I’m rarely moved to criticize the work of other commentators because, even when I don’t share their views, I applaud the airing of the issues their efforts bring. But sometimes a proposition is just so blatantly ill-advised, so prone to unfairly tilt the litigation playing field, that any reader and every writer should stop and say, “Wait a second….” One such article, currently running in the New York Law Journal and called No Disclosure: Why Search Terms Are Worthy of Court’s Protection, charges that judges who require disclosure of search terms “discount or misunderstand” what the authors term the “protected nature of key aspects of the e-discovery process,” namely filtering of data by use of search terms. The authors think that disclosure of search terms used to exclude data from disclosure compromises the work product privilege and argue that judges should “recognize that a search term is more than a collection of words, rather, the culmination of an attorney’s interaction with the facts of the case.”
Espousing the sanctity of work product privilege to an audience of litigators is like saying, “I support our troops.” It’s mom, baseball and apple pie. It’s also popular to paint judges as addled abusers of discretion. But let’s not let jingoism displace judgment. Search terms are precisely what the authors claim they are not: search terms are a collection of words. They are lexical filters. Nothing more.
Search terms deserve no more protection from disclosure than date ranges, file types and other mechanical means employed to exclude data from scrutiny. Search terms strip out information that will never see the light of day nor benefit from the application of lawyer judgment as to their relevance. In that sense, search terms are anathema to the core principles of work product and warrant more, not less, scrutiny.
The authors of the article point to FRCP Rule 26(b)(3) and its language that courts “must protect against disclosure of the mental impressions, conclusions, opinions, or legal theories of a party’s attorney or other representative concerning the litigation.” True, but the authors elide over the other salient points of the Rule, including the requirement that parties seeking to resist disclosure must expressly make the claim AND furnish information about the information sought to be protected that “will enable other parties to assess the claim.” You can’t just claim something is work product. You’ve got to prove it. The authors also fail to address the provision in the Rule that requires disclosure if the information sought is otherwise discoverable or the party seeking discovery “shows that it has substantial need for the materials to prepare its case and cannot, without undue hardship, obtain their substantial equivalent by other means.”
By what other means might a requesting party secure the substantial equivalent of search terms used to exclude potentially responsive material—again, excluding same sight unseen—save by getting them through discovery?
But the overarching reason why the work product argument fails is because it’s based on a false premise. Disclosure of mechanical filters doesn’t compromise counsel’s mental impressions, conclusions, opinions, or legal theories. No one suggests that the searches a lawyer runs to test theories and form mental impressions are per se discoverable. How a lawyer serves to inform his or her own assessment of the case may be protected. But that’s a different kettle of fish from the use of search terms to objectively filter collections for review and production, and we should be careful not conflate the two because they involve similar techniques.
Any student of e-discovery understands that segregation of potentially relevant documents by use of keyword search is a practice fraught with failure. Language is nuanced, and meaning subtle. Fashioning lexical queries that perform well, i.e., that produce estimable precision and recall, is not a task lawyers are trained to undertake or one to which lawyers bring special expertise. Sorry to burst that bubble. In fact, the process by which lawyers typically arrive at search terms in litigation would be pretty funny if it weren’t so pitiful: lawyers mostly guess. Call it “judgment,” if you wish; but, mostly it’s a wild ass guess.
Good lawyers go much further, of course. They test those guesses against representative samples of potentially responsive data, they look closely at the results, winnow away false positives and tweak queries until they are demonstrably more effective and efficient. This is where judgment may come into play, and it’s the part of the process that may deserve limited work product protection. But, the terms that emerge—the terms actually run to mechanically segregate information items–are not worthy of courts’ protection.
Because if integrity of process means anything in e-discovery, it means that opponents should be able to understand, test and challenge error-prone mechanisms pushing potentially responsive information beyond the reach of attorney scrutiny. The reasons why a party selected particular search terms may not be discoverable; but, the terms used should be freely discoverable absent good cause shown. If search terms are not routinely disclosed, then much of the waste, abuse and failure of e-discovery will continue unchallenged, safely ensconced beneath the misapplied mantle of work product protection.
Let’s be honest: Parties don’t seek to conceal search terms because they reveal what their lawyers think. They seek to hide search terms because they’d rather opponents not see how thoughtless they were.
Andy Wilson (Logik) said:
Awesome post, Craig. Confidimus autem de certa.
LikeLike
bmschulman said:
Craig: Search terms may reveal substantive, protected communications. Illustration: A client in an employment litigation tells counsel about a rumor that a specific and highly offensive nickname may have been used in conversation to refer to plaintiff behind plaintiff’s back. It seems prudent (perhaps even required) to search relevant custodian emails for that nickname, but counsel certainly does not want to disclose to the adversary that he ran that search (regardless of results), particularly at the outset before documents are even reviewed for actual responsiveness and privilege. The use of that term reveals the content of an attorney-client communication as well as counsel’s judgment that the search term should be used. In your view, is disclosure required in this instance?
