”) would have preferred a loss to a foolish risk. In BECO, the players that consider themselves more vulnerable, e.g., non-state entities, may be attacking more aggressively to avoid a certain loss. Considering that in cyberspace the actual capabilities are concealed and perceptions are more potent than in physical realms, irrationally-motivated attacks are more likely. The agency problem is also evident in the botnet phenomena where the owners of infected computers (“adversary users,” or aU) are “agents” who don’t really suffer from the Distributed Denial of Service (DDoS) attacks that they precipitate.

Pseudo-certainty is a bias in multi-stage decision-making of choosing the certain outcome of the last stage while disregarding the probability of reaching that last stage. This situation frequently arises in international politics where decision makers focus on the certainties of the final stage and disregard the contingency of the final stage on preceding stages, which may strongly depend on the decision makers’ choices. Thus, “actors under-emphasize the effect of their own actions” (p. 19). In BECO, this means that an actor may focus on its strength in a full-blown cyber conflict and disregard the statistical uncertainties of the preliminary actions leading to it. Field experiments may indicate that in BECO individual hawkish biases might benefit from a consolidated approach if any of them reinforce or diminish the influence of others.