Just when I thought I had heard every excuse possible, along comes a case with a brand new guffaw. Northstar Marine, Inc. v. Huffman, Case 1:13-cv-00037-WS-C (Ala. S.D., 08/27/13). Although both sides stumbled with e-discovery in this Admiralty case, the truly jaw gaping goof comes from the plaintiff’s counsel. He claimed he could not find an e-discovery vendor to provide computer-assisted search. Yes, that is a new excuse for me. But wait, it gets worse. Apparently plaintiff’s counsel was looking for real deals on not-so-fancy wheels.
Here is the quick background. The parties entered into a stipulation where each side agreed to use Computer Assisted Review (CAR) to search and produce documents to each other. You might assume these Alabama attorneys were pretty sophisticated, and ahead of the curve, to specify predictive coding. That is, after all, what most attorneys would think was intended by a CAR stipulation. But no. Nothing could be further from the truth. We are not talking about fancy wheels here at all.
They agreed to CARs alright, but the old Model-T buggy types with keyword search engines. Their stipulation had nothing to do with predictive coding, or any type of advanced analytics. It did not even have anything to do with the proper use of keyword technology, as dated as it is. They were stipulating to Go Fish type keyword guessing production, the kind I dissed years ago. Losey, R., Adventures in Electronic Discovery (West 2011); Child’s Game of ‘Go Fish’ is a Poor Model for e-Discovery Search. This kind of child’s guessing game is based on blind hope and irrationality.
I thought nearly everyone had abandoned that approach. But no. Here it is, alive and well in Admiralty Court in Alabama. Both sides stipulated to a set of keywords, with no apparent testing in advance at all, and moreover agreed to produce to each other all documents with those keywords in them. Here is the court’s description of the parties stipulation to use antique CARs and long discredited driving methods:
Among other things, the parties agreed that:
Both parties have or will immediately arrange to use computer-assisted search technology that permits efficient gathering of documents, de-duplication, maintaining the relationship between emails and attachments, full text Boolean searches of all documents in one pass, segregation or tagging of the search results, and export of all responsive files without cost to the other party. Both parties shall share with the other party the specific capabilities of their proposed computer-assisted search technology, and will endeavor to agree on the technology to be deployed by the other party.
In addition, the parties agreed to use certain search terms and agreed that “[a]ll documents in the search result sets shall be produced immediately to the other side in native format including all metadata.” … On June 11, 2013, the Court entered a Supplemental Rule 16(b) Scheduling Order (Doc. 22) adopting the parties’ plan with regard to ESI.
Ok, all of this is not exactly smart search, but all of my readers have seen and heard this kind of Go Fish stuff before, many times. But I doubt you have seen the plaintiff’s excuse for not complying with the stipulation, namely cannot find a vendor who can do such sophisticated things as “deduplication,” boy that must be hard, or “full text Boolean searches” and not only that, “of all documents in one pass.” I love that phrase. Full text in one pass? Incredible! And maintain the relationship between emails and attachments at the same time! Are there any vendors out there that can meet such stringent requirements?
Now if counsel had figured out that the stipulation was stupid and should be changed, or maybe even if he had objected on the basis of expense, but no, he just says he cannot find a vendor. That’s right. He claims he cannot find any CARs to drive, even simple keyword search engine powered CARs. Ok, to be fair, he does qualify that with “affordable” CARs, so maybe he thought you could just pick up a used CAR for a couple of hundred dollars or something. Here is how the court describes this rather pathetic excuse not to produce documents:
While the plaintiff has not filed a motion requesting modification of the Rule 16(b) Scheduling Order, it should have filed such a motion given its ongoing discovery delays. In any event, the Court finds that there is no showing of due diligence or good cause to support a modification of the order. The initial disclosures required by Fed. R. Civ. P. 26(a) were originally due to be exchanged by May 10, 2013. … However, ESI was not produced at that time because the parties had not yet conferred and agreed upon the parameters of ESI discovery. … The parties subsequently reached an agreement regarding ESI discovery and, on June 10, 2013, the parties filed a joint supplemental report stating that all ESI documents returned after performing the searches agreed upon by the parties will be produced immediately. … To date, the plaintiff has not even begun collecting its ESI material because it is still attempting to locate an inexpensive electronic search technology provider to assist with the process. … Plaintiff’s attempts to find an inexpensive provider certainly do not constitute due diligence.
