Stability of Relationship Structures: A Cooperative Game Theory Approach

Welcome back. In our previous lessons, we developed a rigorous framework for analyzing the stability of a cooperative arrangement. We defined the core as the set of all payoff allocations that leave no individual or subgroup with a rational incentive to break away, and we explored transferable utility (TU) as the medium through which negotiations and side payments can steer a relationship into this stable core.

This lesson brings these concepts to a head to address the learning outcome: Evaluate the stability of different relationship structures (e.g., V-polyandry vs. triad) using cooperative game theory. We will move from analyzing a single structure to comparing two common polyamorous configurations. By modeling each as a distinct cooperative game, we can investigate which is inherently more stable and why. This comparative analysis is central to applying game theory strategically, as it informs not just how to act within a structure, but which structure to form in the first place.

Defining the Relationship Structures

Let's consider two common three-person relationship configurations involving our familiar participants: Charles (C), Hannah (H), and Ben (B).

The Symmetrical Triad: In this structure, all three individuals are in a relationship with each other. The connections are {C, H}, {H, B}, and {C, B}. It forms a complete triangle, where every pair is a functioning partnership.
The V-Polyandry (or "Vee"): In this configuration, one person acts as the "pivot" or "hinge" and is in a relationship with the other two, but those two are not in a relationship with each other. They are "metamours." For our example, let's assume Hannah is the pivot, with connections {H, C} and {H, B}. There is no direct romantic or sexual relationship between Charles and Ben.

This diagram illustrates a V-polyandry structure, also known as a 'Vee'. Here, Bob is the 'pivot' in relationships with Alex and Carol, who are metamours and not in a direct relationship with each other.

The fundamental difference between these structures will be captured in their respective characteristic functions, which assign a value to every possible coalition.

Modeling and Analyzing the Structures

To compare stability, we must first translate each structure into a cooperative game. This involves defining the characteristic function, $v (S)$ , for each.

1. The Triad

We can use the characteristic function for the triad that we developed in previous lessons, which assumes some synergy between all pairs:

$v ({C}) = 2, v ({H}) = 3, v ({B}) = 1$
$v ({C, H}) = 8, v ({H, B}) = 7, v ({C, B}) = 4$
$v ({C, H, B}) = 15$

As we established, this game has a non-empty core. For instance, the allocation $x = (6, 5, 4)$ is in the core. This means that a symmetrical triad, given these values, is a potentially stable structure. There exist ways to distribute the total "utility" of 15 such that no one has an incentive to leave.

2. The V-Polyandry ("Vee")

Now, let's model the Vee structure where Hannah is the pivot. The values of the individual players and the dyads involving Hannah can remain the same, as those relationships exist in this structure too. The crucial difference lies in the coalition of the two metamours, {C, B}.

Since Charles and Ben are not partners, the value they can generate together, without Hannah, is likely minimal. They might offer each other some platonic support, but there is no romantic synergy. A reasonable assumption is that their combined value is simply the sum of their individual values (i.e., no synergy).

Let's construct the characteristic function for the Vee:

$v ({C}) = 2, v ({H}) = 3, v ({B}) = 1$
$v ({C, H}) = 8, v ({H, B}) = 7$
$v ({C, B}) = v ({C}) + v ({B}) = 2 + 1 = 3$ (This is the key difference)
The grand coalition $v ({C, H, B})$ value might also differ. Lacking the C-B link, the total synergy might be less than the triad. Let's set $v ({C, H, B}) = 14$ .

Now, we analyze the core of this Vee game. An allocation $x = (x_{C}, x_{H}, x_{B})$ is in the core if it meets these conditions:

Efficiency: $x_{C} + x_{H} + x_{B} = 14$
Individual Rationality: $x_{C} \geq 2$ , $x_{H} \geq 3$ , $x_{B} \geq 1$
Coalitional Rationality:
- $x_{C} + x_{H} \geq 8$
- $x_{H} + x_{B} \geq 7$
- $x_{C} + x_{B} \geq 3$

Is the core non-empty? A quick check of the balance conditions confirms that it is. For example, the allocation $x = (4, 7, 3)$ is in the core:

Efficiency: $4 + 7 + 3 = 14$ .
Individual Rationality: $4 \geq 2$ , $7 \geq 3$ , $3 \geq 1$ . All satisfied.
Coalitional Rationality: $4 + 7 = 11 \geq 8$ , $7 + 3 = 10 \geq 7$ , $4 + 3 = 7 \geq 3$ . All satisfied.

Since the Vee also has a non-empty core, it too is a potentially stable structure.

The Stability Comparison: Where Does the Difference Lie?

Both structures can be stable, so what is the practical difference? The answer lies in the constraints that define the core.

