TACITUS ◳

The Vision

At Concordia Discors, we examine conflict as a failure of information legibility before it becomes a failure of coexistence. Polarization is not yet conflict, but it is its precursor: a stage where shared frames erode, narratives harden into tribal certainties, and institutions lose the capacity to mediate disagreement. By the time a dispute becomes visible, it has usually been structurally illegible.

TACITUS | Making Conflict Legible to Machines

LLMs cannot reason about conflict—they lack temporality, causality, and provenance. TACITUS transforms scattered conflict traces into a structured knowledge graph that AI systems can actually query and reason over.

Conflict Legibility Infrastructure for AITACITUS

TACITUS ◳ extends this analytical mission into the domain of technology.

Named for the Roman historian who understood that power is exercised as much through what remains unsaid as through what is declared, Tacitus is our conflict-intelligence and resolution technologies hub. It applies AI not to arbitrate truth or manufacture consensus, but to restore visibility into how conflicts form, escalate, and calcify, across politics, diplomacy, institutions, and digital publics.

The premise is simple and, we believe, urgent: before you can resolve a conflict, you must be able to see it. Not as polemic. Not as mood. As structure.

Why Graphs, Not Averages

Most large language models process conflict the way a blender processes fruit: the output is smooth, but the structure is gone. Ask a conventional LLM to analyze a polarized debate and it will produce a balanced summary that no participant would recognize as their own position. It averages semantically. It compresses competing claims into a single coherent narrative. It resolves tension by dissolving it.

This is precisely what conflict analysis cannot afford to do.

Real disagreements are not clouds of sentiment waiting to be averaged into a midpoint. They are structures: networks of actors, claims, interests, moral frames, institutional constraints, and asymmetries of power and information. A territorial dispute between two nations is not a difference of opinion; it is a relational system in which geography, history, legal precedent, domestic politics, and third-party interests form a web of mutual constraint. A corporate governance crisis is not a communication failure; it is a collision of incentive structures, fiduciary obligations, and competing narratives about what the organization owes and to whom.

Tacitus uses graph-based modeling because graphs preserve what summaries destroy: the relations between positions, the distances between actors, the asymmetries that determine who has leverage and who does not. Where a traditional LLM might tell you that "both sides have legitimate concerns," a graph tells you which concerns are structurally incompatible, where interests unexpectedly overlap, and who occupies the bridging positions that make resolution possible.

In technical terms, Tacitus constructs conflict knowledge graphs: structured representations in which actors, claims, interests, constraints, and relationships are modeled as nodes and edges rather than compressed into prose. These graphs can be queried, compared, stress-tested, and evolved over time as new information enters the system. They make disagreement navigable without making it disappear.

This is the difference between a map and a postcard. Both depict the same territory. Only one helps you find your way.

The Platform

Tacitus is a multi-agent conflict intelligence platform. It turns information asymmetry into structured evidence and common ground for boards, teams, mediators, and institutions. Its architecture rests on three core capabilities:

1. Making Disagreement Legible Tacitus uses multimodal AI agents to extract claims, frames, and incentives from unstructured text, including policy documents, media coverage, internal communications, legal filings, and public discourse, and translates them into structured, queryable representations. This reflects Concordia Discors' founding premise: pluralism survives only where disagreement remains intelligible. When disagreement becomes opaque, coexistence breaks down.

2. Treating Polarization as a System, Not a Mood Polarization is not an emotion. It is a patterned phenomenon with a discernible grammar: moral framing, identity signaling, strategic positioning, narrative divergence, and the systematic inflation of threat perception. Tacitus models these elements as interrelated components of a system, enabling them to be decomposed, compared, and stress-tested. The goal is reframing and de-escalation without erasing principled difference, because the point of pluralism is not to eliminate disagreement but to keep it from becoming destruction.

3. Mapping Conflict for Resolution Through a growing conflict ontology and graph-based modeling, Tacitus represents disputes as relational systems of actors, interests, constraints, and asymmetries. This allows practitioners to identify leverage points, veto players, and bridge actors: the positions where intervention can still shift trajectories toward resolution before positions calcify into permanent enmity.

The Apps

Two public-facing applications demonstrate the Tacitus approach in action:

🌬️ Wind Tunnel — windtunnel.tacitus.me

Stress-Test Any Idea Against Society

Wind Tunnel is a simulation environment that subjects proposals, policies, and ideas to structured adversarial analysis. Rather than asking "is this a good idea?", a question that invites confirmation bias and semantic averaging, Wind Tunnel asks: how does this idea behave under pressure?

Users submit a proposal, and Tacitus deploys multiple AI agents representing distinct societal perspectives, institutional interests, and value systems. These agents do not "debate" in the conventional sense; they stress-test the proposal from structurally different positions, surfacing vulnerabilities, unintended consequences, coalition fractures, and implementation risks that a single-perspective analysis would miss.

The output is not a verdict. It is a map of the pressure points: where the idea holds, where it cracks, and where adaptation could strengthen it. Wind Tunnel embodies the concordia discors conviction that ideas are improved by structured opposition, not by applause or suppression.

Think of it as a democratic wind tunnel for policy: the same principle that engineers use to test whether a structure can withstand real-world forces, applied to the social and political environment in which ideas must survive.

🧭 Conflict Compass — compass.tacitus.me

AI-Powered Conflict Intelligence

Conflict Compass is the analytical core of the Tacitus platform. It takes a conflict, whether geopolitical, institutional, or civic, and builds a structured, graph-based intelligence picture: who the actors are, what they claim, what they actually want, where their interests collide, where they secretly converge, and what structural constraints shape the space of possible outcomes.

Conflict Compass does not offer opinions. It offers legibility. It transforms the fog of competing narratives into a navigable architecture, allowing mediators, analysts, boards, and decision-makers to see the conflict as a system rather than experiencing it as noise. It identifies:

• Narrative divergence: where the same facts are framed in structurally incompatible ways
• Interest overlap: where opposing parties share objectives they may not recognize
• Leverage points: where small interventions can shift large dynamics
• Veto players: whose opposition can block resolution regardless of broader agreement
• Bridge actors: who occupies positions of trust across dividing lines

The result is conflict intelligence that respects complexity rather than smoothing it away. Compass does not tell you what to think. It shows you what the conflict actually looks like when you stop averaging and start mapping.

AI in the Service of Concordia

The name Concordia Discors, discordant harmony, contains within it a theory of peace. Harmony does not require agreement. It requires structure: a common space of understanding within which disagreement can persist without becoming annihilation. The great pluralist thinkers, from Isaiah Berlin to Raymond Aron to Leszek Kołakowski, understood that the dream of final consensus is not the friend of liberty but its most seductive enemy. The task is not to resolve all disagreement but to build the institutional and epistemic conditions under which disagreement can endure without becoming destructive.

AI, properly constrained, can strengthen this project. By making conflict analyzable rather than merely reactive, and by enabling earlier, more disciplined forms of mediation, it can extend the reach of human judgment without replacing it. The danger of AI in the conflict space is the same danger that haunts every technology of legibility: the temptation to optimize for resolution at the cost of pluralism, to treat disagreement as a bug rather than a feature of free societies. Tacitus is built against that temptation.

Tacitus is not an authority. It is an instrument, designed to help preserve the conditions under which disagreement can endure without becoming destructive. In a world where the loudest voices and the smoothest algorithms conspire to flatten complexity into tribal simplicity, the preservation of structured disagreement is itself a form of resistance.

>> Tacitus.me ◳ | Wind Tunnel 🌬️ | Conflict Compass 🧭