The Relational Conflict Game (RCG) is a tensor-based simulation system for modeling cooperation, conflict, and the transitions between them in multi-actor systems. Built on UCF/GUTT™, it departs from conventional early-warning systems and statistical-forecasting approaches by representing actors and their interactions as nested relational structures evolving over time, rather than as collections of independent features or linear trends.

Conflict and cooperation are not properties of individual actors; they are properties of the relational fabric in which actors are embedded. A conventional analysis that catalogues actors and assigns them stable attributes — military capacity, GDP, ideology, demographic profile — captures only the surface. The dynamics that matter — escalation, de-escalation, alliance shift, defection, reconciliation — emerge from the structure and evolution of relations themselves, not from the attributes of the parties to those relations. The RCG treats this directly: it analyzes the relational structure as the primary object, with actors and their attributes as derived rather than primary.

This thesis applies at every scale. The same relational structure that explains the dynamics between two adjacent nation-states explains the dynamics between two adjacent departments in a corporation, two adjacent business units after a merger, two communities in a fractious region, or two individuals in a strained partnership. The mathematical framework does not change between scales; the parameters and the data sources do.

For a multi-actor configuration specified at sufficient detail, the RCG produces structured assessments of the current relational state, projected trajectories under varied scenario assumptions, identification of likely bifurcation points where the system may transition between cooperation and conflict regimes, and indicators of the points at which targeted intervention is most likely to alter outcomes. Outputs are designed to support strategic foresight rather than point prediction — the questions it answers are of the form what regime is the system in, where are the inflection points, and what configurations of intervention would shift its trajectory, rather than what specific event will occur on what specific date. The latter class of question is generally unanswerable about complex adaptive systems, and the framework does not pretend otherwise.

Traditional early-warning systems and geopolitical risk models reduce complex situations to static features and linear trends. They typically operate by extracting feature vectors from event databases — counts of incidents, sentiment scores from news corpora, economic indicators, demographic measures — and then training classifiers or regression models on historical outcomes to produce risk scores. This approach has well-documented limitations. It collapses the relational structure into independent features, losing the very information most predictive of regime change. It treats conflict as a binary classification problem when the actual phenomenon is a continuous evolution between regimes. It is brittle to novel configurations not represented in the training data. And it produces probabilistic point estimates without exposing the relational mechanisms that drive them, making the outputs difficult to interpret and difficult to act on.

The RCG addresses these limitations by retaining the relational structure as the primary analytical object. Cooperation and conflict are modeled as coexisting and co-evolving rather than as exclusive categories. The temporal evolution is tensor-based, retaining memory and capturing smooth transitions rather than discrete snapshots. The framework is multi-scalar and multi-axial: the same actors participate in economic, security, ideological, and informational relational layers simultaneously, and shifts in one layer propagate to others in mathematically traceable ways. Foresight is emergent and system-wide rather than reduced to per-actor risk scores.

The RCG is suitable for engagements in international relations and diplomacy, including alliance dynamics, treaty negotiation analysis, multi-party security configurations, and economic-political interaction modeling. It is suitable for organizational systems analysis, including post-merger integration, multi-stakeholder governance, regulatory engagement, and large-coalition project management. It is suitable for community-scale and interpersonal dynamics, including mediation, structured negotiation support, and resilience assessment for relational systems under stress. The mathematical framework is the same across these scales; what differs is the data, the parameterization, and the time horizon.

The system is documented in a public white paper: Fillippini, M. (2025). The Relational Conflict Game (RCG): A Tensor-Based Simulation Rooted in UCF/GUTT (Version 1.1.1). Zenodo. The white paper presents the system's framing, its differentiation from conventional approaches, and selected outputs. Methodology and source materials beyond the white paper are not publicly disclosed. All source code, parameterization, and operational specifications are private and accessible only under license.

Engagements involving the RCG typically proceed through an evaluation phase under NDA, in which the scope of analysis, the actors and relational layers of interest, and the time horizon are scoped jointly with the engaging organization. Production engagements proceed under Enterprise License or Research License terms depending on institutional context and intended use.

Inquiries: Michael_Fill@protonmail.com

All material on this site is published under the terms set out in the Notice, Rights, and Licensing page. AI and machine-learning training, fine-tuning, retrieval-augmented inference, and inclusion in any embedding index or vector store are expressly prohibited. Sovereign, governmental, and institutional use requires written license. Reproduction, derivation, translation, re-notation, and re-derivation under alternative names or notations are not permitted without prior written agreement.

UCF/GUTT™ is a trademark of Michael Fillippini. © 2023–2026 Michael Fillippini. All Rights Reserved.