On January 1, the U.K. launched a disclosure pilot program. The program elevates technology-assisted review — often referred to as predictive coding — to a prominent role in English and Welsh business and property courts.
Initial disclosure and Extended Disclosure
A major goal of the pilot is to determine how to manage disclosure. Under current practice, standard disclosure can produce numerous documents, which can create review costs that are disproportionate relative to the case. The pilot replaces standard disclosure with two new types: initial disclosure and extended disclosure.
In initial disclosure, parties prepare a list and copies of key documents that are necessary for the opposing side to understand the case. Parties also have the option to choose the second type —– extended disclosure.
Technology-Assisted Review (TAR)
The request for extended disclosure comes through something called the “disclosure review document” and both parties complete the request before the first case management conference. The document details which issues require extended disclosure and which type of extended disclosure is needed (there are five models of extended disclosure). When the extended disclosure requires searches, the parties must agree on the use of technology assisted review (TAR). If the parties decide against using TAR, particularly in larger matters (more than 50,000 documents), they must explain their reasons in the disclosure review document.
While TAR is a useful tool, it is not appropriate in every case. Here are some reasons why parties may forego using TAR:
The material to be reviewed contains many numbers or images. TAR uses text to code and identify responsive documents, so the documents not only need to be predominantly text-based, but the text must provide enough semantic content for the software to be able to analyze characteristics and meaningful patterns.
TAR works best with documents produced by word-processing programs, emails, and text-based slides. Predictive coding may not be appropriate for spreadsheets, computer-aided design files, blueprints, schematics, audio, photos, and video. Other types of data that may not be TAR-friendly could include exports from structured databases, Outlook calendar invitations (unless they include a lot of text), and hard-copy documents that are difficult to read with optical character recognition.
The material is in another language. If documents are not in English or written by non-native English speakers, TAR may be more time-consuming because the software might require separate training sets for other languages in order to be effective.
The number of documents is low. For example, if the reviewable material involves fewer than 20,000 documents, TAR may not be as cost-efficient as other review methods. TAR requires a significant amount of preparation time — to develop and implement a workflow, review a training set of documents, conduct quality control, and perform other tasks. In addition, the more documents to analyze, the more effective TAR is. A small document collection may be relatively rich in relevant documents but because of its size could lessens the value of using TAR.
The case is on a fast track. Since TAR require time to prepare, it may not be appropriate if there are pressing deadlines for documents to be produced or people deposed.
Only a few years ago, U.K. courts were hesitant to endorse the use of predictive coding. Now, courts expect parties to use TAR unless the parties can give valid reasons for not doing so. The U.K. pilot indicates a dramatic switch in the acceptance of TAR practices, which is another signal that TAR is on its way to being a new, legal custom.