eDJ Migrated

Recently, Recommind briefed eDiscoveryJournal on the software vendor’s predictive coding. In the Recommind context, predictive coding starts with a sub-set of data (derived by various techniques such as concept searching, phrase identification, keyword searching, metadata filters, etc) and users review and code the seed data set for factors such as responsiveness, issue, and privilege. Once that review is complete, the user can hit a “train” button that tells the Axcelerate application to identify conceptually similar documents based on the attributes of the first set of coding. Recommind refers to this as machine learning – the engine learns from the document coding conducted by humans; and vice-versa, with predictive coding the human reviewers also learn from the suggested relevant documents that are returned by the machine. Basically, it is a process where reviewers are presented with more relevant documents, more often and see much less non-relevant document that slow down the process of completing review. There are checks built in so that case managers can continue to review sets of potentially low relevance documents. If any of those documents are in fact responsive, they are re-coded and the system can apply this learning back to the rest of the corpus.

Our earlier post on predictive coding potentially being the future of document review included a poll on whether or not predictive coding is defensible. The results so far are somewhat surprising in that the majority of folks believe predictive coding is not defensible. I suppose I shouldn’t be surprised; the legal industry is slow to change (and for good reason – it makes sense to proceed with caution when going into unchartered territory).

Analysts are told they are crazy all the time (and in some cases, it’s probably true). Years ago, I suggested that the business models of law firms would have to change because of the realities of eDiscovery. Most people I knew told me I was crazy. I even knew some general counsel that told me that if they were still paying hourly fees to their law firms in five years (this was circa 2005), that they would have failed in their jobs. Meanwhile, the law firm folks I talked to said it didn’t matter what corporations did – good law firms can command the hourly rates that they want. The recent Fulbright and Jaworski Litigation Trends Survey results pointed to an increase in fixed fee engagements, so it looks like the business model is, in fact, changing.

Consultants and analysts fly a lot. It is part of the job, but not a part that I particularly enjoy in this age of overbooked flights and mandatory security theater. I am lucky enough to have the flexibility to push several December engagements into January so that I do not have to fly while the new security procedures are debated. I like to think of it as waiting for the first service pack instead of upgrading to a new full release from Microsoft. So how does this relate to eDiscovery? Think of the TSA screening process as a search and review challenge. You have millions of unrelated travelers (substitute email here) with a very few bad actors that have to be identified and acted on. This is remarkably similar to what a corporation with an immature eDiscovery process goes through with any major case. The TSA is in the same position as the corporation’s law firm. The corporation dumps millions of email on the firm and then makes the senior partner (politician) responsible for any privileged or trade secret email that accidentally leak through the review.

In the normal course of business, I am excited to see EDRM project content incorporated into caselaw, articles, research and by other experts in eDiscovery, especially when it is a piece that I contributed to. In a recent Law.com piece titled “A Strategy to Sample All the ESI You Need” attorney Nick Brestoff leveraged Section 9.5 of the EDRM Search Guide to propose forcing the opposition to produce samples of ‘irrelevant or nonresponsive’ ESI. His proposal is a stark reminder of the adversarial nature of our business. As one of the primary contributors on the validation sections of the Search Guide, I can assure you that I envisioned the producing party using sampling and the other methods to maximize precision, accuracy and completeness of discovery searches. I agree with Mr. Brestoff’s demand that parties disclose the scope, confidence levels and criteria used to sample. However, he basically challenges the producing parties ability or right to determine relevance and presents ‘three easy steps’ to demand all ESI not produced. That seems to effectively negate relevance review completely and excludes only duplicates, system files and privileged ESI.

Barry’s article, “Who Owns eDiscovery?” focused on the challenges of corporate investment in the overall eDiscovery process without clear ownership between IT, Legal, Compliance, Records Management and even the business units. It occurred to me that there is an entirely different ownership battle being waged over who is actually managing and performing the actual document review. The law firms have traditionally owned the review process and born the risk of inadvertent production of privileged and unprotected confidential documents. With eDiscovery slowly evolving from an ad hoc reactive fire drill into a managed business process, corporations have begun to take back the upstream or ‘EDRM Left’ phases of the eDiscovery lifecycle. This is happening at the same time as public corporations are experimenting with Business Process Outsourcing (BPO) and SAAS solutions for infrastructure and internal services.

While conducting my discovery scenario testing on Exchange 2010, I found that Microsoft had made two steps forward along with what seems like several strange steps back. In the early days of corporate networks, personal computers and enterprise software, administration was reserved for wizardly geeks who had mastered various esoteric command line languages. I recall the blessed feeling of relief when I stumbled through my first administrative GUI (Graphical User Interface). The admin GUI brought mastery of systems and applications into the realm of the merely mortal user. When evaluating software, I tend to view applications that force the user to learn command switches and syntax as immature. As a counterpoint, I acknowledge that the command line functionality is fantastic for an advanced user to run batch scripts or even automate functionality. When I first looked at Exchange 2010 SP1 Beta (only version available at the time), I was astounded to find that they had killed off ExMerge, the administrative utility used for years to import and export PST files from mailboxes.

So what is a ‘white paper’? Wikipedia says that they are authoritative reports or guides on a specific issue, which sounds great. Just ‘below the fold’ on the definition we find that since the 1990’s ‘commercial white papers’ have become marketing or sales tools designed to promote a specific company’s solutions or products. I think that the eDiscovery market is cynical enough to understand the embedded perspective of most white papers. This bias or perspective does not invalidate the content of good papers, but it does force the reader to filter and interpret carefully. Don’t get me wrong here, I have contributed my share of commissioned white papers to the body of eDiscovery perspective. Every one was a bit of a struggle with the ‘client’ to keep my content free of market messaging. It was exactly this kind of ethical and professional conflict that drove us to create eDiscoveryJournal in the first place.

Barry’s reflections on the 2010 market space got me thinking about last year’s technology trends. We saw a lot of privately developed review platforms rolled out with different licensing models, but I wanted to figure out what was fundamentally new and different. No. 1 ECA = ESI AccessWell before LTNY 2010, the marketing machines began to hype Early Case Assessment (or Early Data Assessment) as the new eDiscovery usage case for corporations and law firms. But what is the fundamental function that enables ECA/EDA? I posit that it is the ability of a relatively non-technical user to directly access ESI, whether in the wild or from a collection. This basic marriage of search indexing and a review GUI gave Clearwell a big jump on the market, followed closely by StoredIQ, Kazeon and a host of others. Effectively, we are foreshadowing the ‘death’ of processing as a separate EDRM phase. My main concern about merging processing into a one-click function is that it can over-simplify complex options and bury exception handling. As long as users understand the different quality requirements for Identification/Investigation versus the actual discovery request, ECA/EDA tools are here to stay.

The clear trend is that corporations are taking control of eDiscovery and bringing functions on the left side of the EDRM – identification, collection, preservation, and elements of processing, analysis, and review – in house. The goal is a more proactive approach to an inherently reactive process. But, the evolution is still a slow one. Inevitably, most corporations still take a matter-by-matter approach to eDiscovery.