📈 economics

The algebraic structures of social organizations: the operad of cooperative games

This paper establishes a conceptual framework for cooperative game theory by endowing the collection of all games with an operad structure, proving via the Möbius transform that it is isomorphic to the operad of commutative triassociative algebras, and demonstrating how this algebraic perspective unifies previous composition methods, characterizes stable game classes, and yields explicit formulas for solution concepts like the Shapley value and the core.

Original authors: Dylan Laplace Mermoud, Victor Roca i Lucio

Published 2026-04-08

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Dylan Laplace Mermoud, Victor Roca i Lucio

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where every social interaction, from a family deciding on dinner to a parliament passing a law, is a game. In this game, people form groups (coalitions) to achieve a goal, and the size of the group determines how much "reward" they get. This is the world of Cooperative Game Theory.

For decades, mathematicians have studied these games. But there was a problem: they treated every game as a giant, isolated list of numbers. If you had 10 players, you needed a list of over 1,000 numbers to describe every possible group. It was messy, complex, and hard to see the big picture.

The Big Idea: The "Lego" of Social Games
This paper, written by Dylan Laplace Mermoud and Victor Roca i Lucio, proposes a revolutionary way to look at these games. They suggest that complex social situations aren't just random lists of numbers; they are built by stacking simpler games on top of each other, like building a tower out of Lego bricks.

They use a mathematical tool called an Operad. Think of an operad as a set of instructions for how to snap Lego bricks together.

The Bricks: These are simple games (like a 2-person bargaining game or a "dictator" game where one person decides everything).
The Snapping: The paper defines a precise rule for how to replace one player in a big game with an entire new sub-game.

The Magic Formula: The "Möbius" Translator
The authors discovered something incredible: there is a secret code, called the Möbius Transform, that translates these complex social games into a language mathematicians already understand perfectly.

They proved that any cooperative game, no matter how complicated, is just a combination of two basic building blocks:

The Dictator: A game where one person holds all the power.
The Bargainer: A game where two people must agree to get anything done.

It's like saying that every symphony in the world is just a specific arrangement of just two musical notes. This discovery simplifies the entire field, turning a chaotic jungle of possibilities into a structured, predictable system.

Real-World Examples: The European Council
Why does this matter? Let's look at the European Council.

The Big Game: 27 countries voting on laws.
The Sub-Games: Each country isn't a single person; it's a complex system of its own voters, parties, and elections.
The Connection: The behavior of "France" in the big game is actually determined by the "game" happening inside France.

The authors' framework allows us to mathematically "zoom in" and "zoom out." We can see how the internal politics of one country (the sub-game) snap into the larger European voting game. If the internal game changes, the whole tower shifts. This helps us understand how local decisions ripple out to affect global outcomes.

The "Core" of the Matter: Keeping the Peace
In game theory, the "Core" is the set of fair deals where no group of players feels cheated enough to break away and form their own group. If a game has no "Core," it means the group is unstable and will likely fall apart (social conflict).

The paper shows that if you build a complex game out of smaller, stable games, the big game is likely to be stable too. It's like building a house: if every brick and every beam is strong, the whole house stands firm. They provide formulas to predict exactly when a complex social situation will hold together and when it will collapse.

The Solution: Fairness Indices
Finally, the paper looks at how we measure power.

The Shapley Value: A famous formula that calculates how much each person contributes to the group's success.
The Banzhaf Index: A formula that measures how often a person's vote is the "deciding" one.

The authors found that when you combine games, these power measurements combine in a beautiful, predictable way. You don't have to recalculate everything from scratch; you can just multiply the power of the "big picture" by the power of the "small picture."

In a Nutshell
This paper is a "User Manual" for social complexity. It tells us that the messy, chaotic world of human cooperation is actually built from simple, repeating patterns. By understanding how to snap these patterns together, we can better predict how groups behave, how to keep them stable, and how to distribute power fairly. It turns the art of social organization into a precise science of building blocks.

1. Problem Statement

Cooperative game theory models scenarios where groups of individuals (players) form coalitions to achieve a collective payoff. A fundamental challenge in this field is the complexity of aggregation: real-world social organizations (e.g., parliaments, the European Council, corporate boards) are rarely monolithic; they are often composed of subgroups where individual behavior is determined by internal sub-games.

Historically, researchers like von Neumann, Morgenstern, Shapley, and Owen have proposed various notions of game composition (decomposing a large game into smaller sub-games). However, these approaches lacked a unified, systematic algebraic framework. Existing definitions were often ad-hoc, specific to certain classes of games (e.g., simple games or normalized games), and did not provide a general structural theory for how games combine, decompose, or interact under iteration.

The paper aims to:

Establish a rigorous algebraic framework for the composition of cooperative games.
Unify previous disparate notions of game composition (sums, products, tensor compositions) under a single structure.
Analyze how solution concepts (Core, Shapley value, Banzhaf index) behave under this composition.

2. Methodology

The authors employ the mathematical theory of Algebraic Operads. An operad is an algebraic structure that encodes operations with multiple inputs and one output, along with rules for composing these operations.

