Migrated from eDJGroupInc.com. Author: Michael Fluhr. Published: 2013-05-08 05:00:02Format, images and links may no longer function correctly. This is the continuation of my article from last week entitled “Partial Recall: Why Lawyers Can’t Have (And Really Don’t Need) All Relevant Documents”
I find it exceptional the Court found it reasonable and proportional for Biomet to follow eDiscovery procedures predicted to result in the production of only 40% of the relevant documents (40% recall)—a number that might make some attorneys and judges cringe. I don’t know of any other case that has so clearly approved of such a low recall value. And I think the Court was right to do so.
For one thing, studies show that even using the best information retrieval technologies at extraordinary expense, recall in many cases doesn’t reach much higher than 70%. Perfection or near-perfection are not even options.
Additionally, historical eDiscovery practices have likely resulted in significantly lower recall values. As many commentators have noted, parties frequently (even still) filter document populations using untested, blindly constructed full-text searches—now known to result in abysmally low rates of recall. And, of course, the past 20 years have seen the explosion of human document review, also now known to be quite unreliable. Thus, we’ve likely accepted recall rates well below 40% for decades (though we didn’t know it).
Finally, and I think this deserves future study, I question the marginal utility of higher recall rates. The widely held belief among attorneys that they need all or nearly all relevant documents in a case stands quite at odds with the way pretty much everybody else consumes information. Just as an example, if I want to learn a bit about how transmissions work (much of my caseload deals with vehicles), I will Google “transmission”, read a few articles, and be done with it. I don’t need to read every book, article, or word ever written about transmissions. Even if you wanted a masters level understanding about a complex subject, say French history, you still wouldn’t (and couldn’t) read every piece of material on the subject. Certainly if you tried, much of the material would be redundant. The question is, why do we believe that our normal, tried-and-true practical methods of information collection and analysis can work for doctors, and engineers, and historians, and scientists—but not lawyers? Why must we consume and analyze every single possibly relevant document—without regard to whether it actually aids our understanding of a case? I’d love to see a study of medium and large cases analyzing the ways in which the majority of crucial documents (those used at summary judgment or trial) were located. Was it during early case assessment? The first hours of rigorous document search? Or am I wrong, and were they found in the 10,000th hour of document review. Ultimately, we’ve lost our innocence and in the stark light of day must face head-on the reality of partial recall. For this reason, I expect that courts will increasingly and explicitly accept recall rates of 70%, 60%, 50%, or lower. And that’s a good thing.
eDiscoveryJournal Contributor Michael Fluhr*
*The opinions expressed in this commentary are solely those of Michael Fluhr