The legal and practical challenges surrounding discovery in the age of electronically stored information (ESI) continue to center around the preservation of an ever-growing volume of data and documents. Practitioners note that managing both legal risk and litigation costs are tied to developing effective strategies and processes around records retention and review, and new technologies have been developed to streamline these tasks. Our panel of experts addresses these topics, as well as how they select e-discovery vendors, and the influence of court decisions addressing ESI in Pension Committee, Victor Stanley, and Zubulake. They are Anthony Knaapen of Chevron Corporation; Jim Maroulis of Oracle Corporation; and Randall Burrows of Xerox Litigation Services. California Lawyer moderated the discussion, which was reported by Krishanna DeRita of Barkley Court Reporters. Moderator: In complex litigation, what e-discovery problems do you face? Maroulis: There are a series of problems: preservation, production, and the avoidance of spoliation. Production is probably the least difficult to deal with, although it can be expensive. Preservation creates more challenges because the field of data that you must retain can be almost limitless. At some point, it becomes an issue of diminishing returns because there may be 500 custodians who have documents somehow related to your litigation, but their utility is virtually nil if you have documents from the 50 people with the most interesting and relevant data. Nonetheless, if you read Federal Rules of Civil Procedures Rule 26, it imposes a broad obligation to preserve data. That bleeds into the issue of avoiding spoliation, where you have to make sure that you don't somehow lose data—even if it's only tangentially related to your case—through departing employees or under a data retention schedule. Knaapen: One difficulty is identifying the right custodians. It is impossible and impractical to send every preservation order to 60,000 employees. You have to determine which companies and business units are involved, and identify the people who have the right records who can transfer some knowledge to you about the types, volumes, and location of data that you need to preserve and collect. Burrows: I see three areas of risk. One is a sanctions risk, which goes directly to the preservation issue, and the possibility of being called into question for missing data. We see sophisticated law firms using e-discovery as a bludgeon. The second risk is related to cost because you are trying to synthesize a vast amount of data and much of the cost is an upfront cost of processing and searching large quantities of data. The third risk is the risk to people. In Qualcomm Inc. v. Broadcom Corp. (539 F. Supp. 2d 1214, (S.D. Cal. 2007)), lawyers got into trouble with the State Bar, and attorneys in Victor Stanley, Inc. v. Creative Pipe, Inc. (269 F.R.D. 497 (D. Md. 2010)) are looking at prison time. Knaapen: To Randy [Burrows's] point about the bludgeoning aspect of e-discovery, it is important to follow your established processes and document what you do because your adversaries will try to find a mistake, regardless of whether it has any materiality to the facts of the case, and then that becomes the focus of the case. Fortunately, in Pension Committee (Pension Comm. of the Univ. of Montreal Pension Plan v. Banc of America Securities, LLC, 685 F. Supp 2d 456 (S.D.N.Y. 2010)), Judge Shira Scheindlin said that the absence of a document that would have been responsive to a request is not enough to merit sanctions—you must show that the document would have been helpful in proving the case. It is also important to remember that the standard for preservation is reasonableness and proportionality. Burrows: Courts have never required perfection in discovery, only reasonableness. But people do get caught up in the volume of data and in the technology, and sometimes that overrides rational thinking. However, the entire process around preservation is important because if you preserve the right data, you can always fix things, and you can be more aggressive in your strategies for reviewing data and producing records. Moderator: How do you mitigate e-discovery risk and costs? Knaapen: One component of mitigating risk is the underlying culture of the company. If you have a history of high ethical and business standards that have been conveyed, reinforced, and trained into the employee base, then you will probably get a better response on record retention. We provide training on our preservation responsibilities, as well as on information management and record retention. We also rely on our COBALT® process (Chevron Objective Base Litigation Technique), which requires the engagement of our managing attorney, our business client, and our outside counsel to collaboratively evaluate each matter and prepare an appropriate strategy, case plan, and case budget to achieve the best result. We have established processes that we try to use consistently and uniformly across the enterprise and have a group of project managers who help to reinforce those processes. Proper planning, budgeting, and reporting all result in better cost control. Burrows: Preservation is the fundamental strategy, but many tools are available to reduce costs. There's a lot of talk about early case assessment (ECA), which