Methodology
How the Observatory models digital attention — and what these numbers are, and are not.
Demonstration Data Every measurement currently shown is a seeded demonstration value, not verified live collection.
The canonical model
Attention Observatory → Z-Objects → Z-Facts → Z-Factors → Z-Rank → Intelligence Products. A Z-Fact is a verified, time-stamped observation with a source. A Z-Factor is a pattern formed from multiple Z-Facts. Z-Rank measures a Z-Factor's current significance.
Z-Rank is significance, not popularity
Z-Rank blends attention with breadth, durability, and evidence, and subtracts a hype penalty so that loud-but-shallow spikes do not dominate. Scoring is isolated from presentation and is fully replaceable.
| Component | Weight |
|---|---|
| attention Velocity | 0.22 |
| source Breadth | 0.12 |
| organization Breadth | 0.12 |
| platform Breadth | 0.08 |
| research Activity | 0.1 |
| developer Activity | 0.1 |
| adoption Evidence | 0.1 |
| persistence | 0.08 |
| novelty | 0.08 |
| hype Penalty (subtracted) | 0.25 |
Signal lifecycle
Weak Signal → Emerging → Accelerating → Breakthrough → Mainstream → Saturated → Declining → Re-emerging.
Data strategy (progressive)
- Phase 1 (now): curated source registry, seeded Z-Objects/Facts/Factors, demonstration scores, functional public pages.
- Phase 2: automated headline collection, entity extraction, deduplication, signal clustering, historical scoring.
- Phase 3: multi-source measurement, automated Z-Rank, attention-flow detection, public API.
Labels
Values are labelled by valuation: seeded, estimated, simulated, or verified. Nothing on this site is currently verified live data. Historical records are preserved, not overwritten.