Perhaps plaintiff’s counsel should get on the Ball and read Craig’s articles on how to do e-discovery on a shoestring budget with cheap, readily available software. See Ball, Craig, E-Discovery for Everybody: The EDna Challenge (2009).
Conclusion
There are at least two morals to this Labor Day 2013 short story. Be careful what you stipulate to. Court’s do tend to require that you keep your word, and if you cannot, that you should at least file some kind of objection, such as I did not realize how expensive my stipulation was, or maybe, I had no idea that the keywords would be in over a million documents, I did not test them in advance. Do not just shuffle around and claim that due diligence did not turn up any vendors with keyword search capacities. There are more vendors out there with such modest capabilities than you can shake a stick at. Even most law firms claim to be able to do it, and they are just supposed to provide legal services, right?
The second moral is a little more tricky, but also more fun. Many cases that hit your standing keyword search for CAR and TAR may have nothing whatsoever to do with predictive coding, even ones that says computer-assisted like Northstar Marine does. At least the CAR false positives usually have something to do with legal search. Just think what the false positives for TAR must be like – I’m not going there. Yet another reason to use the CAR acronym, and not TAR.
In reality, both are not very good word-handles for predictive coding, as the Northstar Marine opinion that went the rounds of new predictive coding cases last week illustrates. It does not take much for a review of documents to be considered computer-assisted, or technology-assisted. Using a computer for review is no big deal, and neither is using some kind of technology. An air conditioner is, after all, technology, correct? So if you did a paper review in an air conditioned room, would it not be technology assisted? So I am tired of saying TAR or CAR. Moreover, according to the ultimate TAR glossary by Cormack and Grossman, they both mean either rule based linguistics or machine learning:
Some TAR methods use Machine Learning Algorithms to distinguish Relevant from Non-Relevant Documents, based on Training Examples Coded as Relevant or Non-Relevant by the Subject Matter Experts(s), while other TAR methods derive systematic Rules that emulate the expert(s)’ decision-making process.
Since the only kind of predictive coding most people care about is machine learning, not rule based, it is just too vague and imprecise to use the keywords CAR or TAR, since they could mean either. So from now on I’m just going to say active machine learning or predictive coding. Please join me. Let’s not get stuck on TAR. And please, will not one of you good e-disco CAR salesman call the plaintiff’s counsel in Northstar and make him an offer he cannot refuse? I know full well you have some clunkers in the back that you really ought to get off your lot.
Reading the supplement to 26(f) report it does not look like the agreed to use CAR or TAR – where the R means Review. They agreed to use Computer Assisted SEARCH. So you assume too much when you assume “what most attorneys would think was intended by a CAR stipulation.”
It’s a fair enough criticism that they are producing every document that hit on a search term, but the 26(f) report provides for tuning the keywords and custodian lists.
The other interesting tidbit is that Plaintiff’s counsel thinks the $3,000 price quote is high. The relief requested is $1,000,000.
I RECAP’ed a bunch of the docs here:
http://ia601608.us.archive.org/34/items/gov.uscourts.alsd.53527/gov.uscourts.alsd.53527.docket.html
Thanks. P’s idea of proportionality is another one for the record books. He thinks $3,000 is too much to spend for a million dollar case!
Ralph – I appreciate the work that you do in this area (CAR/TAR). Being on the vendor side of things, I see this scenario far too often. Counsel focuses on complying with a basic agreement (deduplication, maintaining family relationships, etc.) and miss the bus when it working efficiently. Perhaps the nextgen of attorneys will fill the void and move the industry in a positive direction as a whole.
[…] Losey’s article Poor Plaintiff’s Counsel, Can’t Even Find a CAR, Much Less Drive One tells of an Alabama law firm which entered into an eDiscovery agreement with opponents on terms, […]