A Major Leap Forward In E-Discovery

Monday, December 6, 2010 - 01:00

Editor: Would you give us your background? Tell us how you came to join Xerox Litigation Services.

Burrows: I was a litigator for 20 years in the San Francisco office of McKenna & Cuneo (now McKenna Long & Aldridge LLP). During the early '90s in the course of managing progressively bigger and bigger cases I started to see many electronic evidence issues arising in discovery, stimulating my interest in that area. The focus of my interest shifted at that point, and the last ten years of my professional life I have concentrated on e-discovery.

Editor: What is the size and professional make-up of your team?

Burrows: Xerox Litigation Services is a business unit of Xerox Corporation. Within Xerox Litigation Services (or XLS), we currently have 250 employees - all devoted to electronic discovery services - representing growth of about 20 percent this year.

Editor: When did Xerox acquire this particular division?

Burrows: Xerox acquired a start-up known as Amici LLC (now XLS), an electronic discovery company, in 2006. Amici was founded in 2002. Since the acquisition Xerox has made a substantial investment in the growth of XLS; both in terms of the team and the technology.

Editor: How do you help your clients decide what tools they should be using on a feature-rich review platform?

Burrows: Our client service managers, a dedicated team of about 50 people, work exclusively with clients using our technology. They are walking-talking manuals on best practices, spending all of their time everyday working on electronic discovery projects. These people are best able to help clients decide how to approach their particular challenges in a given case, advising not only on technology features but best practices in terms of getting to the end point efficiently and accurately. We also have a team of 40 software developers that work closely with our client managers to make sure we are providing the features that our clients need most. For instance, E-mail Analytics is a tool that was developed directly from a client request. CategoriX, our automated document classification system, was developed internally by Xerox R&D in the U.S. and Europe.

Editor: How many lawyers are on these teams?

Burrows: Our teams have varied backgrounds. Seventy percent of the professionals in our client services group have some advanced degree - whether in law, business or technology - providing a mix of professional backgrounds and covering all facets of skills required to meet the needs of our clients.

Editor: What is the E-mail Analytics tool?

Burrows: E-mail Analytics is a tool designed to help identify the communications of key individuals or witnesses related to a litigation or investigation, helping to identify who they were communicating with and what they were communicating about during critical time periods. In litigation it is vital to be able to see who critical witnesses are communicating with, how often, and on which topics - this process also helps you identify additional witnesses perhaps previously unknown in the case.

Editor: The legal market is not known for being an early adopter in technology. Is the market ready for tools like CategoriX that take all the review out of the hands of the attorneys?

Burrows: I think it is, although use of machine-aided review tools have been very slow in gaining acceptance. The legal community has been slow in trusting technology over human review, even knowing that this is proven technology. But the single most important reason that we think the market is ready for it is quite simply because document review costs are climbing out of control. Document review is now the single biggest driver of cost in litigation. General wisdom is that 80 percent of all litigation costs come in the discovery phase and 80 percent of those costs stem from document review. Doing the math, it's apparent that over 60 percent of all litigation costs stem from document review. What many partners at the AmLaw firms are telling us is that document review is now the tail wagging the dog. Because of the cost involved in document review, lawyers are not able to provide the value added services - trying cases, motion practice, taking depositions - all those things that they went to law school to do. We think we can give a big part of that budget back to our clients and their law firms to actually practice law and resolve cases on the merits. The market is ready for this type of change.

Editor: Could you explain how CategoriX differs from other document review tools used in e-discovery? Tell us how it does statistical ranking.

Burrows: CategoriX is an automated document classification tool. It uses technology coupled with experts in fields like linguistics and statistics to take a sample set of documents within a population that typically would be slated for review. This subset, a training set of documents, is reviewed by lawyers. What is of supreme importance to this process is that lawyers most knowledgeable about the case can do the initial review of documents, ascribing codes or values to the documents. The attorney-coded subset is then used to "train" CategoriX to classify the larger remaining document population according to the parameters established by the attorney reviewers. The process is iterative and involves refinement of the training set - to ensure that the lawyers' initial coding is consistent and the judgments that will be generalized are accurate. This is one area where the technology helps insofar as it evidences when there are inconsistencies in coding by the lawyers. We will often see that even very senior people have different takes on relevancy calls for various issues. We make sure that the technology first evidences these inconsistencies so that we can normalize the coding. CategoriX then applies the uniform coding to the larger document population. The result is a ranked value of where all those documents stand relative to that original training set - substantially speeding up the document review process in general. In a recent engagement, our client only had to review seven percent of a two million document population. That translates to significant cost and time savings.

