Less than a year after a federal judge issued a seminal electronic discovery ruling blessing the use of computer-assisted review, the technology is already appearing in courts across the country.
In late February of last year, U.S. Magistrate Judge Andrew J. Peck of the Southern District of New York was the first to issue a reported opinion in support of the technology — also referred to as “predictive coding” or “intelligent review” — calling it “an acceptable way to search for relevant [electronically stored information] in appropriate cases.”
“Computer-assisted review appears to be better than the available alternatives, and thus should be used in appropriate cases. While this court recognizes that computer-assisted review is not perfect, the Federal Rules of Civil Procedure do not require perfection,” Peck wrote in Da Silva Moore v. Publicis Groupe.
“What the Bar should take away from this opinion is that computer-assisted review is an available tool and should be seriously considered for use in large-data-volume cases where it may save the producing party (or both parties) significant amounts of legal fees in document review,” Peck added.
Since the decision was issued, there has been a “silent revolution” in favor of the use of predictive coding, according to Jonathan Sablone, head of Nixon Peabody’s e-discovery and digital evidence practice group in Boston.
Although only a handful of decisions have been issued addressing the technology, it is being used by parties in cases across the country without appearing on the public record, Sablone said. “Da Silva did change the landscape.”
Ralph C. Losey, who represented Publicis and fought to use predictive coding in the case, said he is unaware of any court being presented with the technology and deciding not to allow its use. A partner at Jackson Lewis in Orlando, Fla., and chair of the firm’s electronic discovery practice group, Losey recently used predictive coding for a case in arbitration.
“We broke the ice,” he said, adding that 2012 “was the breakthrough year, but  will be the big year when [the technology] really catches on.”
Other courts weigh in
In the months since Da Silva was handed down, other courts across the country have also addressed the issue of computer-assisted review.
In April 2012, a Virginia judge issued an order allowing the use of computer-assisted review over the objection of plaintiff’s counsel. (Global Aerospace Inc. v. Landow Aviation, No. CL 61040 (Va. Cir. Ct., 2012).)
A federal judge in Illinois issued an order in support of the technology in September 2012. (Kleen Products LLC v. Packaging Corp. of America, No. 1:10-cv-05711 (N.D. Ill., 2012).) However, the parties later agreed to a different form of electronic review.
An order issued last July addressed a dispute between the parties about how to build the control seed set of documents in the Actos MDL located in the Western District of Louisiana. (In re: Actos, MDL No. 6:11-md-2299 (W.D. La., July 27, 2012).) (A “seed set” of documents is used to train the software to identify other relevant documents through a series of sample rounds during which it learns to predict the coding of the entire document set. )
Then, in October, a Chancery Court in Delaware issued a sua sponte order requiring the parties to use predictive coding. (EORHB, Inc. v. HOA Holdings, Inc., No. 7409-VCL (Del. Ch., 2012).)
The Delaware case has many practitioners talking.
“The decision jives with what I’ve been seeing in my practice,” Sablone said.
Similarly, he had a judge instruct counsel to use predictive coding without either side suggesting it.
“That is a sea change,” Sablone said. “Federal judges out there are paying attention and are now open to the idea [of using the technology], at least as one of the tools in the e-discovery tool belt.”
Computer-assisted review remains most cost-effective and efficient in cases with large amounts of data.
While Sablone has used predictive coding in a case with just 50,000 documents, he says it starts to become more efficient with 100,000 or more.
Thomas C. Gricks III, a partner in the Pittsburgh office of Schnader, Harrison, Segal & Lewis and chairman of its e-discovery practice group, represented Landow Aviation in the Virginia case, which involved the collapse of a hangar in a February 2010 snowstorm.
“We had a boatload of data,” he said, estimating around 2 million documents. “We knew that, by and large, a lot of that stuff was going to be unnecessary to the case and unrelated.”
Gricks said computer-assisted review was “the most efficient and cost-effective way to find the relevant stuff.”
More and more cases involve millions of documents, he said, and the technology can save clients significant amounts of money by weeding out unresponsive data.
Some practitioners also use predictive coding for other discovery-related purposes, such as prioritizing documents to be reviewed or as quality control for manual review, said Nashville, Tenn.-based Drew Lewis, eDiscovery counsel for the electronic discovery company Recommind.
The year of computer-assisted review
Practitioners agree that computer-assisted review will be a big e-discovery focus in 2013.
“Three years from now, it is my hope and prayer that keyword searching will be dead and predictive coding will have taken over,” Losey said.
Judges and attorneys will increase their comfort level with the technology and its use will grow, trickling down to smaller cases and to state courts, experts predicted.
Clients will also push for a switch to predictive coding given the potential cost savings, Lewis said.
And, Sablone predicted, court decisions will shift from disputes over whether to use the technology to arguments over the amount of information to be shared between the parties.
For example, opposing counsel may seek to obtain the “seed set” used to search for documents and request to see the non-responsive documents that were identified during the process of training the software, he said.
Many lawyers will balk at sharing such information.
How the computer was trained to find responsive results is what attorneys traditionally equate with their mental impressions, something that has historically not been revealed to opposing counsel.
“Technology-assisted review doesn’t happen with the touch of a button,” Sablone said. “It is a multi-step, iterative process with human reviewers making decisions about what is responsive. A couple of mistakes in human review can have great implications for how accurate the computer system is.”
But lawyers are not accustomed to sharing non-responsive search results or the details on how they crafted search terms, Sablone added.
The decisions to date have required a great deal of disclosure, said Losey, who is the author of four books on electronic discovery. He expects future battles to result in a tightening of how much information must be disclosed as parties become more accustomed to the technology.
“Quality is going to be the biggest concern,” Losey said. While challenges will be raised as to the results of a search, “we never had any quality assurance [in discovery] before now.”