Our Insights

Thought Leadership and Industry Trends

Home 9 Insights 9 Winfield v. City of NY: What Happens When Computers and Humans Disagree in eDiscovery?

Winfield v. City of NY: What Happens When Computers and Humans Disagree in eDiscovery?

Jan 16, 2018

Predictive coding has become increasingly prevalent in eDiscovery. Although courts have ruled that parties are allowed to use the technology, they have also recognized that litigants aren’t required to use it. However, once a litigant has decided to employ predictive coding, the question is to what extent courts should permit discovery into, and allow opposing counsel, to re-examine and second guess a litigant’s TAR process? The recent case of Winfield v. City of NY, No. 15-CV-05236 (SDNY Nov. 2017) (LTS) (KHP) addresses these questions.

The Winfield court looked to Sedona Principle Number 6, which states that “responding parties are best situated to evaluate the procedures, methodologies, and technologies appropriate for preserving and producing their own electronically stored information (ESI).” However, as attorneys and clients turn to Technology Assisted Review (TAR), including predictive coding, to save time and costs on large discovery projects, the danger arises of increased scrutiny into the technical details of the processes and workflows that counsel chooses to find and produce responsive and non-privileged documents.

In the Winfield case, after the Defendant reviewed 100,000 documents, Plaintiffs demanded that Defendant review an additional 90,000 documents from new keyword searches and an additional document custodian. To save costs, Defendant implemented a TAR process that allowed Defendant to leverage the existing coding of the first 100,000 documents to predictively code the 90,000 documents. Plaintiffs challenged the TAR process and asked for detailed discovery into the Defendant’s TAR process.

The Court recognized that the Defendant and responding party followed a well-designed TAR process that included a robust quality control procedure that gave human attorneys the ability to check the output of a computer’s TAR algorithm and overturn coding decisions of a computer without triggering a right by opposing counsel. Defendant used a robust “seed set” of documents taken from the review of the first 100,000 documents, as well as the discovery requests and “sample documents.” After running TAR, Defendant then sampled the results to ensure quality. Although the results were not perfect, the Court recognized that perfection is not the standard in discovery. As a result, the Court was willing to uphold the Defendant’s TAR process despite several problems identified by Plaintiffs. These included documents that were arguably responsive yet withheld by Defendant as non-responsive, and documents withheld as privileged from which Defendant later withdrew their privilege designation.

Every trial lawyer knows the danger of arguing a complicated discovery motion and going too deep into the weeds. In rejecting a wholesale effort to conduct discovery and second guess the Plaintiffs’ TAR discovery process, the court applied Sedona Principle Number 6 and prior case law that reached a similar conclusion. The Court noted that traditionally, courts have not micro-managed parties’ internal review processes for several reasons, including because internal attorney ESI work processes may reveal work product, litigation tactics, and trial strategy. The Court agreed that the Defendant is in the best position to determine the most effective way to identify and produce responsive and non-privileged documents, and the judge would not second guess counsel, even when there were instances of mishaps or imperfections in the results of the process. However, the Court also felt some transparency was required. Accordingly, it allowed some additional sampling of documents that Defendant withheld, but did not allow a full review of documents where the computer and the human attorneys disagreed. Plaintiff was not permitted to review all documents “predictively coded” responsive that were withheld as non-responsive. Nor could Plaintiff discover the discovery process details including the scoring threshold that Plaintiff presumed delineated the computer’s distinction between responsive and non-responsive.

Winfield is an important decision. It clearly applies Sedona Principle Number 6 to a complex fact scenario, protecting critical attorney work product, and recognizing the importance of transparency. However, as attorneys turn increasingly to TAR technologies, we will see more and more disputes that focus on critical technological details. TAR technologies require the courts and counsel to have a thorough understanding of how the technologies work and how they impact the rules that govern discovery. As more questions about technology are asked, parties also risk a loss in credibility when they mishandle technology.

Technology is a boon to eDiscovery, allowing litigants to review large quantities of data and find relevant results. However, it’s crucial for parties to follow best practices so they can justify their processes when needed. Consulting eDiscovery experts can help.

CDS’s Advisory Services team regularly consults on the use of TAR in eDiscovery. Contact us today and learn how TAR may be able to help you on your next matter.

About the Author

<a href="https://cdslegal.com/team/william-wallace-belt-jr/" target="_blank">William Wallace Belt, Jr., Esq.</a>

William Wallace Belt, Jr., Esq.

Bill has 25 years of experience as a partner, shareholder and board member of AmLaw 200 law firms, including Williams Mullen where he built one of the first law firm eDiscovery practices. For the last 10 years Bill has participated in and now runs a Corporate-Only Roundtable where corporate eDiscovery teams, judges and industry experts have met to share experiences and discuss emerging challenges and solutions that combine technological and legal solutions.