Method
How we measure the world's pulse. And why you should know.
Important notice: The indicators published on The Dark Mirror are estimates generated by artificial intelligence based on a sample of news. They are not scientific measurements, hold no certified statistical value and must not be used as a basis for political, financial or other decisions. The project is a work of data-art and a sociological experiment, not a research institute.
The Pipeline
Every night, an automated process runs these steps:
News collection
A client queries GNews API for the top 10 stories in 4 languages (English, Italian, German, French). About 30 articles are collected, deduplicated and categorized by keyword.
Dark Analysis
Headlines and descriptions are sent to Google Gemini 3 Flash with a clinical prompt asking for 7 numeric indicators (0-100), a dominant emotion, a summary sentence and a comment for each indicator.
The day's tale
Independently of the world analysis, the system fetches the day's tale from a separate, protected archive. The two data flows never contaminate each other.
Pulse computation
A composite index (0-100) is computed as a weighted average of the 7 indicators. The formula is public and documented below.
Publication
Data is saved in an SQLite database and exported as a static JSON file. The website reads it without ever querying the database directly.
The Pulse Formula
The Pulse is a composite index that synthesizes the 7 indicators into a single number. Higher = world in better shape. The formula:
Hope × 0.25 + Human Rights × 0.20 + Social Cohesion × 0.20 + Environment × 0.15
Negative (high = worse):
Global Tension × 0.25 + Economic Pressure × 0.15
Neutral:
Tech Disruption — not included (can be positive or negative)
─────────────────
Pulse = Positive − Negative + 40 (normalized to 0-100)
The weights reflect an editorial choice, not a scientific methodology. Hope and tension weigh more (0.25) because they represent the primary forces. The environment weighs less (0.15) because environmental change is less volatile day to day.
Sources and known bias
The quality of the indicators depends on the quality of the input news. Here is what we cover and what we don't:
| Geographic area | Coverage | Notes |
|---|---|---|
| North America | Covered (EN) | Dominant English-language sources |
| Western Europe | Covered (EN, IT, DE, FR) | Well represented |
| Eastern Europe / Russia | Partial | Only through Western sources |
| Middle East | Partial | Filtered by Western media, no Arabic sources |
| China / East Asia | Partial | No Mandarin, Japanese or Korean sources |
| Africa | Minimal | Only news reaching Western media |
| Latin America | Minimal | No Spanish or Portuguese sources |
| Southeast Asia / India | Partial | English sources only |
Consequence: the indicators reflect a predominantly Western view of the world. Significant events in uncovered regions may not affect the numbers.
AI Model
Primary model: Google Gemini 3 Flash (Preview)
Fallback: Google Gemini 3.1 Pro (Preview)
The model receives the headlines and descriptions with a prompt that demands a cold, clinical and objective analysis. The prompt is fixed and does not change day to day.
Known model limits:
- It may be systematically more or less dramatic than reality
- It has training biases reflecting the data it was trained on
- It has no access to confidential or classified information
- It may produce different results for the same input (non-deterministic)
- It does not verify the news — if a source reports false information, the analysis includes it
What we are not
The Dark Mirror is not:
- A research institute or scientific observatory
- A registered news outlet
- An early-warning system for international crises
- A financial index or market indicator
- A substitute for human judgment or expert analysis
The Dark Mirror is:
- A work of data-art
- A sociological experiment
- A father's educational legacy for his son
- A daily archive of humanity and imagination
Reproducibility and open data
The raw data for each day is available through the public APIs:
GET /api/today.php— today's dataGET /api/today.php?date=YYYY-MM-DD— archive by dateGET /api/history.php?days=N— last N daysGET /feed.php— RSS feed
The data is released under CC BY-NC 4.0. You may use it for research, analysis and non-commercial projects citing the source.