My take on what you really need in eDiscovery technology:
Data Mapping -- I have written about this before in "Mapping: The Basis of Information Management" . Data mapping ferrets out data throughout the storage infrastructure. It should be dynamically updated, searchable and actionable. Examples of eDiscovery usage include collection (obviously) and an efficient way to carry out early case assessment.
Visual Analytics -- Visual analytics are just one tool in the review cycle but it's an important one for human reviewers. These screens enable teams to quickly grasp potentially relevant data and relationships, allowing them to make sense of presented information without diving down one too many rabbit holes. Visual analytics is primarily a tool for ECA although it can be used as a component of a full-blown review and analysis platform as well.
Culling the dataset -- Reducing the size of the collected dataset before processing is a widely accepted procedure. Traditional de-duplication is the big daddy process but so is near de-dup. Prioritized search results that let reviewers dump low-relevance results are also useful. (Caveat: be sure to pick a technology that reports what you dumped and why you dumped it.)