A firm is not a person. It is a group of people who must somehow act as one. The standard assumption in strategy research is to treat the firm as a unitary actor with coherent preferences — as if organizations want things the way individuals do. This is useful and wrong. Useful because it allows clean game-theoretic analysis. Wrong because organizations are coalitions of individuals with conflicting goals, and the way those conflicts are resolved changes what the organization “wants.”
Csáji, Hauser, and Tao (arXiv:2602.20518) formalize what the aggregation structure does. Under unanimity — everyone must agree — the organization's utility function approximates the pointwise minimum of the individuals' utilities. The most cautious voice wins. Under polyarchy — anyone can act — the utility approximates the pointwise maximum. The most aggressive voice wins.
The organization's risk profile is not a property of its members. It is a property of its decision structure. The same group of people, with the same individual preferences, will produce a risk-averse organization under unanimity and a risk-seeking one under polyarchy. The preferences didn't change. The aggregation did.
This has competitive consequences. In Cournot competition, firms with unanimity-style governance produce less (cautious output decisions). Firms with polyarchy-style governance produce more (someone always pushes for expansion). The market outcome — prices, quantities, profits — depends on the internal structure of the firms, not just their cost functions or demand curves.
The general point: when a collective is modeled as a unitary actor, the model inherits a personality from the aggregation rule. The personality is real — it affects strategic outcomes — but it is not located in any individual member. It is a structural property of how disagreement is resolved. The organization's character is its constitution, not its membership.