On May 6, 2013, the U.S. Department of Justice’s (“DOJ”) Antitrust Division approved Constellation Brands Inc.’s (“Constellation”) and Crown Imports LLC’s (“Crown”) request to use predictive coding to determine which documents were most relevant and responsive to the DOJ’s requests. Constellation is a potential buyer of assets from the huge AB InBev-Grupo Modelo merger, and Crown is a joint venture between Grupo Modelo and Constellation. Reportedly, Constellation and Crown identified in excess of one million documents that would require manual review before being handed over to the Justice Department for scrutiny. After several seed sets were run using the automated data review software and compared manually, DOJ was satisfied that the predictive coding software would identify the most relevant documents and approved its use. As reported by the Wall Street Journal, the predictive coding software used by the parties was developed by kCura Corporation, a software vendor for many entities including DOJ.
DOJ’s approval of the use of predictive coding is another step towards increased usage and acceptance of automated rather than human review of documentary evidence. Historically, a “linear” document review of the magnitude in question in the AB InBev-Grupo Modelo merger investigation would involve retention of numerous contract or temporary attorneys, at a rate ranging anywhere from $25 to $40/hour, to sift through the voluminous documents at issue, leading to millions in legal fees. According to estimates, use of predictive coding software will slash the cost of the document review to Constellation and Crown by at least half.
Although the AB InBev-Grupo Modelo merger investigation is the first reported occasion where predictive coding was approved for use in a DOJ investigation, other government agencies, such as the Federal Trade Commission, have already considered its use on a case-by-case basis. Given the current economic climate, the government’s openness to the use of this new technology signals that requests by other investigated companies to use predictive coding in large scale, document heavy investigations will exponentially increase in an effort to significantly reduce legal fees. The end result: government agencies will be positioned to conduct and complete investigations in far less time and at a far lower cost to parties.
Judicial acceptance of predictive coding in civil litigation has been limited. However, virtually every court that has addressed this technology has found that — at least when based on appropriate, quality controlled protocols — predictive coding is an acceptable alternative to linear document review. (For discussions of other recent cases where courts have permitted the use of predictive coding as an e-discovery tool, see here, and here.) The DOJ’s acceptance of its use is yet another indication that these methods are well on their way to general acceptance. Given the changing landscape of document discovery, it is prudent for every attorney whose client is faced with massive document requests to at least seriously consider whether the case might benefit from automated review and, if so, seek the Court’s permission to use predictive coding software to ease the burden of large-scale document review and production.