|This document is available in two formats: this web page (for browsing content) and PDF (comparable to original document formatting). To view the PDF you will need Acrobat Reader, which may be downloaded from the Adobe site.|
| Electronic Discovery at the Antitrust Division: An Update(1) |
It goes without saying that electronic productions have grown exponentially. In 2009, I appeared on a panel addressing electronic discovery issues at the Spring Meeting and remarked that the Division had increased its electronic storage capacity from 12 terabytes in 2003 to 70 terabytes in 2008. Today, we have single investigations that involve over 8 terabytes of data. Recently, the Division has completed several investigations with productions just under or slightly above one million records.
In response to these challenges, the Division continues to review and revise its electronic discovery practices. In 2009, the Division released its civil electronic production letter, which provides a framework for negotiations for electronic productions, in conjunction with the Summation load file specifications. This was a significant step and I believe that the letter has benefited both the Division as well as those responding to the Division's compulsory process. However, there are additional steps that can make producing documents and information to the Division less burdensome and more efficient. Before discussing how the Division is addressing electronic discovery issues, I will briefly address initiatives that are being undertaken by the Department of Justice.
As a result of an internal Department-wide review of discovery practices, the Department established a Civil Electronic Discovery Committee with representatives from all of the litigating components (Antitrust, Civil, Civil Rights, Environmental and Natural Resources Division (“ENRD”), and Tax) and the Executive Office of the United States Attorneys (“EOUSA”). The committee has led to an unprecedented level of communication among the litigating components and enhanced all the components’ ability to address the challenges presented by electronic discovery. This review also led to the creation of my position in the Antitrust Division in order to focus on best practices and address novel questions as they arise. Similar positions have been created in the Civil Division and ENRD. As you might expect this group has devoted substantial effort to training. The National Advocacy Center, in conjunction with the Department’s Civil Electronic Discovery Committee, has conducted numerous training courses for lawyers, paralegals, and litigation support personnel throughout the Department.
The Office of the Deputy Attorney General and EOUSA have also been working on criminal electronic discovery issues. They too have focused on training initiatives for agents, lawyers, and paralegals. These training initiatives have focused on meeting the discovery obligations of criminal prosecutors in the age of electronic communications and information.
Antitrust Division Initiatives
The Division has had an internal electronic discovery working group since 2006. There are representatives from every litigating section and field office as well as the active participation of our Litigation Support Group. The group is a forum for comparing notes, discussing new issues and problems, and developing possible best practices. I am extremely pleased at how we have improved our processing and review of electronic documents and information and believe that we are working well with private parties to obtain the needed information while at the same time avoiding unnecessary burdens. As always, challenges remain. To that end, I want to highlight some issues that remain relevant before addressing newer challenges that both the Division and private bar face.
In 2009, the Division posted a sample electronic production letter for civil investigations for the first time along with a set of Summation load specifications. Quite recently, we removed the Summation load specification because -- I am very happy to report -- the Division now has access to other tools and platforms in addition to Summation. We are in the process of revising these load specifications to be more platform neutral. When the Division posted its electronic production letter and Summation load specifications, the documents were accompanied by explicit warnings that the documents were not to be relied upon without discussions with the staff handling the particular matter. Despite these warnings, too many counsel or vendors used these documents without communicating with investigative staff. Without these discussions, for example, counsel and vendors risk creating load files based on outdated specifications or that fail to include information or data that is necessary for a particular investigation. In light of these issues, the Division has not determined whether to repost them. It is critical for a party who receives compulsory process from the Division to ask staff for the current production specifications that will be used in your particular investigation. These specifications are subject to constant revision and can vary depending upon the investigation and/or the information sought by the Division. Ensuring that you have the current specifications will avoid any unnecessary reprocessing costs and delays.
The Division has emphasized to its staff that technical assistance from our Litigation Support staff is critical to ensure that documents and information are produced in the proper format to the Division.We believe the same thing is true for the producing party. We continue to be surprised that lawyers who are producing documents and information to the Division do not routinely include their electronic production vendors in discussions with the Division and its technical staff. We strongly encourage counsel to include vendors in these calls.
