Shapiro, Lifschitz & Schram

Litigants have been wrestling with the cost implications of managing the collection, review and production of the ever increasing amounts of electronic data ever since the 2006 changes to the Federal Rules of Civil Procedure, which officially made all electronically-stored information (“ESI”) discoverable in US courts.

Today’s typical E-discovery undertaking involves culling client electronically stored information (“ESI”) from the appropriate data-stores, then filtering it according to file types, date ranges and keywords, after which attorneys conduct a linear, eyes-on documents, review.  This linear approach to reviewing electronic production on a document-by-document basis has been the generally accepted method for years, and works well in many varieties of litigation.  However, as clients’ ESI and data volume continue to grow in both size and complexity, this kind of review (on average the most costly and time-consuming aspect of the discovery phase) is no longer practicable or sustainable.  This is the case especially in light of today’s high original data volumes and low proportions of relevant information. 

Attorneys and E-Vendors have been developing search methods that focus on reigning in the increasing costs associated with electronic document review – computer assisted electronic document review.  Over recent years, these strategies have had evolving monikers but they are commonly named “predictive coding”, “machine learning”, “technology-assisted review”, or “analytics”.  These cost reducing technology-driven strategies refer to computer applications that receive input such as coding decisions from human attorneys for a certain subset of documents, and then use that input to help categorize, ‘predictively’ code, or rank the remaining documents in the set.  Predictive coding essentially uses a learning process comprised of a mathematical algorithm to find identical or similar issues or discussion within the rest of the document population.

Attorneys and document reviewers employing such mechanisms on behalf of their clients are able to identify representative document types and then have the software analyze the entire date collection in order to find instances of conceptually related documents (i.e., “more like this”).  Other aspects of technology-assisted review boast being able to:
-    automatically organize entire document collections;
-    automatically prioritize “key” or “hot” documents;
-    automatically categorize documents by key phrases, concepts and names;
-    automatically determine the relevance, responsiveness, and privilege status of any document;
-    locate all similar or related documents irrespective of keywords; and
-    algorithmically reduce the number of documents requiring eyes-on review, which therefore reduced billable hours.

It should be kept in mind that computer assisted electronic document review is just another tool and if used improperly could result in increased costs and headache.  But, if used competently, it can expand the decisions made according to a small sub-set of documents to a wider population of documents and considerably increase efficiency.  It is gaining popularity in litigation involving large amounts of ESI because it can be a powerful document review tool that has the potential to increase efficiency and decrease costs.  The time and expense of predictive coding processes can be a fraction of that of a full manual review, but potential users should always proceed with caution and with good advice from counsel that has significant experience or specialty in this area.