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.
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.