World Observer collects only passive, publicly observable signals. Observers do not initiate network connections, scan hosts, or probe services. Instead, they rely on data that is already visible to the public or produced by routine, non-invasive observation pipelines.
Long-term stability requires repeatable measurement. The project emphasizes consistent cadence and unchanged semantics so that comparisons across months or years remain meaningful. Repetition reduces noise, helps identify structural changes, and supports cautious, evidence-based interpretation.
- No active verification: Passive observation can miss issues that require direct probing to confirm.
- Signal ambiguity: Public signals can be delayed, incomplete, or biased by external factors.
- Ethical constraints: The project prioritizes non-interference, which constrains what can be measured.
These limits are intentional. The goal is to establish a durable, ethically sound baseline rather than to maximize coverage through aggressive techniques.
The dns-tta-stress-index observer estimates DNS stress using only aggregate
signals from minimal A/AAAA probe timing outcomes.
Higher stress indicates slower and less stable DNS answers relative to local historical baselines, combining:
- elevated p95 response latency,
- increased timeout rate,
- reduced success rate,
- increased latency jitter.
A chart is generated only when significance triggers fire (z excursion,
hard timeout threshold breach, or same-day multi-country mass event). This keeps
visual artifacts focused on unusual days, not daily noise.
Tracked outputs exclude hostnames, resolver identifiers, IP addresses, and per-query rows. Local raw samples are stored outside tracked outputs and are excluded from Git.