Brendan Schulman
LikeLike
craigball said:
Brendan: I think you make a persuasive case of good cause for withholding the offensive term on peculiar facts. But, here again, I distinguish between keywords run by counsel for the purpose of informing themselves about the case and terms run to exclude from production information items that do not contain the search terms. These endeavors are not two sides of the same coin.
If the offensive document turns out to exist and is responsive, will it be produced in your scenario or will it be hidden beneath the skirts of privilege? If counsel acts ethically, what, then, is ultimately gained by failing to include the term in the list of disclosed terms use to identify (patently relevant) information items? How does counsel respond when the other side says, “Why didn’t you run a search for [offensive term]?”
Many things we must disclose in litigation are the fruits of attorney-client communications or provide insight into counsel’s judgment. A word is just a word, though the rationale behind its selection may lie beyond the pale of discovery.
LikeLike
bmschulman said:
I don’t think it’s appropriate to assume that a responsive but substantively harmful document would be improperly withheld. Our system has long been based on the premise that we are officers of the court and produce documents in good faith, just as we create search terms or impose any other reasonable limitation on our document review procedures. The rules place this process in our hands in the first instance. Your first paragraph is extremely dismissive, and I am suggesting that there is an argument for work product protection that should not so readily be dismissed. It seems there is at least one example as to which you agree; I’m sure we could think of others
LikeLike
craigball said:
The premise that we, as officers of the Court, will produce responsive, non-privileged documents assumes that we will first see the documents that we might elect to produce. Keyword search turns that assumption on its head. Just one more compelling reason why keywords demand disclosure.
I’m sure we could think of other exceptions as even the most ironclad rule may brook some exception borne of fertile legal minds. Yet, shouldn’t we build sound practice more on the usual than the exceptional or contrived? I left room for exception under the rubric of ‘good cause shown;’ but, the norm should be transparency of process absent showing of exceptional need to suppress, not suppression without cause. What you call “extremely dismissive” was actually the exercise of some restraint on my part. Pulling no punches, the assertion that search terms *must* be deemed protected work product suggests a mindset of those who have abused the discovery process for so long that they’ve come to see obstruction as an entitlement. Tortured application of privilege is the last refuge of a scoundrel (with apologies to Samuel Johnson).
Brendan, I have not dismissed the proposition readily. I’ve actually mulled it over quite some, even if that doesn’t come across. I think we just disagree, and, as someone about to be 27 years married with children, I am routinely reminded that I am not always right even when, as now, I am right. 😉
LikeLike
Jeffrey Bannon said:
Doesn’t the answer to bmshulman’s hypothetical depend on whether the search finds documents confirming the rumor? If there are no documents, then revealing the search just discloses the rumor, but if there are documents confirming the rumor, revealing the search doesn’t really leave them any worse off (assuming they don’t intend to cheat).
LikeLike
Michels, Mark (US - San Jose) said:
Wondered whether you saw the article. I guess so.
I know that this is an area of considerable discussion. I am curious to see what reaction you get.
Are you planning on attending LegalTech?
Mark
This message (including any attachments) contains confidential information intended for a specific individual and purpose, and is protected by law. If you are not the intended recipient, you should delete this message and any disclosure, copying, or distribution of this message, or the taking of any action based on it, by you is strictly prohibited.
v.E.1
LikeLike
Mark Greenwald said:
You are very polite. The position of the commentators is insipid and stupid. They have no idea of the practical mechanics involved. Just smart people living on Pluto with time on thier hands and a need to publish.
LikeLike
craigball said:
Dear Mark:
Thank you for reading the post, but the commentators in question are really none of the things you suggest (except smart). I don’t agree with them this time, but they are NOT insipid, stupid, clueless, idle or (insofar as I know) extraterrestrials. They’ve written many fine articles in the past. As to the need to publish, I’m in a glass house there.
LikeLike
mark greenwald said:
As Forrest Gump said: “stupid is as stupid does.” You may be in error on extraterrestrials.
LikeLike
Pingback: e-discovery 2.0 » Blog Archive » In Re: Biomet Order Addresses Hot Button Predictive Coding Issue
Pingback: If Your Legal Hold is Privileged, Is All Content Privileged? – Legal Hold and eDiscovery Blog
Pingback: Could this be privilege? ELF AND Shelf OR El* OR Shel* AND SANTA BUT NOT CLAUS - Computer Forensics and eDiscovery Experts