Triadic HHAI Conditions in Network Models

• Triadic HHAI conditions are defined by the interplay of preferential attachment, moderate homophily, and intermediate triadic closure that together yield increased global inequality and decreased segregation.
• PATCH model simulations reveal that higher probabilities of triadic closure amplify degree inequality while moderating homophily diminishes segregation, producing counterintuitive network outcomes.
• Empirical applications in coauthorship and citation networks validate that balancing these mechanisms is crucial for designing interventions that address both inequality and integration.

Triadic HHAI conditions refer to parameter regimes in stochastic network models—specifically, those incorporating Preferential Attachment (PA), Homophily (H), and Triadic Closure (TC)—in which the interplay of these three attachment mechanisms generates the joint outcome of increased global degree inequality and reduced segregation. This behavioral regime is central in recent large-scale modeling of social network inequality such as PATCH, which is designed to explain persistent disparities in real-world and synthetic networks by tuning the relative strengths of PA, H, and TC. “Triadic HHAI” thus denotes not a structural motif, but a system-level statistical consequence of interaction between these linking mechanisms, particularly when homophily and triadic closure are both moderate and the modes of attachment combine both global and local rules (Bachmann et al., 27 Sep 2025).

1. Mechanistic Foundations: Preferential Attachment, Homophily, and Triadic Closure

In PATCH, every new node connects to $m$ existing nodes via a two-stage stochastic process that combines mechanisms:

• Preferential Attachment (PA): Targets nodes with probability proportional to their degree $k_j$, implementing a “rich-get-richer” dynamic.
• Homophily (H): Assigns weights $h_{ij}$ depending on whether pairs $(i, j)$ share group identity, with $h$ modulating the in-group preference: $h_{ij}=h$ if same-group, $1-h$ otherwise.
• Triadic Closure (TC): For a link, with probability $\gamma$ the candidate targets are restricted to the “friend-of-friend” set (FoF), otherwise drawn from all nodes (the “global” pool).

These mechanisms are integrated by assigning selection weights in each linking event. In the most general “PAH” form, the probability that new node $i$ links to candidate $j$ combines degree and group similarity:

$p_{ij} = h_{ij}k_j$

This $p_{ij}$ then enters convex mixtures of global and local (TC) candidate pools, mediated by $\gamma$:

$$P(i \to j) = (1-\gamma)\,P_G(i\to j) + \gamma\,P_{TC}(i\to j)$$

where $P_G$ and $P_{TC}$ are normalized by their respective candidate sets.

2. Quantitative Outcomes: Global Inequality and Segregation Measures

The PATCH model formalizes network inequality with two primary metrics:

• Gini coefficient ($G$): Quantifies degree inequality as

$$G = \frac{1}{2N^2\mu} \sum_{i=1}^N \sum_{j=1}^N |k_i - k_j|$$

where $k_i$ is the degree of node $i$ and $\mu$ is the mean degree.

• Segregation index ($S$): The EI-index, defined as

$S = \frac{E-I}{E+I}$

with $E$ the number of out-group and $I$ the number of in-group edges. $S=-1$ indicates total segregation; $S=0$ is neutral mixing.

Additionally, group-level degree disparity $$\Delta \langle k \rangle = \langle k \rangle_{\text{maj}} - \langle k \rangle_{\text{min}}$$ tracks mean degree differences between groups.

3. Parametric Regimes and Emergent Triadic HHAI Effects

Simulation and analysis of PATCH confirm that the individual mechanisms drive network structural properties differently:

• Increasing homophily $h$ (preference for in-group) strongly decreases $S$ (increasing segregation) for any fixed $\gamma$.
• Preferential attachment dominates $G$; increasing $h$ with PA present increases $G$ sharply.
• Triadic closure $\gamma$ typically amplifies $G$ (super-linear degree inequality) due to the “friendship paradox,” making high-degree nodes more frequent as FoF targets, but attenuates $S$ (reduces segregation) especially when local (TC) attachment is uniform or only weakly homophilic.

4. Definition and Delineation of Triadic HHAI Conditions

The triadic HHAI regime is characterized as follows:

• Moderate homophily: $0.6 \lesssim h \lesssim 0.85$
• Intermediate-to-large triadic closure probability: $0.3 \lesssim \gamma \lesssim 0.8$
• Local TC step: Unbiased or only weakly homophilic

In this regime:

• The Gini coefficient increases with $\gamma$: $\frac{\partial G}{\partial \gamma} > 0$
• Segregation decreases with $\gamma$: $\frac{\partial S}{\partial \gamma} < 0$

This generates the paradoxical effect where intensifying the tendency to close triangles (raise $\gamma$) makes the network as a whole more unequal in terms of node degree (with extreme hubs), yet less segregated by group identity. Pushing $h$ above $\approx 0.9$ or $\gamma$ near unity reverts the system to near-complete segregation and eliminates the desegregating effect of triadic closure.

5. Empirical Applications and Theoretical Implications

PATCH has been validated against longitudinal coauthorship and citation networks in physics and computer science, capturing observed gender disparities when parameterized with moderate homophily and substantial triadic closure. The triadic HHAI regime is essential to reproduce cases where certain minority or majority groups experience persistent disadvantage via high-degree hubs (inequality), but where network-wide mixing (segregation) is less extreme than under pure homophily or PA alone.

A plausible implication is that interventions solely fostering triadic closure may not reduce overall power-law disparities and may even intensify degree inequality while improving group mixing. As such, PATCH and the triadic HHAI regime provide a nuanced framework for designing equity interventions in networked systems (Bachmann et al., 27 Sep 2025).

6. Summary Table: Mechanism Effects and Triadic HHAI Regime

Mechanism	Effect on Gini ($G$)	Effect on Segregation ($S$)
Preferential Attachment	Increases sharply with $h$	Amplifies group disparities
Homophily ($h$)	Non-monotonic U-shape in G	$h \uparrow \implies S \downarrow$
Triadic Closure ($\gamma$)	Typically $G \uparrow$	$S \downarrow$ in moderate $h$

Triadic HHAI regime: $h\sim0.7$, $\gamma\sim0.5$, PAH global and uniform or weak-H local; results in $G$ rising with $\gamma$, $S$ falling with $\gamma$.

7. Significance in Network Science and Inequality Research

The identification of triadic HHAI conditions establishes a parameter region where classic social processes yield counterintuitive macro-structure: strong “hubs” coexisting with less segregated groups. This decoupling of group mixing and degree inequality demonstrates the inadequacy of policies or mechanisms focused on a single attachment bias and highlights the importance of combinatorial mechanisms in driving emergent inequalities. The framework provides not only predictive modeling for empirical networks, but also guidance for interventions in systems where both group integration and equality are desired outcomes (Bachmann et al., 27 Sep 2025).