A U.K. court ruled this week that using humans to review 3 million documents would carry "unreasonable” costs and ordered the use of culling software, a sign of the ascending role that technology plays in the discovery process throughout the world.
Master Matthews of the High Court of Justice, Chancery Division in London, characterized his 14-page decision in the case, Pyrrho Investments Limited v. MWB Property Limited, as the first in the U.K. to explicitly approve predictive coding software, and he noted judges in both the U.S. and Ireland that have already done so.
The case centers on a breach of fiduciary duty business dispute between several businesses and individuals, in which an original set of 17 million documents had been reduced — by removing duplicates — to roughly 3 million documents that need review for relevance and possible disclosure to the other side.
“The cost of manually searching these documents would be enormous, amounting to several million pounds at least,” Matthews wrote. “In my judgement, therefore, a full manual review of each document would be ‘unreasonable’... at least where a suitable automated alternative exists.”
He estimated the cost of review with “predictive coding software” could run as high as $672,000, plus monthly hosting costs of $30,000 — which he characterized as “obviously” less expensive than a full manual review by humans without technology.
Throughout the order, Matthews uses several terms — such as “technology-assisted review” and predictive coding — interchangeably to refer to proprietary software that can analyze and score documents for relevance to the issues in the case, rather than delegating such tasks to humans alone. One main benefit of such software. he wrote, is that the cost of review does not increase at the same rate as the number of documents to be reviewed.
“Doubling the number of documents does not double the cost,” Matthews wrote, adding “Primitive versions of this kind of process were being demonstrated to [sometimes skeptical] litigation lawyers in the mid-1980s.”
David Horrigan, legal content director at kCura who reviewed the opinion, said the parties in the case are not bound to use technology-assisted review, but at this point have all agreed to it.
The case is not scheduled to go to trial until June 2017, according to the opinion.
Horrigan said some U.K. lawyers already used technology-assisted review, but many resisted out of fear that the technology would be indefensible in the event that a mistake was made — for example, privileged material was accidentally turned over or critical documents were not produced. The concern was that such a mistake could land a lawyer in disciplinary proceedings or a malpractice suit.
“The lawyers have a greenlight now,” he said. “It definitely was being used, but this ought to really jumpstart it.”
Tom Barnett, a data scientist at Paul Hastings, said it is important to note that discovery obligations in the U.K., known as e-disclosure, are not as broad as in the U.S.
“I think this is a really important recognition by the court that even in a setting where the scope of whats expected to be produced is less, the amount of data is so vast,” he said.
Emma Kettleton, a director at the London-based e-Discovery vendor Millnet, which is working with one of the law firms involved in the case, said using technology-assisted review to cull a document set “isn’t a push button solution,” in that you still need trained-individuals to manage the process.
But the skill sets required for document review are changing, so it’s no longer just lawyers reviewing documents, according to Kettleton. Now, in addition to attorneys, the process requires people who understand the technology and who understand statistics well enough to know if the machines are reliably culling documents,” she said.
“It’s utilizing technology, and experts at the technology to achieve a result,” said Kettleton.