Live monitor · synthetic feed

Methods

How the index is constructed, what the baseline vs. AI-assisted comparison means, and where the model breaks down.

Demonstration notice. The current release uses structured mock data and illustrative logic. Scores are computed from simulated vectors for demonstration purposes only and are not derived from classified or proprietary intelligence feeds. Future versions may ingest validated open-source corpora.

01 · What the scores represent

Escalation Score

A 0–100 index that measures how quickly an incident could spiral from a localized event into a wider confrontation. It captures decision speed, the chance of misidentifying the actor, the time available for de-escalation, and how tightly allies are drawn in. A higher score means a narrower margin for error.

Economic Spillover Score

A directional estimate (in USD billions) of how much economic value is at risk within roughly 30 days of the incident. It draws on commodity, equities, insurance, and trade-flow signals. It is not a point forecast; rather, it flags which incidents carry outsized market or supply-chain exposure.

Escalation drivers

  • Decision speed

    How fast decision-makers must respond before the window for choice closes.

  • Misattribution risk

    The probability that an actor or intent is wrongly identified, triggering retaliatory moves.

  • Escalation latency

    The remaining time and diplomatic channels available to walk an incident back down.

  • Public signaling

    How much the incident is amplified in public or social media, constraining private bargaining.

  • Allied coupling

    The degree to which an ally is treaty-bound or politically pressured to join the response.

Spillover drivers

  • Energy-market exposure

    Share of global supply chains, LNG flows, or grid capacity that could be disrupted.

  • Trade-flow density

    Volume of goods and capital passing through the choke point in question.

  • Financial-system coupling

    Extent to which credit, FX, or derivative markets propagate shocks across borders.

  • Regional stability multiplier

    Existing political fragility that can magnify a localized event into broader instability.

02 · Score breakdown illustration

Below is a representative decomposition of how the five escalation vectors shift between a human-only baseline and an AI-assisted loop. Values are illustrative.

Decision speedBaseline 45 → AI 78(+33)
Misattribution riskBaseline 55 → AI 88(+33)
Escalation latencyBaseline 40 → AI 65(+25)
Public signalingBaseline 50 → AI 80(+30)
Allied couplingBaseline 35 → AI 55(+20)

03 · Analytical pipeline

Every incident moves through the same four-stage pipeline. Analysts encode the event, score its inherent risk factors, apply a scenario adjustment (e.g., AI assistance, oversight strength), and produce a composite score.

Incident input

Event metadata, actors, domain, and region are recorded.

Risk factors

Five escalation vectors and four spillover drivers are rated.

Scenario adjustment

Detection speed, confidence, oversight, and uncertainty are tuned.

Score output

Baseline and AI-assisted scores are computed and compared.

04 · Scope

ERL studies incidents where AI-enabled systems plausibly altered the decision loop or information environment during a crisis. The corpus spans geopolitical, cyber, and energy domains. Incidents are selected when an AI system—whether recommendation, generative, or autonomous—was present in the escalation chain.

05 · Limitations

The model assumes approximately linear coupling between capability dials and escalation outcomes. It does not yet account for adversary adaptation, treaty effects, or non-rational behavior under stress. Spillover estimates are directional and should not be used as trading or policy triggers without additional expert review.