consensus building

This study addresses the challenge in online deliberation where existing opinion selection algorithms often marginalize minority viewpoints and struggle to balance proportional representation with diversity. To tackle this issue, the paper introduces social choice theory into opinion selection for the first time and proposes a novel algorithm that jointly optimizes diversity and balanced representation. The method integrates representative sampling, diversity-aware optimization, and a multi-criteria evaluation mechanism. Through systematic comparison with alternative strategies, the authors demonstrate that their approach achieves a superior trade-off between proportional representation and diversity, significantly outperforming existing methods that optimize only a single objective.

This SoK paper systematically investigates the CAP trade-offs, security vulnerabilities, and scalability limitations of Directed Acyclic Graph (DAG)-based consensus protocols in distributed ledgers. Methodologically, it establishes a unified evaluation framework integrating CAP theory, functional modeling, and cross-protocol comparative analysis to enable systematic classification and evolutionary trajectory assessment of DAG consensus protocols. Key contributions include: (i) identifying characteristic consistency–availability trade-off patterns inherent to DAG consensus; (ii) uncovering recurrent design deviations and common attack vectors across protocols; (iii) pinpointing critical research gaps—particularly in fairness guarantees, formal verification, and heterogeneous network adaptability; and (iv) proposing a verifiable, forward-looking research roadmap. The study provides both theoretical foundations and practical guidelines for protocol design optimization, standardization efforts, and rigorous security evaluation of DAG-based distributed ledger systems.

In multi-source heterogeneous systems (e.g., supply chains), primal, dual, and proximal agents coexist with immutable interfaces—yet existing consensus optimization methods, such as standard ADMM, assume agent homogeneity and cannot accommodate such heterogeneity. Method: We propose the first distributed consensus planning framework supporting collaborative optimization among all three agent types. By unifying linearized ADMM, dual ascent, and standard ADMM, we design a novel relaxation-and-enhancement update mechanism that accommodates structural mismatches without requiring interface modifications. Contribution/Results: Under mild assumptions, we establish rigorous convergence: O(1/k) rate under weak convexity and two-step linear convergence under strong convexity. Experiments on mixed-agent scenarios demonstrate both robustness and efficiency, validating the framework’s plug-and-play applicability. This work provides the first theoretically grounded, algorithmically practical solution for decentralized decision-making in heterogeneous multi-agent systems.

In decentralized systems—such as blockchain, P2P networks, and crowd navigation—individual agents must spontaneously coordinate and collectively generate order without a global authority. Existing models often rely on consensus protocols or synchronization assumptions, limiting their generality. Method: We propose *semi-topology*, a generalized topological framework that relaxes the requirement of closure under finite intersections of open sets. Agents are modeled as points; admissible coalitions (e.g., PoS majorities, node clusters, pedestrian coordination groups) are formalized as “open sets,” capturing locality, voluntariness, heterogeneity, and autonomy simultaneously. The framework integrates set theory, algebraic structure, and distributed semantics—without assuming consensus or global time. Contribution: This is the first purely topological unification of self-organized order emergence across diverse decentralized systems. Semi-topology provides a foundational abstraction for decentralized cooperation, enabling rigorous, protocol-agnostic analysis of collective behavior and structural stability in open, asynchronous environments.

This paper studies information aggregation via asynchronous majority dynamics in agent networks: each agent initially receives a biased binary signal (accuracy $1/2+delta$) and iteratively updates its public opinion based on neighbors’ opinions, aiming to reach global correct consensus. Focusing on Erdős–Rényi random graphs $G(n,p)$, the work precisely characterizes convergence behavior across the sparse-to-dense transition. When $log n/n ll p = o(1)$, the process converges to the correct consensus within $(1+o(1))nlog n$ steps with high probability; in contrast, when $p = Omega(1)$, a phase transition occurs where incorrect consensus emerges with non-vanishing probability. The key contributions are: (i) establishing a tight time scale of $O(n log^2 n / log log n)$ for convergence, and (ii) revealing a critical interplay between graph sparsity and consensus robustness—quantifying how diminishing edge density fundamentally limits resilience to initial bias.

This work addresses a critical gap in algorithmic collective action research, which has largely focused on single collectives and thus fails to capture how multiple, decentralized groups—each pursuing related yet distinct objectives—jointly influence shared learning systems. The paper proposes the first statistical theoretical framework for multi-collective algorithmic action, specifically examining how distinct collectives coordinate data modifications to collectively steer model behavior in classification tasks. Under partial observability of other collectives’ strategies, the authors integrate statistical learning theory, game-theoretic analysis, and probabilistic bounds to derive a computable success criterion. This bound explicitly quantifies the interplay between collective size and alignment of objectives, and its predictive accuracy for multi-collective coordination efficacy is validated through simulations of climate adaptation interventions in smart cities.

This work challenges conventional multi-agent systems that rely on majority voting or hierarchical aggregation, which often treat consensus as a terminal goal and discard critical reasoning information. Instead, the authors propose aggregating complete reasoning trajectories as fundamental units, generating diverse trajectories through semantic-preserving input perturbations. Their approach integrates an anchoring refinement strategy with provable non-degeneracy guarantees to enable trajectory-level synthesis. Notably, it reveals a “aggregation paradox”: even when all agents converge on an incorrect answer, the correct solution can still be recovered from their collective reasoning traces. Experiments demonstrate that perturbation-induced trajectory variations from a single model significantly outperform ensembles of heterogeneous models across structured reasoning, doctoral-level scientific problems, competitive mathematics, and programming tasks, yielding substantial gains in accuracy.

This study investigates how human societies transform decentralized information into collective judgments and coordinated action under constraints of limited computational resources—specifically time and communication—and reveals the fundamental limitations that computational complexity imposes on social organization. By introducing a novel computational model that transcends traditional Turing-machine paradigms and worst-case complexity assumptions, the work formally characterizes key social mechanisms—such as distributed consensus, hierarchical structures, and external memory—as computational processes for the first time. Integrating tools from distributed computing theory, complexity analysis, and modular modeling, this project establishes a new theoretical direction termed “social computation,” providing a formal framework for analyzing social coordination, scalability, and institutional evolution.

This study investigates how graph gluing strategies influence the convergence rate of consensus in multi-agent systems. Focusing on two interconnection schemes—bridge-based and interface-based gluing—the work employs algebraic graph theory and consensus dynamics to establish a theoretical link among the gluing structure, spectral properties of the graph Laplacian, and system convergence performance. The key contribution lies in elucidating how the number and topology of inter-agent communication links modulate the Fiedler eigenvalue, thereby enabling a quantitative assessment of the enhancement in algebraic connectivity imparted by different gluing mechanisms. Theoretical findings are validated through numerical simulations, offering spectral optimization principles for modular design of multi-agent networks.