Watch it on demand by clicking here.
HSR Second Requests are some of the most voluminous, fast-paced, high-stakes, and costly ediscovery responses in the industry. Technology Assisted Review (TAR) is a powerful tool used to tackle many of these constraints. While challenges around timing, cost, and data volumes are not unique to HSR Second Requests, they are certainly amplified in this setting. TAR approaches and applications continue to evolve, and there is certainly not a one-size-fits-all approach to TAR for HSR Second Requests. One decision point for TAR experts and antitrust attorneys is which TAR approach to use: TAR 1.0 or TAR 2.0. Both approaches have been used to comply with HSR Second Requests to both the DOJ and the FTC, so what is different about these two approaches and when would a client want to choose one approach over the other?
TAR 1.0 v 2.0
First, let’s start by addressing the question: what are TAR 1.0 and 2.0? At a high level, TAR 1.0 can be defined as simple learning, where attorneys train the system upfront, develop the model, and then apply the learning to the broader dataset. TAR 2.0 involves continuous learning, where the model learns and continuously updates throughout the review.
Generally, a 2.0 model provides more flexibility than a 1.0 model when needing to get started quickly with their review, and when introducing new data sources into a TAR workflow that is already underway. As opposed to investing significant time upfront in training rounds and control sets by subject-matter experts (SMEs) (typically higher-cost firm attorneys) in a 1.0 model, review teams can immediately jump in and start to review in a 2.0 model. This speed and flexibility can be critical during a HSR Second Request when time is of the essence. Negotiations with regulators around custodians and collections can drag on for weeks, and those negotiations can result in additional custodians and data sources requiring incorporation into a TAR workflow that is already off to the races.
TAR 1.0 has significant precedence as an accepted model by the DOJ and the FTC, so negotiations with the government around approach and validation can be streamlined when using a 1.0 model. Another potential advantage of TAR 1.0 is getting to your responsive set and stabilizing a model with fewer overall documents requiring review when compared to a 2.0 model. For example, Lighthouse has seen cases where as few as 5,000 documents were required to stabilize a TAR 1.0 model for a 1 million total document population. While fewer documents may require eyes-on review in a TAR 1.0 model, however, 1.0 requires eyes-on review from higher-cost SMEs, so it may not necessarily translate into overall cost savings.
Negotiating TAR, Validation, and Best Practices
While a TAR 1.0 model may be more familiar to the DOJ and the FTC, the negotiation and validation approaches for each model are essentially the same. Both TAR 1.0 and 2.0 models have been successfully negotiated with the DOJ and FTC. Developing the facts and remaining transparent about your TAR process is critical to a successful negotiation. Collaborating with and involving TAR SMEs both within your firm and from your service provider early on is also very important. Many TAR SMEs have demonstrated success in negotiating directly with these regulators and can leverage historical information and context to drive a successful result.
In terms of best practices, understand that TAR is not a silver bullet for HSR Second Requests. Mobile, chat, and other collaboration platform data sources are increasingly under the microscope of regulators and, depending on the data type, may not be amenable to TAR. It’s critical that the case team understands their data and is prepared to carve out separate workflows to address non-standard data types, foreign language, and other corner cases. Privilege review and privilege logging continue to be a significant burden for HSR Second Requests, even after TAR streamlines the binary classification of responsive vs not responsive documents. Antitrust attorneys and ediscovery providers alike need to continue to leverage other technologies and best practices to streamline privilege review and logging efforts.
TAR 1.0 and 2.0 can both be used for HSR Second Requests, and each model has its unique advantages and disadvantages. The speed at which one can get started with a 2.0 model without upfront training and control sets and the flexibility it offers for incorporating new data sets midstream make it a very attractive option in a HSR Second Request context where timing is so critical. Leading antitrust attorneys are comfortable negotiating the 2.0 model with government agencies, and have demonstrated success validating a TAR 2.0 process. The TAR 1.0 model still has its advantages with its well-established validation processes and potential for getting to a stabilized model (and responsive set) with fewer documents requiring eyes-on review. It will be interesting to see how TAR 1.0’s lack of agility and nimbleness will play out over time, and if we will continue to see a trend towards TAR 2.0 in HSR Second Requests.
To discuss this topic further, please feel free to reach out to me at firstname.lastname@example.org.