initially was just another name for filtering. Now tools are much more analytic so you can make earlier decisions about what you need to review, process, and produce. These tools can help you better analyze your searches, look at relationships between witnesses—things that really go to the substance of the case, and which can create a cost savings. Maroulis: There needs to be transparency about what you are doing. If you don't have transparency, it leads to countless arguments. If you do, you can agree with your adversary that you will only preserve a certain amount of data from certain custodians, with the possibility of supplementing it in the future. And you can agree on specific search terms. Whether it's an adversary in a civil case or an attorney with the U.S. Department of Justice of the U.S. Securities and Exchange Commission, by having that dialogue and transparency, you can establish and document expectations, and then you've fulfilled the preservation or production obligations. We have also negotiated search terms, and in symmetrical litigation, things work out quite well. We typically run the terms that the other side is proposing against our database to figure out how burdensome it would be to review everything they are asking for. We try to narrow it down from there, and they do the same. Knaapen: It's important to engage outside counsel early when deciding upon search terms. Do some sampling and show them the results. They need to understand which search terms in the context of your corporate culture, your business environment, and your infrastructure, will produce the most relevant documents. Then they are in a better position to advocate for you. Maroulis: Another option is to offer to shift the costs. You can tell your opponent that you're willing to review one million documents, but because their search terms generate five million hits, they have to pay for my outside counsel if they want me to review additional documents. They then have to articulate to the court why these tangential searches are necessary. Moderator: How do you decide which e-discovery vendors to use? Maroulis: We have a director of e-discovery who regularly meets with vendors, and we work together as a team to figure out who is the best fit. Making the relationship with the vendor work requires communication between the company and the vendor because without a constant stream of communication about the kind of data you are dealing with, deadlines, and how the data needs to be processed, you are going to run into delays or problems. For example, we may have encrypted data. We also have proprietary systems that the vendor may never have seen before, and the vendor may have no idea how to process them. Burrows: Communication and collaboration among vendors and clients is crucial. In some of our cases, we are talking to clients several times a day. We are synthesizing new information as it comes in and changing priorities as necessary. Knaapen: Collaboration and communication between in-house and outside counsel is also important. We issue preservation orders in anticipation of litigation. When you retain outside counsel, you have to explain your processes, the steps you have taken, and the vendors you require your retained counsel to use. Counsel has to be comfortable with what you have done and the vendors you select. You also need to keep decision logs about what you have done so that a year later, when an issue arises, you can show that you took reasonable steps, that you were paying attention, and that you acted in good faith to fulfill your preservation discovery obligations. Maroulis: One of the most critical things to document is all of the dead ends that you run into. You need to be able to say, "I interviewed these 50 custodians and they don't have any documents." Documenting every step is essential if you are going to have a defensible plan. Knaapen: We have some legacy cases that go back for years and that will continue on into the future. We didn't want to lose the work product or build a different database with every law firm that we work with. So we chose a document review tool that had a strong repository feature. We choose vendors who can process data and conduct reviews using that platform. Our IT infrastructure is complex. We don't want vendors coming in and mucking up our system, and we don't want to constantly explain that system to a new vendor, so we have developed an internal IT forensic group. Burrows: For clients, there are only two ways you can make a vendor selection, and that is you either try them out, or you get good references. And nobody does an exclusive. Clients choose more than one preferred provider because different vendors may be appropriate for different cases. Moderator: What new technology or protocols do you use to curb costs related to document review? Burrows: More than half the cost of litigation is now related to document review. So that presents a prime opportunity for cost savings. There are new tools that will allow machines to assist with review by triaging the material so that attorneys don't have to read each document. No one expects that attorney review is going to be replaced entirely, but given how much data there is, technology can eliminate large quantities of irrelevant material. Maroulis: The new technology has helped tremendously. We