Compare the coalitional rationality conditions:

Triad: $x_{C} + x_{B} \geq 4$
Vee: $x_{C} + x_{B} \geq 3$

The condition for the metamour pair {C, B} is weaker in the Vee structure. Because Charles and Ben can only generate a value of 3 on their own, it is much easier to find a payoff distribution within the Vee that keeps them satisfied. In the triad, where their dyad is assumed to create a value of 4, any stable allocation must grant them a combined payoff of at least 4.

This reveals a crucial insight:
The V-polyandry structure, by design, limits the power of the metamour-only sub-coalition. This reduces one potential axis of instability, making the overall structure less vulnerable to a breakup by that specific pair.

The "price" for this added stability is the opportunity cost of the value that a C-B relationship might have generated. The triad aims for a higher total value ( $v = 15$ ) but faces tighter constraints to maintain stability. The Vee settles for a lower total value ( $v = 14$ ) but has looser constraints, potentially making it easier to manage and sustain.

This demonstrates how cooperative game theory provides a formal language to analyze the trade-offs between different relationship architectures. The "most stable" structure is not an absolute; it depends on the values the participants can generate in various combinations.

Beyond the Core: When Perfect Stability is Impossible

The core is a powerful but stringent concept. It demands an allocation that is invulnerable to defection by any subgroup. In reality, many relationships function even if their core is empty, meaning no perfectly stable state exists. This is where more nuanced solution concepts become essential.

These concepts model stability not as a complete absence of conflict, but as a dynamic equilibrium where objections can be met with counter-objections. The following technical report provides an excellent overview of these ideas. Given your background in operations research, you will find its formal treatment of these concepts clear and direct.

[PDF] On Game Theory and Coalition Formation. - DTIC

This report discusses the limitations of early solution concepts and introduces more sophisticated ones that are critical for analyzing games without a non-empty core.

Please read section 1.2, "Game Theory and Coalition Behaviour." Pay close attention to the paragraphs describing alternatives to the core. Focus on the introduction of these three key concepts: The "bargaining sets," which model stability based on a system of objections and counter-objections. The "kernel," which is a subset of the bargaining set. The "nucleolus," which identifies a unique point, often considered the "fairest" or most stable single allocation.

These solutions—the bargaining set, kernel, and nucleolus—provide ways to predict outcomes even in inherently unstable situations. They represent states of negotiated truce rather than perfect harmony.

This diagram shows the relationship between several key solution concepts for a three-player game. The core is the shaded hexagonal region of stable allocations. The Shapley Value, Nucleolus, Kernel, and Pre-Kernel are typically single points that represent 'fair' or particularly robust outcomes, often lying within the core if it exists.

A Broader Perspective from Matching Theory

The problem of forming stable relationships can also be analyzed using the framework of two-sided matching theory, which studies how to pair agents from two distinct sets (e.g., men and women, or in our context, couples and bulls) based on their preferences. This field offers a compelling high-level result on the stability of different systems.

[PDF] Towards a theory of mutual mate choice: Lessons from two-sided ...

This paper from evolutionary biology applies two-sided matching theory to mating systems. It provides an accessible, non-mathematical argument for why some structures are inherently more stable than others.

Please read the subsection titled "Is monogamy unusually stable?". Note the conclusion that monogamous systems are "unusually stable" because they always guarantee the existence of a stable matching, whereas many other systems, including general polygamy and coalition formation, may not.

The insight from this reading is that simpler structures with fewer degrees of freedom (like monogamy) are more likely to have stable solutions. As relationship structures become more complex (like a triad, quad, or larger network), the number of potential sub-coalitions grows exponentially. This creates a web of interlocking interests and constraints, making it much more likely that the game will be "unbalanced" and possess an empty core, leading to inherent instability.

Conclusion

In this lesson, we applied the tools of cooperative game theory to evaluate and compare the stability of different relationship structures.

Key Takeaways:

Relationship structures like a Symmetrical Triad and a V-Polyandry can be modeled as distinct cooperative games with unique characteristic functions.
The stability of a structure is determined by the properties of its core, which in turn depends on the values that can be generated by all possible sub-coalitions.
A Vee structure can be more robust against breakups by the metamour pair because that sub-coalition has little value on its own, weakening its bargaining position. This stability comes at the cost of the potential value that the metamour relationship might have generated.
When the core is empty, other solution concepts like bargaining sets, the kernel, and the nucleolus provide a more nuanced analysis of stability, modeling it as an equilibrium of threats and counter-threats.
More complex relationship structures with more participants and interconnections have a greater number of potential sub-coalitions, increasing the mathematical difficulty of finding a stable allocation and making them potentially more prone to instability.

Preview of the Next Module:

We have now concluded our study of cooperative games, which focuses on coalition formation and the stability of payoff distributions. We've analyzed which coalitions are stable. We now pivot to a new question: how do we design the rules of engagement within a relationship to guide behavior toward desired outcomes? This is the domain of Mechanism Design. In our next lesson, we will begin this new module by learning to "Frame the creation of relationship rules as a principal-agent problem, identifying potential moral hazard and adverse selection."

Recommended resources