Last updated: 20 October 2026
GridResilience is built by a named researcher, on a documented method. Every paid response is timestamped and carries the version of the scoring rules it was produced with, so a number can be traced back to the rules and the data window behind it. This page exists so that a buyer, an auditor or a compliance reviewer can check those claims independently — not take them on trust.
Dr Parádi József, CSc — retired border-guard colonel, retired college professor; candidate of military science and member of the public body of the Hungarian Academy of Sciences; founder and president of the Szemere Bertalan Hungarian Law Enforcement History Scientific Society (SZBMRTT), a learned society founded in 1990.
The Society is a scientific body, not the vendor. GridResilience is operated by Dr Parádi József as a sole proprietor; the Society's institutional record is cited here as the methodological lineage, in the same way that a research group cites the laboratory it came from.
The pipeline is deliberately boring, because boring is auditable:
| Stage | What happens |
|---|---|
| Collection | 140 open RSS sources in 9 languages, refreshed every 5 minutes. |
| Scoring | Goldstein-scale weighting per matched keyword domain; article stress normalised to a 0–4 scale. |
| Aggregation | Volume-weighted stress per country and per region; thematic sub-indexes; source-language mix. |
| Critical infrastructure | KIVP grid criticality: attack risk, recovery risk, defence priority and redundancy ratio, normalised across covered networks. |
| Scenarios | A fixed logistic model over stress, chokepoint exposure, event triggers and 12-hour news momentum. The weights are constants in the source; identical inputs always produce identical outputs. |
| Abstraction | No article text, links or channel names leave the system. Outputs are numbers and categories. |
The product is built on information gathering and structuring, governed by a logic that follows a military and border-security methodology. That logic concentrates on prevention and on the identification of so-called decoy phenomena. Decoy logic recognises events that are real but display attention-diverting and capacity-draining patterns, and reads such patterns as an assessment of the balance of forces and preparation for the execution of the real objective.
Outputs are model estimates from a baseline that is not yet calibrated. They are not forecasts, not investment advice, and not a substitute for a compliance decision. Every escalation response carries its own calibration statement and the number of signals it was built from. If a number is uncertain, the response says so.
Every response is timestamped and versioned: the body carries generated_at, and the
country, chokepoint and thematic aggregates carry the stress_version of the scoring rules
(stress_rules_v1.3). The scorer is deterministic and keyword-based — the same inputs and
the same rules version produce the same numbers, with no model in the loop.
Every published data snapshot — the corpus, the daily brief and the imagery index — is hashed with SHA-256. Each hourly manifest carries the SHA-256 of the previous manifest, so the manifests form a chain, and the chain head is signed with the operator's GPG key:
| Signer | Dr Parádi József CSc, operator of GridResilience |
|---|---|
| Key ID | 2E0844B46B3AC477 |
| Fingerprint | 512A 10A6 83CD DAFA A00C BECF 2E08 44B4 6B3A C477 |
| Published | keys.openpgp.org |
To check: recompute the SHA-256 of the files listed in the manifest, verify the detached signature
with gpg --verify, and confirm that each manifest's prev_manifest_sha256 matches
the previous entry in the chain. The instructions and the live manifest are at
/verify. A signature proves who published what; it does not make the
numbers correct — that is what the validation page is for.
All inputs are publicly available open-source reporting. We do not use privileged or classified sources,
and we say so in every assessment. Coverage is denser for Europe, MENA and Eurasia than elsewhere; the
data_sources and warnings fields of each response state which components were
available for that particular query.
| Input | Used for | Licence |
|---|---|---|
| 140 public RSS feeds, 9 languages | stress score, brief | public feeds; no article text is redistributed |
| OpenStreetMap (grid & road geometry) | network vulnerability | © OpenStreetMap contributors, ODbL 1.0 |
| ENTSO-E, via PyPSA-Eur | grid topology | ENTSO-E Transparency Platform terms |
| Copernicus DEM (GLO-90) | terrain → attack feasibility | Copernicus: free, commercial use, attribution |
| NASA GIBS (VIIRS/MODIS) | daily imagery | NASA, public domain |
| NASA FIRMS (thermal anomalies) | fire/flare context, aggregated | NASA open data |
FRED DCOILBRENTEU | backtest target (Brent) | US Government, public domain |
| OFAC SDN, EU FSD, UK sanctions lists | entity mentions only | public; list itself not redistributed |
We do not use AIS vessel data, and no paid or classified source is involved. Only derived numbers and our own paraphrased brief leave the system.
We take no payment from any government, political party or covered entity, and we do not accept sponsored coverage. Revenue comes only from subscriptions and API usage.