Search queries fail silently in eDiscovery, returning entire document sets and hiding relevance

Summary

A litigation support professional runs an apparently valid query against a 65,102-file SharePoint corpus expecting a handful of hits, and the platform returns all 65,102 files with no error, no warning, and no indication that filtering did not occur. The failure is silent: the user only notices because they manually spot-checked, meaning in ordinary practice the miss would go undetected. This is the central anxiety of the whole discipline — when a search reliably cannot be trusted, you cannot know whether the smoking-gun email exists, whether you have over-collected and inflated review costs by orders of magnitude, or whether your pre-hearing preparation rests on an incomplete record. No mainstream eDiscovery or case-intelligence tool offers query validation, result-set sanity checks, or an explicit 'this query matched nothing / matched everything' audit trail. As disclosure arrives in tranches and chronologies must be rebuilt repeatedly, every untrusted search compounds the cost. This is directly aligned with the pain of finding a message read weeks ago across hundreds of emails and with the need for source-anchored, verifiable answers.