Specification document for methodology for advanced heuristics and mitigating faulty measurements

[jbonisteel]

This is an epic for the anon credentials project

Due Date: March 31

We will develop a set of (more advanced) heuristics to detect measurements which present inconsistencies (e.g. measurements submitted from a location different than that of the probe) that can be a sign of faulty data.

Output: specification document outlining methodology for more advanced heuristics.

Based on the previous heuristics for detecting if OONI measurements are inconsistent, we will create a set of strategies for what to do when we have determined that a measurement is “bad” (by some definition of “bad”).

These strategies might involve (in the most “light” case) annotating relevant measurements with a flag indicating that it should not be trusted, while in more extreme cases, we might even decide to reject the submission altogether to prevent storing bad data at all.

Output: specification document outlining the strategies for mitigating faulty measurements.