skip to content

Looking Forward from 2017 on Ediscovery

by Joshua Gilliland


There have been hundreds of cases with ediscovery issues in the past year, ranging from motions to compel, to the use of predictive coding, to proportionality analysis. Judges have resolved disputes in the spirit of Federal Rule of Civil Procedure Rule 1’s mandate to reach just and speedy resolutions, demonstrating strong understanding of complex issues from search terms to the scope of discovery.  Sifting through these many cases, the ones that may have the most value to attorneys for how to handle thorny ediscovery issues are the “nuts and bolts” opinions on discovery. Here are a few of the lessons from case law in 2017 as we look toward to the future.


The first line of analysis in any discovery dispute is proportionality. Judges have become very effective at balancing the value of the discovery being sought with the burden of producing the requested information. 

Proportionality analysis typically begins with a similar cadence:

Federal Rule of Civil Procedure Rule 26(b)(1) states that parties are entitled to “obtain discovery regarding any nonprivileged matter that is relevant to any party’s claim or defense and proportional to the needs of the case.”

The proportionality requirement was moved up to Rule 26(b)(1) to address the “explosion” of information that “has been exacerbated by the advent of e-discovery.”

The 2015 Amendments impose a burden on both parties and the court to “consider the proportionality of all discovery and consider it in resolving discovery disputes.”

The proportionality standard set forth in Federal Rule of Civil Procedure 26(b)(1) must be applied in each case when formulating a discovery plan. Requests for production of ESI and related responses should be reasonably targeted, clear, and as specific as possible.

The proportionality considerations under Rule 26(b)(1) include “the importance of the issues at stake in the action, the amount in controversy, the parties’ relative access to relevant information, the parties’ resources, the importance of the discovery in resolving the issues, and whether the burden or expense of the proposed discovery outweighs its likely benefit.” Where a party seeks the production of electronically stored information (“ESI”), the party from whom discovery is sought may avoid production if it can show that the ESI is “not reasonably accessible because of undue burden or cost.” Even upon such a showing, however, the court may order the material to be produced if the requesting party shows good cause. Further, the court “may specify conditions for the discovery.”

Courts across the United States uniformly begin analysis of discovery disputes with the above proportionality principles. In a discovery dispute over the production of timekeeping data in native format, the Court ordered the production of the data because 1) it was reasonably accessible; 2) critically important to the claims in the case; and 3) the cost of $5,000 and five days to conduct the work was not unduly burdensome compared to the amount in controversy.

Proportionality has been used as the reason to justify searching data of custodians, limiting depositions Judges have even drafted specific limits on what discovery could be requested. Rule 26(b)(1) arguably has appeared in every case with a discovery dispute, even if not specifically named, as the analysis to determine whether to grant or deny a motion to compel. The best advice for attorneys in 2018 is to understand how to argue the proportionality arguments, as it is the responsibility of all parties and the court.

Requests for Production

The art of drafting requests for production is the backbone of civil discovery. Without a narrowly tailored request that is proportional to the needs of the case, there is nothing to review for motions, depositions, or trial.

Federal Rule of Civil Procedure Rule 34 governs requests for production. As a preliminary matter, the party requesting discovery can state the form of production; the responding party can object to that form and state the form they intend to use. If no form is stated, the data must be produced as it is ordinarily maintained, which means in native file format.

In a case where two parent associations of children with disabilities sued California, Rule 34 was at the center of a large discovery dispute. The Plaintiffs requested data in native file format with associated metadata. The Defendants did not object to the discovery request. However, the Defendants produced the discovery with load files (and presumably static images) instead of native files with metadata. The Court held the Defendants could not ignore the Plaintiffs’ form of production, contrary to the requirements of Rule 34. The Court opined that requesting parties would take the ease of review as one of their considerations for stating the form of production, which was why the Plaintiffs sought the data in native file format. The Defendants were ordered to re-produce the discovery in native file format with metadata.

Search terms were at the heart of many discovery disputes. In one case, the Defendant produced 11,687 documents after conducting review of court ordered search terms, but withheld 44,337 documents as being non-responsive. The Plaintiff challenged the designation of the documents as non-responsive.

Chief United States Magistrate Judge C.J. Williams noted that the only claim the Plaintiff could make that the documents could be relevant was that the documents had at least one search term hit in the content. The Court found it would be disproportionate for the Defendant to go through documents to classify records that the Plaintiff thought could be relevant. As such, the Defendants were ordered 1) to produce all unresponsive ESI as Attorney Eyes Only; 2) that production would not be an admission to relevance; and 3) that the Plaintiff would bear their own review costs.

