Sequencing of TAR & Keyword Search Terms: Which Should Come First?
The case law surrounding technology assisted review (or “TAR”) is slowly – but deliberately – developing. This welcome occurrence flows from the fact that TAR is gaining greater adoption, especially in complex civil actions and fast-moving investigations. With greater adoption comes more questions and disputes – which in turn increases the number of reported decisions – as litigants look to the courts for guidance.
Most attorneys are familiar with conducting keyword searches in the “traditional” way – that is, by using the Boolean operators of AND, OR, and NOT, together with some additional flourishes. Very often, we learned such search methodology either in law school or in practice as a way to perform electronic searches for relevant case law, statutes, and the like. And we carried over such learning and methodology to the searching of documents in the discovery process.
TAR, on the other hand, has become a bit of a catch-all for a range of techniques and processes that are used to bring greater efficiency and consistency to the document review process. In the beginning, TAR was largely limited to email threading, concept searching, and various flavors of automated document clustering. But as the technology has evolved, it now also may refer to the more robust task of “predictive coding,” where human subject matter experts “seed” the software with examples of relevant or otherwise probative documents, and the software “predictively extrapolates” such judgments to other documents automatically.
As they begin the work of negotiating their eDiscovery protocols and accompanying proposed orders, litigants often find themselves facing the following question.
Should Boolean keyword searching be conducted (if at all) before or after the application of TAR resources are brought to bear against a given set of documents?
The answer – as it is so often in the law, and at least for now – is that it depends, albeit with a developing slant in dicta toward TAR first, keyword searching second.
FCA USA v. Cummins, No. 16-cv-12883 (E.D. Mich. Mar. 28, 2017), is one of the latest cases to address the so-called “sequencing” of TAR and keyword search terms. At issue in Cummins was the division of financial liability for the cost of an automotive recall. In a short memorandum opinion, the Court praised the parties for their ability to “agree on many issues relating to discovery and particularly electronic discovery.” And yet, the Court went on to note that “unfortunately [the parties] cannot reach agreement over whether the universe of electronic materials subject to TAR review should first be culled by the use of search terms.”
As is often the case when courts are faced with ruling on issues concerning eDiscovery, in addition to reviewing the papers submitted by the parties, the Court in Cummins also turned to the thought leadership provided by The Sedona Conference®. Specifically, the Court noted that it had read and consulted the Conference’s recently published TAR Case Law Primer, 18 Sedona Conf. J. 1 (2017) (available here).
In a section entitled “Disputed Issues Regarding TAR,” the Primer (at 27) collects an additional four cases concerning the sequencing of TAR and keyword search terms. Following review of same, the Court in Cummins held that “[a]pplying TAR to the universe of electronic material before any keyword search reduces the universe of electronic material is the preferred method. The TAR results can then be culled by the use of search terms or other methods.”
But it should be noted that the Court in Cummins made its ruling – in its own words – “reluctantly.” As it noted, in light of the “magnitude of the dispute and the substantial matters upon which they agree, the parties should have been able to resolve the discovery issue without the Court as decision maker.”
And therein is perhaps the most valuable takeaway from Cummins. Even with parties before it who were praised by the Court for how much they were able to agree upon outside the courtroom, the judge in Cummins still did not seemingly want or enjoy having to rule on the parties’ narrow discovery dispute.
The American system of discovery in civil cases is designed to be self-executing – and self-regulating – by the parties, with only limited (and then only reluctant) involvement from the courts. Litigants are almost always – aided by their own counsel and eDiscovery experts – in a better position than judges to reach decisions about the technical aspects of eDiscovery. And that’s why “calling the judge” should be a chess move that is used sparingly, and then only after every other move on the board is considered and rejected.
Contact Us to learn more about the ways to balance the use of both TAR and keyword search terms to achieve more efficient, consistent, and cost-effective eDiscovery review.