In the US at least, arguments about the use and acceptance of technology-assisted review (TAR) seem to be moving from whether this technology is appropriate to how it is to be used.
The problem with this, or one of them, is that the the debates quickly (and understandably) go over the heads of the lawyers, as technical terms multiply and as apparently conflicting opinions come from different courts in different cases.
David Wallack of NightOwl Discovery has written a paper focusing on one particular difficulty, the disclosure of a seed set’s contents and the inclusion of irrelevant documents while building the sets.
The paper is an easy read compared with some of those which are around. It includes links to other papers for those who want to read further.