Institutions, Networks, and Survival: An Agent-Based Model of Organizational Adaptation Under Sustained Warfare
收藏资源简介:
Replication materials for the paper “Institutions, Networks, and Survival: An Agent-Based Model of Organizational Adaptation Under Sustained Warfare”, presented at SSC-2026. Purpose The model investigates how organizational and self-organizational pathways of adaptation interact under sustained warfare, using wartime Ukraine (2022–2025) as the empirical reference system. It formalizes 27 empirically identified mechanisms as 38 computational rules within a BDI-inspired agent architecture organized around the Coleman Boat. Approximately 22,200 simulation runs across four experiment batteries (mechanism exploration, counterfactual scenarios, complex systems analysis, robustness testing) produce the results reported in the paper. Contents File Description org_adaptation_warfare.nlogox NetLogo 7.0.3 model (main file) 01-declarations.nls Variable and breed declarations 02-setup-initialization.nls Setup and initialization 03-go-loop.nls Main simulation loop 04-landscape.nls Spatial environment and resources 05-civilian-behaviors.nls Civilian BDI cycle and actions 06-military-behaviors.nls Military echelon and combat logic 07-government.nls Fiscal policy and recruitment 08-demographics.nls Birth, death, displacement 09-utilities.nls Helper procedures 10-morris-screening.nls Morris sensitivity analysis (*) 11-calibration-sweep.nls LHS calibration sweep (*) 12-calibration-fitness.nls GA fitness functions (*) ODD.pdf Full ODD protocol model description README.txt Plain-text version of this document (*) These files require external input data (Morris trajectories, LHS design matrices) not included in this deposit. They are provided for code transparency but cannot be executed standalone. See Limitations below. Requirements NetLogo 7.0.3 or later: https://ccl.northwestern.edu/netlogo/. No external extensions are required. All .nls files must be in the same directory as the .nlogox file. The model runs on any platform supported by NetLogo (Windows, macOS, Linux). Quick Start Place all files in a single directory. Open org_adaptation_warfare.nlogox in NetLogo. Select a society preset from the society-preset chooser (default: baseline). Click Setup, then Go. A default run of 300 ticks covers approximately 2.5 years of the wartime scenario. Replicating Paper Results The paper reports approximately 22,200 runs across four batteries. Most can be reproduced using NetLogo’s BehaviorSpace (Tools › BehaviorSpace, select the experiment, click Run). Per-condition run counts are listed in the paper (Table 2) and in the ODD. S1 · Mechanism exploration (160 conditions, 4,800 runs) S1-Mechanism-Exploration 5 presets × 32 toggle combinations × 30 replications. The five toggles are cultural-transmission, trust, reputation, hub-brokering, and long-ties. Mobilization is always active and is not part of the toggle design. B1–B11 · Counterfactual scenarios (4,100 runs) B1-Preset-Comparison B2-Government-Collapse B3-Volunteer-Mobilisation B4-Cultural-Dynamics B5-Trait-Differentiation B6-Migration-Network B7a-Default-Ukraine through B7e-Frozen-Conflict B8-Cascading-Disruption B9-Decentralisation-x-GovStrength B10-Behavioral-Permeability B11-Requisition-Impact C1–C7 · Complex systems analysis (11,650 runs) C1a-Collapse-CivAttrition through C1e-Collapse-MemoryDecay C2-Bimodality C3-Early-Warning C4a-NetFrag-MemoryDecay through C4c-NetFrag-HubDegree C5-Phase-Diagram C6a-Hysteresis-ShockRecovery through C6c-Hysteresis-Occupation C7 (spatial early warning) reuses C3 output and adds no runs. R1–R11 · Robustness (1,620 runs) R1-Perturb-initial-altruism through R10-Perturb-mil-requisition-take-rate Each perturbs one parameter ±20% across 5 levels, 30 replications (1,500 runs total). R11 (noise injection, 120 runs) requires a modified decision procedure and was run separately; it is not in this deposit’s BehaviorSpace. A single default run completes in under one minute on a modern machine, but full batteries benefit from parallel execution. Limitations Several experiments cannot be reproduced from this deposit alone: Morris-Screening and OFAT-Sensitivity require external input files (Morris trajectory matrices and OFAT design points) generated by R scripts not included here. The .nls files containing these procedures are included for code transparency. calibration-sweep-v082 and calibration-refine-v082 depend on Latin Hypercube Sampling design matrices. R11 (noise injection) requires a separate code variant. V1–V8 are validation experiments used during model development; their results are reported in the dissertation rather than in this paper. All other experiments reported in the paper (S1, B1–B11, C1–C7, R1–R10) can be run directly from the deposited code. Key Parameters Parameter Default Role society-preset baseline Structural configuration cultural-transmission? on Cultural tag blending trust-enabled? on Three-route trust model reputation-enabled? on Network reputation effects hub-brokering-enabled? on Hub-mediated introductions long-ties-enabled? on Displacement-resistant ties Five society presets (baseline, total-defense, decentralized, centralized, dual-economy) crossed with all 32 on/off combinations of the five binary toggles above produce 160 experimental configurations. Mobilization is always active as a situational precondition and is not part of the toggle design. Model Overview The model contains 800 civilian organizations, 120–140 military formations, and one government super-agent on a 51×51 patch grid with an east-to-west safety gradient. Five network link types (trade, supply, requisition, coordination, long-tie) connect agents. Each tick represents approximately three days, and a default run of 300 ticks covers about 2.5 years. The warfare scenario follows nine phases calibrated to the documented timeline of the russian invasion of Ukraine. A full model description following the ODD protocol (Grimm et al., 2020) is provided in ODD.pdf. License This work is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0): https://creativecommons.org/licenses/by/4.0/ Contact: a.serdiuk@ukma.edu.ua



