Methods and evidence limits
Understand observations, inferences, unknowns and inapplicable checks, and what one run can establish.
Updated · Egress Observatory
Evidence before conclusions
An observation is a response from a named source at a stated time. An inference applies site rules to those observations. Neither should be confused with an official platform score.
Unknown means insufficient evidence, including missing integration, timeouts, permission limits or absent fields. Inapplicable means the check does not apply to this protocol or scenario. Neither means passed.
One run and one measurement point
HTTP, DNS, WebRTC and remote probes can take different paths. Compare observation time, measurement point and source coverage first. Starting a new run does not turn cached data into a fresh observation.
Save A, change one setting, start a fresh run and save B. Comparison explains observed differences; unmeasured items remain unknown.
Common misinterpretations
A high score does not guarantee account access. Different addresses are not necessarily leaks. A negative result from one list does not mean no list has a record. These checks do not predict approval or bans.
Traffic-intensive checks require an explicit start. Review report contents before sharing; local exports can contain full addresses, while public reports retain only allowed summary fields.