Since my server died with the entire collection of DocReviewMD blog posts, I figured I would update these posts based on where the predictive coding field is today and treat these posts as a re-release. I remember drafting this post last year when I thought rather naively that we had just changed the litigation world. Global Aerospace then continued with the predictive coding process being run, meet and confers with the opposing lawyers, and an agreement to produce the documents we coded using predictive coding. While the case eventually settled, Global Aerospace for me stands as the best example of a successful predictive coding case. We got the order we needed to use the process when opposing counsel wouldn’t agree to use predictive coding. We then ran the software program and trained it using an expert reviewer who was a partner in the law firm’s construction litigation group. We stabilized the tool then met with opposing counsel and talked through how we coded documents including sharing the null set. Opposing counsel disagreed on a small number of documents we had coded not responsive. We added these documents into the language model and pushed the button. No other case has made it this far. I believe it was the transparent approach we followed to make discovery about getting agreement on the process which is best accomplished when you are transparent. Contrast that with Da Silva Moore which is still an ongoing case, and you can get a sense of how successful we were. The end result is we culled the 1.3 million documents in the collection to 200,000 documents with family members included and got the review done with a handful of attorneys. A great outcome which resulted in two Wall Street Journal articles. Yet for some reason today, the acceptance of predictive coding hasn’t taken off. Which is one reason why we have been touring the country offering inexpensive CLE’s through the Predictive Coding Thought Leadership Series on how to validate predictive coding results. Read on and enjoy….
Originally posted on April 25, 2012
Setting the Record Straight on Global Aerospace and Predictive Coding
When the order was signed on Monday at 4PM on April 23rd in the Global Aerospace, Inc. v. Landow Aviation, LP case in Virginia State court, we understood the significance of the moment. For the first time, a court was issuing an order in the face of an adversary who did not want predictive coding to be used that the defendant should be allowed to use predictive coding. We also understood that there were only 20 people in the room and it was possible that there could be some confusion over the case in the rush to get the story onto the Internet. While I disagree with much of what has happened in Da Silva Moore case in NY, I do agree that the lobby supporting predictive coding and the interest the legal community has in this topic does attract a significant following of readers. So now that many of the pundits have weighed in on the impact, I am going to try and play clean up duty and write what I see as the impact of Global Aerospace, with the advantage that I was there. My analysis will be restricted to facts with some clean up on a few posts by some of my blogging friends who attract a good deal of attention on e-discovery issues.
One of the initial mistakes is the observation that this is “the second case” to approve of the use of predictive coding. As I posted in an analysis of Da Silva Moore, which some opine is the first case to approve the use of predictive coding, the parties in that case were not contesting the use of predictive coding. At the time of the agreement to use predictive coding, the only dispute was how to validate the results of predictive coding. Secondly, the Da Silva Moore case involved a Judge who is well recognized as a proponent of using technology such as predictive coding to make electronic discovery more efficient.
Contrast that with Global Aerospace where the plaintiff was opposed to using predictive coding and submitted a brief and oral argument fighting the use of these tools in this particular case. Nor were the Judge’s predilections concerning the use technology assisted review or predictive coding as well known. So when the parties walked into the court room, all that was known was the opponent was saying the defendant should not be allowed to use predictive coding. In a nutshell, this was not like “dying and going to heaven” as Judge Peck told the defendant advocating predictive coding use and for that reason, Global Aerospace is indeed the first case anywhere where the Court issued a protective order allowing the use of predictive coding over the objection of opposing counsel. So like most discovery matters, this defendant will be able to review the documents as they see fit, and the plaintiff will need to wait till they receive data before they can challenge the production.
Some analysis have shrugged and said this case reflects the fact that judges may not care about discovery so they might sign off on any reasonable approach that a party proposes. The other salient point from Global Aerospace is the efficiency of the decision and approach proposed. In effect, this case avoids the traditional key word dance that parties can spend weeks, months or longer fighting about. That is extremely unproductive time in the process. The analogies used in this case showed a complete understanding by the Judge that advancing into technology assisted review is nothing different from the first time 40 years ago when a partner in a law firm said “I am going to let my associate review the documents.” Another example given was when paralegals were used. If the progression is looked at, the Court made a reasoned decision that without seeing problems in the output, it would be acceptable for a defendant to try an approach which the defendant believes is much more efficient. So while no one knows for sure what the Judge was thinking, he provided a clear opportunity for the parties to try and improve the process.
That being said, the transcript is well worth reading. I would suggest ordering it as it is a lot shorter than the other transcripts flying around the Internet in Da Silva Moore and Kleen. Predictive coding or technology assisted review is an evolution and there will be more opportunities for parties and judges to grapple with predictive coding issues. I also suggest taking the time to watch these cases argued. I have now seen two of the cases argued and there are lots of differences between the cases and the approaches being taken in making the case to use predictive coding when at least one of the parties wants these tools to be used.
- DocReviewMD AKA Karl Schieneman