Operadic Definition: The authors define the collection of all cooperative games, $\mathcal{G}$ $G$ , as an operad.
- Objects: $\mathcal{G}(n)$ is the vector space of all $n$ -player grounded cooperative games (where $v(\emptyset)=0$ ).
- Partial Composition ( $\circ_i$ ): The core operation is defined by replacing the $i$ -th player of a "quotient" game $\Gamma_A$ with an entire "component" game $\Gamma_B$ .
- Formula: For games $\Gamma_A = (A, \alpha)$ and $\Gamma_B = (B, \beta)$ , the partial composition $\Gamma_A \circ_i \Gamma_B$ is a game on the player set $(A \setminus \{i\}) \cup B$ . The value function $\gamma$ for a coalition $S$ is:
  $\gamma(S) = \beta(B)\alpha(S_A) + \partial_i \alpha(S_A) \cdot \beta(S_B)$
  where $S_A = S \cap A$ , $S_B = S \cap B$ , and $\partial_i \alpha$ is the marginal contribution of player $i$ in game $\alpha$ .
Isomorphism via Möbius Transform: The authors utilize the Möbius transform (which maps a game to its Harsanyi dividends) to construct an explicit isomorphism between the operad of cooperative games and a known algebraic structure: the operad of commutative triassociative algebras ( $\text{ComTriass}$ ).
Suboperad Analysis: They investigate whether specific classes of games (e.g., convex, simple, balanced) are closed under this composition, thereby forming suboperads.
Solution Aggregation: They study how standard solution concepts (Core, Shapley, Banzhaf) interact with the partial tensor product induced by the operad structure.

3. Key Contributions and Results

A. The Operad of Cooperative Games

The paper proves that the collection of all cooperative games forms a unital partial operad.

Generalization: This structure generalizes all previous composition methods:
- Shapley's composition of simple games.
- Owen's tensor composition of normalized games.
- Shapley's sum and product of simple games (recovered via specific compositions with 2-player bargaining and dictator games).
Generators and Relations (Theorem B): The authors provide an explicit presentation of the operad. They prove that the Möbius transform induces a canonical isomorphism:
$\mathcal{G} \cong \text{ComTriass}$
Consequently, any cooperative game can be expressed as a linear combination of iterated compositions of two fundamental 2-player games:
1. The 2-player bargaining game ( $B$ ).
2. The 2-player dictator games ( $d_1, d_2$ ).

B. Stability of Game Classes (Suboperads)

The authors demonstrate that many important classes of games form suboperads (they are stable under the composition operation). This includes:

Normalized games ( $v(N)=1$ ).
Simple monotone games (equivalent to combinatorial clutters).
Capacities (non-negative, monotone games / fuzzy measures).
Convex games (supermodular games).
k-monotone games (including totally monotone games/belief functions).
k-alternating games (including submodular games/plausibility measures).
Balanced games (games with a non-empty core).

C. Geometric Implications: Generalized Permutahedra

A significant consequence of the stability of convex games is the discovery of an operad structure on generalized permutahedra.

Convex games are in bijection with generalized permutahedra (via their cores).
Since convex games form a suboperad, the family of generalized permutahedra inherits an operad structure. This provides a new algebraic perspective on polytopes used in combinatorial optimization and discrete geometry.

D. Aggregation of Solutions

The paper analyzes how solution concepts behave under composition:

The Core: The authors define a partial tensor product map between the cores of component games and the core of the composite game.
- They prove that if the cores of the components are non-empty, the core of the composite is non-empty (under certain conditions).
- They provide conditions for injectivity and surjectivity of this map, showing that the "balancedness" of a social situation is robust under the addition of internal details (sub-games).
Shapley Value: An explicit formula is derived for the Shapley value of a composite game using the Möbius transform.
Banzhaf Index (Theorem E): A striking result is found for the Banzhaf index. The index of a composite game is the partial tensor product of the Banzhaf index of the quotient game and the Shapley value of the component game:
$\psi(\Gamma_A \circ_i \Gamma_B) = \frac{1}{2^{|B|-1}} \left( \psi(\Gamma_A) \otimes_i \phi(\Gamma_B) \right)$
This reveals a deep structural link between the Banzhaf index and the Shapley value within the operadic framework.

4. Significance

Unification: The paper resolves the fragmentation in cooperative game theory by providing a single algebraic language (operads) that encompasses decades of disparate definitions of game composition.
Structural Insight: By identifying the operad of games with $\text{ComTriass}$ , the authors unlock powerful algebraic tools (generators, relations, homological properties) to study game theory problems.
Interdisciplinary Bridge: The work connects cooperative game theory with:
- Combinatorics: Through clutters and generalized permutahedra.
- Geometry: By endowing polytopes with an operad structure.
- Decision Theory: By formalizing the structure of fuzzy measures and capacities.
- Evidence Theory: By showing that belief functions and plausibility measures form suboperads.
Practical Application: The results on the stability of the Core and the explicit formulas for Shapley and Banzhaf indices provide robust tools for analyzing complex, hierarchical social organizations (e.g., voting systems, corporate governance) where decisions are made at multiple levels.

In summary, this paper elevates cooperative game theory from a collection of specific models to a structured algebraic discipline, demonstrating that the "composition" of social interactions is not just a heuristic but a fundamental algebraic operation with rich geometric and combinatorial consequences.