As a more general matter, we believe that early and continuing communication between the Division staff and the producing party, which must include technical staff on both sides, is a critical component of a successful electronic discovery negotiation. This communication should occur as early as possible and certainly before the producing party begins collecting information for production. If the collection process does not, for example, include all the metadata requested by the Division, your client might have to incur unnecessary costs. In addition, as I will discuss in more detail below, the Division now has access to a number of litigation support tools, so relying on “what I did last time” is no longer a safe approach.
The Division fully embraces the Sedona Conference Cooperation Proclamation.(2) Given the size of most Antitrust investigations, it is simply not possible to negotiate and produce an electronic production without significant cooperation and communication between the Division and the producing party. The Division is keenly aware that there is no “perfect” electronic production. There are, however, some steps that can be taken to reduce the risks and minimize the costs associated with problems and errors in electronic productions.
The most simple, straightforward means of avoiding costly production errors is to provide the Division with a sample of the type of production you intend to make. The Division makes every effort to carefully review and load sample productions to identify field or format issues or errors that would be costly to correct after a load file or entire hard drive has been produced. To be most useful, the sample must be derived from the producing party’s documents and information. We also encourage vendors to ask any questions or raise any issues that they believe may create errors or problems in a production.
Reducing unnecessary burden on the companies and individuals that receive compulsory process is a priority for Division staff. Although the reality is that antitrust investigations, by their nature, often require the production of large quantities of documents and information, we have no more interest in, or use for, unnecessary or irrelevant information. Of course, the problem remains determining what is necessary (vs. unnecessary) and relevant (vs. irrelevant). There can, of course, be honest disagreements between the staff and parties on these issues. Staffs are keenly aware of the limited resources and, in some cases, time to conduct our investigations. It is counterproductive to request information that we do not have the time nor space to load or review. This is a point of emphasis in training conducted by both the Department and the Division. Therefore, we encourage you to propose ways to reduce the size of productions. However, if you do make a novel proposal, you should be prepared to explain the rationale behind your proposal and answer questions about the impact it will have on the collection produced to the Division. In particular, you should be prepared to ensure that complete information about all the persons within your organization that had custody of responsive documents or information remains intact.
Deduplication is a good example of how the Division responds to new ways to reduce the burden of productions. The Division began allowing parties to deduplicate their productions as early as 2005. After careful negotiations at the beginning, we created standard guidance that is widely used by the Division’s civil litigating sections. (It is important to remember that there may be circumstances in which staff decides that deduplication is inappropriate in any particular investigation. )
We have recently negotiated an agreement to allow a producing party to consolidate email threads and produce a single instance of an email thread, subject to certain restrictions such as custodian information and an agreement to authenticate and produce individual emails, among other things. We are still in the process of refining this procedure and parties should expect to engage in detailed negotiations before obtaining the agreement of the Division to similarly proceed.
In addition to these initiatives, the Division is also revising its Model Second Request to better address electronic discovery issues.
The first change is a revision to Instruction O.9 that would require the producing party to disclose the applications or techniques that are used to assemble the collection of responsive documents and information. The Division believes that the best approach that would permit producing parties to take advantage of these new tools and platforms while ensuring that the Division receives the documents and information necessary to conduct its investigations is to seek a cooperative approach that requires the parties to disclose what processes are being used.
The promised benefit of new tools like predictive coding(3) and concept searching is the notion that only the “really” relevant documents could be produced, instead of all the documents that are technically responsive to the Second Request. While the Division strives to be sensitive to the burden of a Second Request, the use of these tools raises obvious questions. Relevance, of course, is in the eye of the beholder. It is not at all uncommon for an investigative staff and counsel to have significant disagreement about whether a collection of documents are relevant to a particular investigation. Therefore, we believe that if the parties wish to use these platforms and tools to identify “more” relevant documents and information, as opposed to responsive documents and information, the producing party must seek a written modification of the Second Request.
