Treat World Monitor (and any other external source) as a high-quality sensor array.
Your World Signal remains the personal interface, the aesthetic, and the pattern-recognition layer you train.
1. Core Philosophy
World Signal is not a clone of any large dashboard.
It is a personal, high-signal, provenance-first situational awareness tool with a calm, slightly luminous presence.
- You control the schema and the meaning
- You decide what becomes an “event”
- You train the pattern recognition on your own judgment
- External sources only supply raw or lightly processed signals
2. Soft Data Import Architecture
Preferred approach: keep the browser reading static (or locally generated) JSON. A small script on your machine does the fetching and normalization.
Flow
External Source → Local Script → Normalized events-live.json + feed-health.json → Your World Signal dashboard
3. Focus Domains
Flights
Primary source: OpenSky Network (free non-commercial tier)
- Category:
aviation
- Focus on anomalies, military activity, unusual routes, or region-specific spikes
- High instrumental confidence when position data is present
- Location derived from lat/lon or ICAO region
Stocks / Markets
Primary sources: Finnhub, Alpha Vantage, FRED (macro)
- Category:
market
- Do not dump every ticker — only surface anomalies, volume spikes, sector moves, fear indicators
- Model-generated interpretations should be clearly marked
- Useful for training on “fear pushes and pumps”
Radio / Frequency
Sources: KiwiSDR network, Priyom / numbers stations logs, Reverse Beacon Network, your own future captures
- Category:
signals or radio
- Events can be frequency activity, numbers stations, unusual HF bursts, new broadcasts
- Location often approximate or propagation-based
- This is the most personal and trainable layer
4. Event Schema Additions
Keep your existing rich schema. Only extend carefully:
category: add "aviation" | "market" | "signals"
Optional extra fields (in tags or a flexible meta object):
• frequency_khz
• callsign / icao24
• percent_change / volume_z
• receiver_id (for radio)
5. Implementation Sequence
1
Flights first
Build a small OpenSky → your schema importer. Write events-live.json.
2
Stocks second
Add anomaly detection (simple z-score or threshold) for a watchlist of symbols/sectors.
3
Radio third
Start with public schedules and logs. Later add your own SDR/KiwiSDR observations.
4
Pattern layer
Begin manually labeling events that feel significant. Store labels locally. This becomes the training data for your personal recognition machine.
6. Aesthetic Guardrails
- Never import another project’s UI, colors, or layout
- Keep larger, more open typography and soft luminosity
- Live data indicators should be quiet (soft pulse, subtle badge)
- All model-generated content remains clearly marked
- Your dashboard must still feel calm and intentional
7. Long-Term Vision
A personal home pattern-recognition instrument.
External world (flights, markets, radio, disasters, conflict) flows in as clean signals.
You observe, label, and train.
Over time the system begins to surface the patterns you have learned to see — not the ones a generic model was trained on.