Editor: This is really a kind of exercise in artificial intelligence, isn't it?

Burrows: There is definitely an artificial intelligence aspect to it. We are using mathematical algorithms, very importantly coupled with human expertise - statisticians, linguistic experts and lawyers - the best of all worlds. It is human-assisted computerized review.

Editor: Can metrics be used in other areas besides automated document classification?

Burrows: Metrics are increasingly important, especially as document volumes grow. Our clients want primarily to be able to get to the most important data quickly, eliminate irrelevant document sets up front, and make sure that the actions they are taking with regard to their documents are consistent. There are review metrics that show you how quickly a team of reviewers is proceeding so that you can tell whether you are going to meet your discovery production deadline. Metrics can also show if there are reviewers who are outliers, that is, particular coders proceeding a lot faster or a lot slower than the rest of the group. These measures would cause you to investigate to make sure that the review is being conducted in a consistent and optimized fashion.

Editor: How does XLS ensure high-quality standards?

Burrows: Quality control is one of the ways in which we distinguish ourselves. In every step of the process we use both technology as well as manual checking, including peer review. From preloading of documents through final production all documents are checked and re-checked by human beings and by technology each step of the way.

Editor: What other services do you find your clients asking for?

Burrows: Our clients tend to be Fortune 500 corporations and the large law firms representing them who are looking for the type of services and assurances that large companies expect. So, for example, they are very concerned that a platform is reliable. We consistently run at over 99.95 percent up-time. Clients are interested in security and redundancy. One of the big buzzwords right now is "early case assessment" or ECA. Clients are very interested in getting to key data quickly. This involves more than simply filtering and culling data, which is what some companies are calling ECA. ECA should be about finding out quickly what are the critical documents and where are the documents that affect the value of the case. The tools that we use - Search Analytics, Document Analytics, E-mail Analytics, and other analytical tools - help our clients get to the essentials of a case expeditiously so they can make a decision about whether their case is worth pursuing or defending, or whether to seek an early compromise and resolution.

Editor: Clients also want to know what e-Discovery is going to cost because that also determines whether they want to settle quickly.

Burrows: Absolutely. The cost of e-discovery at the beginning of a case can sometimes be so substantial that it eliminates the possibility of actually contesting a matter on the merits. That is why a tool like CategoriX, which can help eliminate a lot of those costs in early stages of review, is so valuable. Clients also want predictability in costs, which is a hallmark of XLS.

Editor: There is a lot of talk about bringing e-discovery in-house. What do you tell your clients when they ask about this?

Burrows: There is not a one-size-fits-all solution. What we typically tell our clients is there is definitely a place for in-house solutions where a technology platform of the scale and capacity of Xerox Litigation Services is not needed. I think our clients benefit by having the best of both worlds: while in-house tools can be used to save clients money on smaller matters, there are also advantages to being able to outsource, especially when big spikes in litigation occur, where a client wants quick results and doesn't want to invest in technology and additional staff to run it. I think the right balance is to have access to each.

Editor: Tell us about your foreign language support, because this feature would appeal to Fortune 1000 companies that are overseas.

Burrows: We are capable of handling data in over 130 different languages, (including those with non-Latin alphabets), and we also have tools that assist with language recognition within documents. Increasingly, you are dealing with three or four languages that can appear in a data set, and our tools can help clients identify what languages a document contains and where those languages appear. This proprietary foreign language identification technology is unique in the e-discovery space because it can parse language at the sentence level as opposed to only the document level.

Editor: What do you see as challenges related to e-discovery in the future?

Burrows: The real challenge is the one that we have already touched on - the overall cost related to the growth in the volumes of e-mail and other data. Finding ways to reduce the cost of dealing with that volume of data in document review - using other tools such as near duplicate identification, de-duplication, e-mail threading and other tools that can eliminate large pieces of review populations - is the biggest challenge.

Editor: Do you think that the time will come when lawyers will no longer be needed in document review - that machines can do their work?

Burrows: Never. Legal judgment based on training and experience will always be required. But the time has come when lawyers do not need to lay hands on every piece of paper related to a case.

Please email the interviewee at randall.burrows@xls.xerox.com with questions about this interview.