Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.16498

From Caricature to Collective Artifact: How Persona Construction Drives Polarization in Generative Social Simulations

Dimitris Michailidis·Petter Törnberg·Fernando P. Santos
View original source ↗Read source document ↗Export metadata ↓

Abstract

Generative Social Simulation (GSS) leverages Large Language Models (LLMs) to create sophisticated personas that define agents with demographics, emotions, and political opinions. While prior work shows that personas compress outputs into stereotypical caricatures, there is no comprehensive study of how these caricatures evolve into collective artifacts when they interact. Using a Twitter-like GSS platform with personas derived from the American National Elections Study (ANES), we ablate three layers of partisan identity: affective polarization markers, party identification, and vote choice, while retaining demographics, ideological positioning and important issues. We find that collective outcomes depend heavily on how survey responses are translated into personas. Removing a few explicitly partisan statements moves the network from near-complete segregation to cross-party ties approaching those of real Twitter networks. We argue that matching one outcome is not evidence for behavioral validity and point to salience as a possible explanation. Surveys ask for a position on every question regardless of its importance to the respondent, and persona construction turns each answer into a direct behavioral instruction. We conclude with a checklist for more comprehensive evaluation of personas in future works.

Explore related subjects

cs.SIcs.AI

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Dimitris Michailidis, Petter Törnberg, Fernando P. Santos. 2026-10-12. From Caricature to Collective Artifact: How Persona Construction Drives Polarization in Generative Social Simulations. https://arxiv.org/abs/2610.16498

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Semantic Communities and Boundary-Spanning Lyrics in K-pop: A Graph-Based Unsupervised Analysis

We examine how lyric representation shapes computational similarity using four 384-dimensional multilingual MiniLM representations of 7,891 K-pop songs by 1,456 artists from charts spanning 2000 to October 2023. Whole-song truncated embeddings provide the strongest standardized separation between neighbouring and random song pairs (1.995), compared with 1.931 for chunked documents, although chunked representations show greater adjusted Rand agreement and line-based representations yield higher modularity. Using whole-song embeddings, we construct a weighted cosine nearest-neighbour graph ($k=30$). A representative Louvain partition identifies 21 operational communities, without a corresponding robust plateau in the Markov Stability analysis. AI-assisted examination of selected lyrics and terms suggests overlapping patterns of memory, separation, longing, self-reflection, attraction, and performance. Community membership is associated with character-script composition (Cramér's $V=0.294$; permutation $p=0.001$). High-participation songs exhibit broader cross-community connectivity, while participation has weak negative associations with embedding-relative novelty. These findings show how representation choices and textual characteristics influence the network organization of a cultural corpus, subject to limitations from truncation, graph construction, and the absence of independent human semantic validation.

cs.SI↗
arXiv2026

Coupling Macro Dynamics and Micro States for Long-Horizon Social Simulation

Social network simulation aims to model collective opinion dynamics in large populations, but existing LLM-based simulators mainly focus on aggregate dynamics while largely ignoring individual internal states. This limits their ability to capture opinion reversals driven by gradual individual shifts and makes them unreliable in long-horizon simulations. We propose MF-MDP, a social simulation framework that tightly couples macro-level collective dynamics with micro-level individual states. MF-MDP explicitly models per-agent latent opinion states with a state transition mechanism, combining individual Markov Decision Processes at the micro level with a mean-field collective framework at the macro level. This allows individual behaviors to change internal states gradually rather than trigger instant reactions, enabling the simulator to distinguish agents that are close to switching from those that are far from switching, capture opinion reversals, and maintain accuracy over long horizons. Across real-world events, MF-MDP supports stable simulation of long-horizon social processes with up to 40,000 interactions, compared with about 300 in the baseline MF-LLM, while reducing long-horizon KL divergence by 75.3% (1.2490 to 0.3089) and reversal KL by 66.9% (1.6425 to 0.5434), significantly mitigating the drift observed in MF-LLM. Code is available at ai4ss.github.io/M2Sim.

cs.SI↗
arXiv2026

DeTox-Fed: Detecting Toxic Conversations in the Fediverse with Federated Graph Neural Networks

The rise of decentralized social networks (DSNs), and in particular the rapid uptake of the Fediverse (e.g., Pleroma, Mastodon, Lemmygrad), introduces new challenges in content moderation. Independent instances host their own data, follow different moderation policies, and often observe only partial views of conversations. We present DeTox-Fed, a federated graph-learning framework for detecting toxic conversations in DSNs without requiring instances to share raw conversations or moderation labels. Each instance constructs a local conversation graph, where nodes represent conversation trees and edges capture shared user participation across conversations. A Graph Neural Network is then trained in a federated learning setup, allowing instances to collaboratively learn a toxicity classifier while preserving data locality. Unlike text-only moderation approaches, DeTox-Fed combines conversational structure, user-interaction patterns, conversation-level statistics, and aggregate sentiment signals. We evaluate the framework on a large Pleroma conversation dataset and a second one from Mastodon, showing that it achieves stable toxic conversation detection under limited local labels, partial client participation, and varying moderation thresholds. Our results indicate that federated graph-based moderation is a promising direction for semi-automated moderation in decentralized social networks.

cs.SI↗