Adversarial Mindset

Reason about a system from the perspective of other agents who pursue their own interests and will exploit your assumptions, not only from how the system is meant to be used. Whether the other party is an attacker probing for a weakness or a rival anticipating your next move, the discipline is the same: put yourself in their shoes, expect them to find the gap you overlooked, and design and decide against their best response.

Reinforced By

• Threat Modeling — enumerate how an attacker would misuse the system, working from their goals rather than the intended flow.
• Trust Boundary — mark where trust changes hands, because that is exactly where an adversary aims.
• Backward Induction — look forward to the other side's future best responses and reason back to your move, the same foresight an attacker uses against you.
• Credible Commitment — read which of an opponent's threats are real and shape their choices by binding your own, rather than assuming good faith.
• Prisoners' Dilemma — expect self-interested defection from the payoff structure, not the goodwill you would prefer.
• Mixed Strategy — deny an opponent a pattern to exploit by being unpredictable where predictability is a weakness.

Why it Matters

Two domains reach the same stance from opposite ends. Threat Modeling builds it for security: systems are attacked by adversaries who reason about your defenses, so you must reason about their attacks, since the gap you do not imagine is the one they use. The Art of Strategy builds it for any strategic interaction: outcomes depend on others who act in their own interest and respond to your moves, so the central skill is to put yourself in their shoes and anticipate the move you would least like. Security threat modeling is one instance of the general case, anticipating an intelligent, self-interested adversary. The optimistic default, designing only for cooperative, intended behavior, is precisely the assumption a real opponent is paid to violate.

Tension

The stance can curdle into paranoia, treating every counterparty as hostile and every interaction as zero-sum, when many are cooperative or positive-sum and trust is what makes them work. The discipline is to model the adversary where the stakes justify it and to recognize when the game is not, in fact, adversarial.