To date, while the Division has discussed using these techniques on several occasions, it has never reached an agreement with a producing party to use one of these platforms or tools. I would think there is a significant opportunity to reduce the amount of material produced in response to the Division using one of these platforms or tools, however, I do believe it would require a great deal of cooperation, transparency, time, and hard work very early in the Second Request process. The first issue would be the means by which the Division could assure itself that the relevant/not relevant designations were consistent with its view of the pertinent issues. I would imagine that this could be accomplished by the producing party providing a statistically significant sample of both relevant and non-relevant documents. The Division would review those sets of documents and raise any concerns. In addition, there would have to be a mechanism by which, if an issue arose that had not been anticipated by the Division at the outset, e.g. , a natural experiment about a critical issue in the investigation, additional information relevant to that issue would be produced. As an initial framework, I believe a plan like this could serve as the basis for a meaningful negotiation between the Division and the producing party.
The second modification to the Model Second Request is additional language in Instruction O.9 that addresses search terms. There can be little doubt that search terms selected in a vacuum are not an effective means for identifying relevant documents in a collection. Notwithstanding this conclusion, they still remain one of the most widely used methods to reduce the size of a production. In the investigative context and particularly in a merger investigation, there is significant concern that Division staff may not know enough about the transaction, the industry, and the company to provide meaningful input into the selection of search terms. For example, there may be code words, company jargon, or industry phrases that are unknown to the Division. Notwithstanding these limitations, and based on recent experience in several investigations, the Model Second Request will require that the producing party provide additional information about the search terms that are used and about the words that are in the underlying collection. This information is sought in order to better prepare staff to evaluate the producing party’s use of search terms and to provide some means by which staff can have input in the selection of search terms.
Databases present special issues in antitrust investigations.(4) Because of the kinds of information they contain – data about transactions, sales, prices, customers, etc. – they are often critical to merger investigations. It is also true that negotiating the production of information from databases (which is more common that negotiating the production of an entire database) is difficult and time-consuming. We encourage the producing party to initiate these discussions as early as possible. However, it is essential that counsel have collected the necessary information about various corporate databases – from the information they contain to how various databases relate to one another. For an organization that is prepared for discovery, this information should be readily at hand. E-discovery experts have been urging corporations to be prepared to disclose this information by creating a data map for their organizations. If counsel has been advising the organization about its preservation obligations, by necessity counsel should have undertaken a review of all the potentially relevant sources of information – including databases.
The final significant change to the Model Second Request involves the information that the Division requests about the producing party’s databases. The Division has incorporated a new Specification that seeks information about the kinds of information contained in the database, how it is structured, the fields it contains, types of reports, and similar information. We added this specification to request not just existing documents, but where the company lacks this information, to assemble the information. The purpose is not to have counsel create long, tedious interrogatory answers. Rather, the purpose is to have counsel identify the information necessary to have meaningful negotiations. In addition to this information, I would always recommend that counsel provide sample reports from databases to facilitate negotiations.
E-discovery presents significant challenges to both the Division and the private bar. I believe the Division has made significant progress, but significant challenges remain. Communication and cooperation are the cornerstones of any successful electronic production negotiation. The electronic productions that have gone the most smoothly have involved frequent communication with counsel, company representatives, and the company’s electronic production vendor. I encourage you to raise any issues or problems as early as possible. Electronic production issues will continue to evolve. Staff will continue to evaluate its practices to ensure that an appropriate balance is struck between our need to complete our investigations and the cost and disruption to the corporations that receive our compulsory process.
1. The views presented in this paper are my own, they do not reflect those of the Department of Justice or the Antitrust Division.
2. The Sedona Conference Cooperation Proclamation (July 2008), available at www.thesedonaconference.org/content/tsc_cooperation_proclamation.
3. Predictive coding, generally speaking, at its base, leverages the review of a small subset of the documents in a collection to identify similar documents in the entire collection. The review relies on sampling techniques and computer-assisted review to validate the review process.