have designated document review vendors, and they are doing it at lower cost than the law firms we employ. In terms of automated processes, we've compiled a list of every attorney we have worked with in recent years. We run that list against any dataset that we get, which segregates out most of the privileged attorney-client communications. Automated tools that allow you to identify related documents are also helpful. You can run into problems if opposing counsel is making motions to compel because you produced an email but didn't produce the response to that email because two document reviewers coded the emails differently. Knaapen: We are moving away from using law firms for document review; we prefer that our vendors do the first and second pass. We are trying to maintain control so that we have consistency, higher quality, and cost predictability. But you don't do this in isolation; you engage outside counsel in designing the review methodologies and protocols. Outside counsel get the final call on privilege and responsiveness, but initial reviews will be done by a document review company using review tools with clustering and concept-searching technology. Burrows: If you can categorize documents, whether it's by near-duplicate identification, clustering, or other tools, then you can provide subsets of documents to specific reviewers, and they can focus on those, instead of having them learn the entire case. Things go faster and the near-duplicate identification tool allows you to only look at a particular document once, which is important, given the proliferation of email and the multiple copies that can be sent to multiple parties. Maroulis: Having said so many wonderful things about technology, let me point out one risk, which is that there is also the possibility for a catastrophic error. With the push of a button, all of your privileged documents could be marked "to be produced," and after it all goes out the door, you must now try to claw it back. Knaapen: Here, again, you can do sampling to mitigate against error. You can look for the false positives and check what has been marked nonresponsive. By documenting and keeping a detailed log, you are in a much better position to defend against discovery or evidentiary motions that arise later. If anybody claims that you missed something, you can show that you ran a test. Burrows: One tool we developed performs what we call "automated document classification," which essentially combines mathematical theory, computer power, and human input. Given an entire dataset of documents, attorneys code a subset, which we call a training set. You use the attorneys who are most knowledgeable about a case to train the technology to apply those parameters to the larger set that hasn't yet been coded. Senior partners at major law firms sometimes do not agree whether the same document is privileged or not, but our process addresses that because we normalize the coding for consistency before we release the training set. You end up with a probability ranking of each document related to relevance or privilege. It allows you to triage the documents, eliminate manual review of large sub-populations of documents, and potentially save a significant amount of money on review. Moderator: How have recent decisions addressing ESI, such as Pension Committee, Zubulake, or Victor Stanley influenced how you manage e-discovery? Knaapen: In Pension Committee, Judge Scheindlin did a great job in defining what is considered negligence, gross negligence, and willful negligence. But then, in Rimkus (Rimkus Consulting Group, Inc. v. Cammarata, 688 F. Supp. 2d 598 (S.D. Tex. 2010)), Judge Lee H. Rosenthal highlighted the fact that different jurisdictions have different standards and requirements. Regardless of jurisdiction, if you make sure that you have applied the reasonableness and proportionality standards, and you have a process that is followed consistently, then you will be all right. If you find an error, disclose it. Also remember that search technology is not perfect and adamant statements that you have found absolutely everything, can set you up for failure. Burrows: The good news about these cases is that they got everyone's attention. But interestingly, a lot of these cases revolve around preservation. So they really underscore the importance of preserving properly. All of these cases also underscore the need to stay up on the latest technology. The courts expect you to understand technology, or that you will bring in the right people to understand it, to help you properly use it, and to explain what you did. Maroulis: In Victor Stanley and Pension Committee, both judges dropped footnotes in the decision about how annoyed they were that they had to get into e-discovery issues at all, but that the parties had forced it upon them since their conduct did not live up to the standards. So as long as you are living up to the standards, show transparency to the court and the other side, and you have a defensible process, you should be OK. Zubulake (Zubulake v. UBS Warburg LLC (220 F.R.D. 212 (S.D.N.Y 2003)), seemed revolutionary at the time. But now, the understanding around adhering to legal holds, what constitutes accessible versus inaccessible data, and how that affects your obligations to preserve or review data, are just the way you do business.