## Facts

- Purchasers of methylene diphenyl diisocyanate (MDI) and toluene diisocyanate (TDI) sued BASF Corporation and other diisocyanates manufacturers (collectively, BASF), alleging antitrust violations related to pricing and supply.
- The actions were consolidated in a multidistrict litigation in the U.S. District Court for the Western District of Pennsylvania.
- During ESI discovery, plaintiffs sought an order compelling BASF to use technology-assisted review (TAR) and to implement plaintiffs’ proposed TAR protocol.
- The parties agreed that an estimated recall rate in the range of about 70–80% would be acceptable, but they disputed how recall should be calculated and validated.
- In typical TAR workflows, parties narrow the documents subject to review by identifying custodians and applying negotiated search terms to create a TAR “review set,” then use TAR to classify documents and test results through sampling.
- BASF proposed a methodology that used custodian and search-term culling to define the TAR review set and then limited validation sampling and recall calculations to that narrowed review set, rather than to the full collection of potentially responsive ESI.
- Plaintiffs proposed a more demanding methodology that would validate against the larger potentially responsive population and continue iterative searching and sampling until BASF could reasonably believe it had produced “substantially all” responsive documents.
- The court appointed an e-discovery special master to address plaintiffs’ motion to compel and BASF’s motion for a protective order regarding the competing TAR protocols.

## Issues

1. Whether plaintiffs were entitled to an order requiring BASF to adopt plaintiffs’ TAR protocol, including validation and recall measurement against the full population of potentially responsive ESI and a production standard aimed at “substantially all” responsive documents.
2. Whether BASF’s proposed TAR protocol—defining a TAR review set through custodian and search-term culling and validating recall within that review set—was a reasonable and proportional discovery method under Federal Rule of Civil Procedure 26(b)(1).

## Decision

- Plaintiffs’ motion to compel adoption of plaintiffs’ proposed TAR methodology was denied.
- BASF’s request to proceed under its proposed TAR approach was granted in substantial part, subject to court supervision and agreed safeguards.
- The court permitted BASF to define a TAR review set using custodians and negotiated search terms and to perform validation sampling and recall estimation within that TAR review set rather than the full ESI corpus.
- The court indicated that plaintiffs could seek additional, targeted discovery later if specific deficiencies emerged from what was produced, but it declined to impose plaintiffs’ up-front, open-ended completeness framework.

## Legal Principles

- A producing party generally may choose a reasonable method to locate and review ESI, including TAR used in combination with search terms and other culling steps, absent a concrete showing that the method is deficient.
- Rule 26(b)(1) requires discovery proportional to the needs of the case; it does not require a process designed to maximize recall regardless of burden and cost.
- Where parties have defined a reviewable universe through custodian selection and search terms, a court may allow TAR validation and recall sampling to be conducted within that agreed review set rather than across the entire collected data set.
- Courts may reject protocols that impose repeated rounds of sampling and re-training intended to approximate “complete” production when the likely marginal gain does not justify the expense and delay.
- Approval of a TAR protocol does not prevent later, focused follow-up discovery if later-developed information shows non-cumulative gaps tied to specific custodians, topics, or time periods.

## Conclusion

In this MDL discovery order, the court refused to require plaintiffs’ full-population, “substantially all documents” TAR framework and instead approved defendants’ more conventional approach: custodian and keyword culling to create a TAR review set, with validation and recall estimated within that set, consistent with Rule 26(b)(1)’s proportionality limits and subject to later targeted supplementation if warranted.


In re Diisocyanates Antitrust Litigation, 2021 WL 4295729 (2021)