It’s hard not to feel a bit embarrassed for all the bloggers who await, and write about, nearly everything Magistrate Judge Andrew Peck (S.D.N.Y.) writes about the search for and production of electronically stored information (ESI). We’re far more guarded over here at Discovery Advocate, only commenting a few times (here, here, here, here, here and, um, here).

What do you need to know about Judge Peck’s recent decision in Hyles v. New York City? It builds on multiple themes that appear in his prior opinions. The main point in Hyles is that a requesting party cannot force a responding party to use technology assisted review (TAR) or, for that matter, any other particular method, to search for and produce ESI. This is not a surprise. Judge Peck previously made this point in Rio Tinto PLC v. Vale S.A., relying on some of the same precedent he now invokes in Hyles – the Tax Court’s recently revisited 2014 ruling in Dynamo Holding (“[T]he Court is not normally in the business of dictating to parties the process that they should use when responding to discovery”) and The Sedona Conference Principle 6 (“Responding parties are best situated to evaluate the procedures, methodologies, and technologies appropriate for preserving and producing their own electronically stored information”).

But Judge Peck’s recent treatment of this theme is important because:

  1. It comes with a caveat. While a court typically will not dictate the methodologies and technologies a party must use in responding to discovery, that may change where a requesting party is able to demonstrate deficiencies in the production.
  2. It does not matter that the alternative preferred by the requesting party is the “best and most efficient search tool” or, for that matter, the tool preferred by the court. While Judge Peck believes some TAR methods are the “best and most efficient” to search for relevant ESI, and even states his preference that he “would have liked the City to use TAR in this case,” he concludes: “While Hyles may well be correct that production using keywords may not be as complete as it would be if TAR were used, the standard is not perfection, or using the ‘best’ tool, but whether the search results are reasonable and proportional.” (Citations omitted).
  3. All that said, Judge Peck leaves open the possibility that – someday – certain search and review methodologies may be so demonstrably superior that the use of other methods may no longer be reasonable:

“There may come a time when TAR is so widely used that it might be unreasonable for a party to decline to use TAR. We are not there yet.”

One additional point should be evident from Hyles. Whether a party chooses machine learning tools, keywords or human eyes on the screen, the process used to deploy each tool is as important as the tool itself. Using TAR? What flavor of TAR? Do you need to build a seed set? How will it be done? How, if at all, will data be screened before it is ingested into the TAR tool? What levels of recall and precision are acceptable and why? What form of quality control will be applied? Judge Peck’s decision should not be read as a global statement that TAR is always superior – it assumes, as it must, that your tool of choice is defensibly deployed.

And, of course, the same is true for keywords. If your keyword selection method involves a lawyer sitting at a desk thinking “this is an antitrust case” and writing down keywords like “antitrust,” you’ve got a problem. If you select keywords without knowing the language of your custodians, for example that they use the phrase “the football” to refer to “the business plan,” then you’ve got a problem. But, if you use a keyword selection methodology that involves, as Judge Peck has laid out, “careful thought, quality control, testing, and cooperation with opposing counsel in designing search terms or ‘keywords,’” it’s unlikely you will be forced to do something different . . .

. . . at least for now.