Homophily is an important social force in many social networks. While homophilous tendencies among positive relations (such as friendship) and heterophilous tendencies in negative tie networks (such as dislike) have been extensively studied, less is known about how such tendencies might become stronger or weaker as a result of past experiences. Focusing on homophilous and heterophilous tendencies between non-Muslim and Muslim students in schools, this paper explores whether, in line with Intergroup Contact Theory, existing friendship connections with outgroup members will reduce their inclination to dislike outgroup peers compared to ingroup peers (i.e., reduce dislike heterophily), and reduce their preferences for ingroup friendships over outgroup friends (i.e., reduce friendship homophily). On the other hand, in line with Attribution Theory and Social Identity Theory, we propose that existing dislike connections with outgroup members are likely to strengthen dislike heterophilous tendencies and friendship homophilous tendencies, thus further reinforcing religious segregation in schools. To test these hypotheses, we apply Stochastic Actor-Oriented Models (SAOMs) to the friendship and dislike networks of 1678 Muslim and non-Muslim students (FIS data) across 17 German school settings where this religious divide is particularly salient. Contrary to what we expected based on Intergroup Contact Theory, our study shows that prior interreligious friendships do not weaken a person’s tendency for friendship homophily. However, it does seem to reduce heterophily in dislike, but only because it makes students more likely to dislike ingroup peers. On the other hand, prior interreligious dislike ties are found to amplify dislike to outgroups, while not affecting friendship homophily. Surprisingly, our study also shows that ingroup friendships reduce dislike for outgroup members. We explore the implications of the findings that friendship homophily appears unaffected by prior interreligious contacts, while dislike heterophily is strengthened by existing negative outgroup ties, but reduced by both outgroup and ingroup friendships.
We analyze the structure of the world trade network over 2019–2024, spanning the Covid-19 crisis and the 2022 large-scale invasion of Ukraine. Using a network-based approach grounded in non-local indicators, we uncover in particular both the immediate and persistent effects of the Ukrainian conflict on global markets and reveal distinct patterns within clusters such as the G20, EU, and BRICS. Metric evaluations and community detection algorithms based on communicability and resistance distances show that, from a trade-network perspective, the 2022 large-scale invasion of Ukraine constituted a major structural shock—one that affected country blocs in markedly different ways and contrasted sharply with the more uniform impact of the Covid-19 pandemic.
Large Language Models (LLMs) have evolved into AI agents that interact with external tools and environments to perform complex tasks. The Model Context Protocol (MCP) has become the de facto standard for connecting agents with such resources, but security has lagged behind: thousands of MCP servers execute with unrestricted access to host systems, creating a broad attack surface. In this paper, we introduce AgentBound, the first access control framework for MCP servers. AgentBound combines a declarative policy mechanism, inspired by the Android permission model, with a policy enforcement engine that contains malicious behavior without requiring MCP server modifications. We build a dataset containing the 296 most popular MCP servers, and show that access control policies can be generated automatically from source code with 80.9% accuracy. We also show that AgentBound blocks the majority of security threats in several malicious MCP servers, and that the policy enforcement engine introduces negligible overhead. Our contributions provide developers and project managers with a foundation for securing MCP servers while maintaining productivity, enabling researchers and tool builders to explore new directions for declarative access control and MCP security.
Conformal prediction offers a powerful framework for building distribution-free prediction intervals for exchangeable data. Existing methods that extend conformal prediction to sequential data rely on fitting a relatively complex model to capture temporal dependencies. However, these methods can fail if the sample size is small and often require expensive retraining when the underlying data distribution changes. To overcome these limitations, we propose Reservoir Conformal Prediction (ResCP), a novel training-free conformal prediction method for time series. Our approach leverages the efficiency and representation learning capabilities of reservoir computing to dynamically reweight conformity scores. In particular, we compute similarity scores among reservoir states and use them to adaptively reweight the observed residuals at each step. With this approach, ResCP enables us to account for local temporal dynamics when modeling the error distribution without compromising computational scalability. We prove that, under reasonable assumptions, ResCP achieves asymptotic conditional coverage, and we empirically demonstrate its effectiveness across diverse forecasting tasks.
