Intelligence Methodology

What Darchow is

Darchow is a real-time geopolitical intelligence platform that synthesizes open-source data from 9 signal types into mechanically graded intelligence. Every inference card is produced by algorithmic fusion — no human analysts in the loop. The platform monitors 5 conflict theatres across the Middle East, Ukraine-Russia, East Asia, South Asia, and the Sahel.

Data sources

“Raw only” means data is ingested and stored but does not currently contribute to scoring or card generation. These sources are retained for future research and correlation analysis when higher-resolution data becomes available.

Grading system

Darchow's grading system is inspired by NATO STANAG 2511 (the Admiralty Code), adapted for automated open-source intelligence fusion. Each inference card receives a grade reflecting how well the underlying event is corroborated across independent sources and sensor domains.

Source uniqueness is measured by publishing domain or channel, not by event count. Three posts from the same Telegram channel count as one source. Three RSS articles from the same domain count as one source.

Physical proof gate

Only kinetic sensors (FIRMS satellite thermal with high confidence, USGS seismic) can set physical proof. Internet outage and traffic anomaly data cannot — these are digital signals, not physical confirmation of a kinetic event.

Evidence URLs

Grade A cards include direct links to NASA Worldview satellite imagery at the event coordinates, showing thermal detection and true-color views before and after the event.

Darchow Index

The Darchow Index is a composite escalation score (0–100) computed from weighted open-source signals. It measures assessed current intensity — not a prediction of future events.

Signal weights

The index uses relative weights that automatically normalize to 100%. When a signal returns no data for a region (e.g., no Polymarket prediction markets exist for the Sahel), that signal's weight is excluded from the denominator. The index is computed only from signals that have actual data.

Signals with zero weight (data collected but not scored): FIRMS thermal, internet anomaly (IODA + Cloudflare), ACLED fatalities, seismic.

Smoothing

The published index uses asymmetric exponential moving average smoothing. Escalation is tracked quickly (α=0.30, ~2-day half-life). De-escalation is smoothed slowly (α=0.07, ~10-day half-life). This prevents a single quiet news cycle from collapsing the index during an active conflict.

Level mapping

Hysteresis

Escalation moves fast and de-escalates slowly. When signals cross a level boundary upward, the new level confirms immediately. Downward movement requires 16 consecutive compute cycles below the threshold before the level drops — roughly four hours at 15-minute intervals. Level 5 has an additional gate: at least one event in the region must carry physical proof from a kinetic sensor. Manual overrides can bypass this gate but are labeled transparently in every API response.

Minimum escalation floor

Regions with confirmed active armed conflict have a minimum escalation level set by the platform operator. This prevents algorithmic de-escalation during overnight news lulls in active war zones.

Regional framework

Darchow's regional boundaries are anchored to the United Nations M49 geoscheme, with two documented deviations for conflict-theatre coherence.

M49 deviations

*Chad (M49: Middle Africa, 017) is placed in the Sahel region for Lake Chad Basin conflict continuity. *Sudan (M49: Northern Africa, 015) is placed in the Sahel region for active civil war spillover into Chad and Niger.

Data quality and transparency

Darchow reports what it knows and what it doesn't. Every API response includes a data quality block with confidence level, active signal coverage percentage, and explicit warnings about unavailable data sources.

Known limitations

The Darchow Index measures current assessed intensity. It is not predictive. Leading indicators (NOTAM airspace closures, GPS jamming, embassy drawdowns) are planned but not yet implemented.

Cross-region article leakage exists for articles that legitimately reference multiple conflict theatres. A single article covering both Iranian and Russian military cooperation will appear in both region feeds.

Telegram geocoding covers approximately 21% of messages. The remaining ~79% are commentary without geocodable location references — these contribute to velocity signals but not to map positioning.

ACLED conflict event data is configured but inactive pending API access. Its escalation weight is zeroed.

API and MCP

Darchow exposes structured intelligence through a REST API and an MCP (Model Context Protocol) server. The API provides JSON responses with full grading metadata, data quality transparency, and evidence URLs. The MCP server enables AI agents to consume geopolitical context directly.

API documentation →