FORDHAM FORENSICS When It Really Matters770-777-2090
Protecting Privilege from Inadvertent Disclosure in E-Discovery
Document reviews in any litigation can be onerous, perilous and expensive. Privilege reviews are particularly troublesome. While other reviews can correct omissions over a longer review period, the window for privilege review is typically much shorter.
Electronic techniques for locating privileged documents are not infallible. For this reason, some advocate manual reviews for privilege.
Manual efforts, however, are also known to have limitations. In fact, statisticians for years have demonstrated how sampling can better predict the existence of a characteristic in a population than complete manual inspection.
In cases involving e-discovery the decision whether to perform a manual review or an electronic review can easily be decided by the volume of documents. This fact is recognized in by the Sedona Conference in Practice Point 1 of its publication on search and information retrieval methods.
While many cases like Rhoads Industries, Inc. v Building Materials Corp. of America, 254 F.R.D. 216 (E.D. Pa. 2008) illustrate the consequence of poorly implemented electronic search methods for privilege reviews, the new FRE Rule 502 adopts the standard that the reasonableness of the search plan will determine whether privilege was waived by an inadvertent disclosure.
In the digital age the most reasonable approach is one that integrates computerized search techniques with statistical sampling prior to manual review.
Under such an approach both computerized search techniques and statistical sampling are used to reduce the volume of data subjected to human review. In addition, the sampling is used to validate the effectiveness of the electronic search methods in identifying privileged documents.
For the computerized search effort every document including e-mail attachments should be subjected to the search terms. Thus, not only are the senders and recipients of messages searched but the document body, attachments and metadata as well.
With regards to the search terms themselves they would be comprised of the attorney’s full names, last names, nick names and e-mail addresses. In addition, the firm names and domain names could be used.
In order to catch instances where individuals have communicated the advice of counsel to third parties the search terms should include “attorney”, “lawyer” and “counsel”.
Finally, in order to identify still other variations of communications about the case, some linguistic analysis should be done with client personnel to determine how the matter is commonly referenced by client personnel.
For those documents having the computerized search term hits, they would be subjected to several layers of analytical testing and human review. For example, the first cut could be to review a list of document attributes which could easily confirm that these are likely privileged documents. Manual review would then be performed to both confirm the privileged nature of the document and to collect data necessary to complete the necessary privilege log information.
One of the initial analytical steps that could be taken would be a review of document names, the number of different search terms found within a document and the type of search terms. For example, a law firm domain name or lawyer e-mail address could be better predictors than simply an attorney’s last name.
Another analytical step that could be taken would be to consider the type of document when making a privilege document type determination. For example, an e-mail document would likely be a better predictor than the OCR’d text from a company promotional piece identifying a contact person with the same last name as one of the attorneys.
For those documents not containing computerized search term hits, discovery sampling techniques could be employed to confirm their lack of privileged information. Discovery sampling is a technique often used to verify the accuracy of another operation. Thus, it would be used here to verify the effectiveness of the computerized search terms in identifying privileged documents. (See http://www. answers.com/topic/discovery-sampling)
In the event that privileged documents are found while performing the discovery sampling, those documents can be examined to determine whether the computerized search criteria should be revised and run again. In the meantime, those having passed the manual review can be cleared and removed from further consideration in subsequent iterations of computerized searches and/or discovery samples.
Clearly, there is no need to manually review every document for privilege. Furthermore, a complete manual review is not assured of having better accuracy. So, perhaps the best and most economic approach is one that implements electronic searches, statistics and limited manual review.