Magistrate Judge Elizabeth Hey began an opinion over disputed search terms with, “Discovery can be described as contentious.” The litigants sought judicial relief over lists of search terms and search strings. The first dispute was over whether search for the corporate names of Patent and Brand would give false positive results, thus should not have been used as search terms. The Court held that the relevance of the names of the parties outweighed the burden of having to cull false positive data. The Court went on to rule on each category of search terms and dates to be searched.

Requests for production arguably generate the most motion practice in discovery. Parties litigate over scope, whether a request is narrowly tailored, and over search terms contained within requests. Courts analyze and rule on each request individually. The overall trends from 2017 were that lawyers and judges spend substantial time drafting requests for production to meet the goals of being narrowly tailored and proportional. This is positive news, because it shows growing sophistication over how to request electronically stored information.

Predictive Coding

Courts continued to discuss, recommend, and analyze predictive coding. Nine court opinions referenced predictive coding in 2017, with one having a detailed analysis of predictive coding in a case.

The Court of Chancery of Delaware ordered a plaintiff to supplement its document production by identifying responsive documents by “applying predictive coding technology to all documents that have not previously undergone a document-by-document attorney-level review for their responsiveness.” The parties were ordered to work together with the service provider to train the predictive coding system.

One district court held that a party was entitled to use their predictive coding model for identifying relevant documents. Another discussed how predictive coding was used in multi-district litigation—that while it required time to train, the predictive coding helped “realistically manage the immense amount of information needed to be produced and reviewed” in the case.

Magistrate Judge Katherine Parker issued a detailed opinion on the use of predictive coding in a discrimination case over affordable housing in New York City. In Winfield v. City of N.Y, the Plaintiffs provided 665 additional search terms to be applied to the Defendant’s review database. The supplemental searches would have added 90,000 more records and cost approximately $248,000 to review. The Defendants agreed to run the additional searches, but stated they would use predictive coding to narrow the data set for review. Ultimately, the Court rejected the Plaintiffs’ challenge that the Defendants’ training of the predictive coding tool was either grossly negligent or unreasonable. The Defendants were ordered to conduct their supplemental review with the blended workflow of search terms and predictive coding.

Predictive coding remains a topic attorneys and judges are interested in, but there has not been much in depth analysis in the 2017 case law. This may mean that parties are using it in litigation without challenges from the opposing party, or that disputes are resolved during the meet and confer process.

Taxation of Costs

Prevailing parties were able to recover ediscovery costs for processing and production in native file format in 2017. This has not always been the result with taxation of cost cases. Exemplifications of recoverable ediscovery costs have often been limited in many cases to costs analogous to copy-making. As such, “intellectual efforts” to prepare a production are not recoverable, such as the work of document review.  

In Dropbox, Inc., v Thru, Dropbox, as the prevailing party, was able to recover $116,040.18 in taxable costs. Dropbox argued that their incurred production costs were in order to comply with the requesting party’s production requirements, which stated: “documents that exist in digital format and constitute or comprise databases or other tabulations or collections of data or information should be produced in the native format, including all metadata.” Dropbox also recovered costs because they explained in sufficient detail that they incurred costs for “collecting, scanning, converting to a usable file format, and processing for production.”

One of the keys to recovering production costs as a prevailing party is to sufficiently explain how the work performed was necessary to complete the production. One prevailing party recovered $1,832.44 for “reproductions costs.” The itemized bill stated the following: “Files transferred from native files to an internal server, where from there it was digitally transferred to an external portable storage. Each file was then named & transferred according to the original source. Quality Check each file that everything was named and transferred accordingly.” While this documentation raises issues of whether the collection was forensically done, it does highlight how to document collection of data for cost recovery.

Properly documenting each step of a production workflow is the best strategy for recovering costs as a prevailing party. Citing to the requesting party’s production requirements, explaining how work performed met those requirements, and providing enough information so the judge understands what happened, are the key takeaways from these taxation of cost cases.

What’s Next for 2018?

As we look ahead to 2018, attorneys should continue building on their past successes and learning from mistakes. Many of the discovery requests from 2017 demonstrated that lawyers attempted to propound narrowly tailored and proportional discovery requests. While many judges did narrow requests for production, it is clear that lawyers in 2017 were far more effective than lawyers in 2007 in drafting discovery requests. This is a positive trend and good news.

Lawyers should continue leveraging predictive coding to expedite document review. In 2017, there was only one case where there was a challenge on whether  predictive coding had been adequately trained to identify relevant data. This is perhaps a sign that parties are using predictive coding and other technology assisted review without any contention.

The recommendation for attorneys in 2018 is to be vigilant about the types of discovery that can appear in a case. Litigation can include messaging appsvideo files, and other data beyond email. Attorneys should keep that in mind during client interviews, meet and confers, and preparing discovery requests.