ESI: Focus on Review and Production Strategy Meredith Lee, Online Document Review Supervisor, Paralegal
About Us Avansic E-discovery and digital forensics company founded in 2004 by Dr. Gavin W. Manes, former Computer Science professor Scientific approach to ESI processing Strong background in academics and research Based in Tulsa, OK with numerous regional offices Meredith Lee, Online Document Review Supervisor, Paralegal Experience managing document intensive cases, class actions Assists clients with collection, review, coding, production Assists clients with strategy and mechanics of document review
Agenda The Problem The Challenge Solutions
The Problem Too much ESI Any type of ESI Files Emails Application Data Other Sources of ESI Computers, tablets, thumb drives, cell phones, ipods Digital cameras, faxes, printers, automobiles
Communication and Vocabulary Good communication is critical jargon is dangerous Technical terms can be defined MANY different ways
ESI Rules FRCP Rules Discoverable documents Planning for discovery Exchange protocols Burden and cost State rules 2/3 states have rules Case law
Reactions
Key Decisions Zubulake v. UBS Warburg (2004) landmark case Victor Stanley, Inc. v Creative Pipe, Inc. (2005) - $1M in sanctions and jail US v. O Keefe Equity Analytics Mancia v. Mayflower Pension Committee - defining the litigation hold
A Solution Litigation habits and customs learned in the days of paper must be revisited and revised. The culture of bench and bar must adjust. - Hon. Lee Rosenthal US District Court for the Southern District of Texas, Chair of the Standing Committee on Rules of the Judicial Conference
ESI Processing
Page Counts (Unscientific) Document Type Average Pages/Doc Average Pages/GB Average Pages/MB Microsoft Word files 8 64,782 63 Email files 1.5 100,099 97 Microsoft Excel files 50 165,791 161 Lotus 1-2-3 files 55 287,317 280 Microsoft PowerPoint files 14 17,552 17 Text Files 20 677,963 662 Image Files 1.4 15,477 15 Review & Production Strategies
Document Review Most expensive part of e-discovery Most time consuming part of litigation Before you begin, answer: What? Who? How?
Review: What What document types will be reviewed? What document volume will be reviewed? What am I trying to find? Responsive documents to produce Privilege or confidential The Smoking Gun
Review: Who Who will do the review? How many reviewers are needed? What skillset and training do those reviewers need? Who will validate the review? Quality control Second pass Who will receive the results of the review? Your litigation team Expert witnesses Opposing counsel Government
Review: How How will we manage the project? Linear review line by line (electronic) Managed review focused on project management and advanced strategies (batching, search filtering, etc.) Batching (not ordering) Quantity Custodian Metadata (Dates, Types) Similar Documents via Clustering
Production Considerations Formats (Review Production) The end of processing is the wrong time to decide production format Native files, TIFF, PDF, Multiple If you ve reviewed TIFF and they asked for native file, you may have to review again Redactions and native issues Delivery Online (repository or secure file transfer) DVDs or hard drives (encryption) Maintain a copy of what you produce Who will prepare productions Allow time for redactions and QC
Scientific De-Dupe vs. Text Near-Dupe Scientific de-dupe removes exact copies Files: use a hash to determine if two documents are the same E-mails: combination of data elements (metadata) or a unique identifier from the e-mail server Text near-dupe groups documents by context Compares the body of a document and accepts minor differences Identify like documents for review Should not be used to automatically eliminate
Search During Review Most review tools offer strong search features to refine and reduce the data during review Simple searching (easy) Relational searching (difficult) Contextual searching (complex) Concept searching (next generation)
Language of Searches Searches are written in their own unique languages (syntax) Boolean Operators Context and Proximity Operators Relational Operators These operators and many others allow for complex search combinations
Search Syntax Different review tools use different search syntax Some let you use & or + for and ADJ is also known as w\1 Not is also known as > or < or! % and Like often mean same
White Space and Noise Words Noise words Common words excluded from indexing in search or review tools In exact phrase searches, these words are treated as white space and not included in hit results it appears likely that a letter of credit is needed = appears ADJ1 likely ADJ2 letter ADJ1 credit ADJ1 needed
Concept Searching Searching documents based on what they are about Language algorithms read the concept of a document Find related documents that don t contain the actual search terms Used correctly, you can quickly get to the core documents you want to review first
Clustering Tool types Black box More like this or find similar Widely-known algorithms Potential workflows Filter, then cluster, review sample set based on those clusters, then apply that sample set coding to the remainder of the set Send small data set through clustering to determine search terms And more. When is it useful? With or without online review component Cases with 10,000+ documents
TAR not HAL Technology Assisted Review Predictive coding Is it ethical to avoid technology to reduce review? Huge document sets NARA case: 200+ million government emails BP case: 1+ billion pages Review & Production Strategies
Conclusion Plan ahead for both production and review Reasonable expectations for turnaround Project management is key Save time and money Eliminate duplicate review Use technology to your advantage Sophisticated software exists Used properly, it can ease the burden of e-discovery Review & Production Strategies
avansic.com Corporate Office Mid-Continent Tower, Suite 1701 401 South Boston Avenue Tulsa, OK 74103 Meredith Lee meredith.lee@avansic.com