The Disclosure Pilot and Technology Assisted Review
June 11, 2018
A working group has recently reviewed the rules by which parties to civil litigation must plan for and provide disclosure. In promoting a new, draft Practice Direction, it has concluded that the regime needs re-structuring. It is expected that the Business and Property Courts in England and Wales will soon commence a two-year pilot of the reforms. The pilot will commence alongside signs that the judiciary is now also placing even greater emphasis on the cost-benefits of deploying automated software in the disclosure process. In this update, we briefly consider some of the headline changes, and the questions that they pose vis-à-vis electronic document review.
Moving away from standard disclosure
Under the extant disclosure rules in multi-track cases, the court can make various orders. These range from dispensing with disclosure, to an order for disclosure on specific issues, to ‘standard disclosure’. In advance of the court’s order, the parties are encouraged to agree a proposal which facilitates the just and proportionate disposal of the claim(s).
In recent times, concerns have developed as to the suitability of this regime. There is a perception that the judiciary has not gone far enough in deploying the breadth of available orders, and a concern that many practitioners are simply treating ‘standard disclosure’ as the default preference. An attempted shift away from this ‘default preference’ perhaps serves as the defining feature of the proposed changes and the pilot.
Standard disclosure requires a party to disclose the documents: (i) on which it relies; (ii) that adversely affect its own case or another party’s case; and (iii) that support another party’s case. ‘Documents’ are defined broadly: any paper or electronic source “in which information of any description is recorded.” A reasonable search for (ii) and (iii) must be carried out.
The question as to what constitutes a reasonable search can be complex vis-à-vis electronic documents. As explained by the then Senior Master of the Queen’s Bench Division, Master Whitaker, in Goodale v Ministry of Justice  EWHC B41:
“… an enormous volume of information is now created… and stored electronically. … The problem is how the parties and (if disputed) the court determines what the scope of that search [for electronic information] should be, [and] how it is going to be made proportionate….”
The proposed changes seek to curtail the occurrence of broad searches that frequently occur under standard disclosure. Under the pilot, parties will need to provide: (i) known adverse documents; and (ii) ‘basic disclosure’. Basic disclosure constitutes the key documents upon which a party relies and which are necessary for the other parties to understand the case to meet.
Notably, the duty to provide known adverse documents does not require an extensive search: it requires a party to consider whether it is aware of the existence of such material. Similarly, basic disclosure can be dispensed with by agreement or by court order. If basic disclosure would result in a party providing more than 500 pages, it falls away.
If a party wants disclosure additional to, or in place of, basic disclosure, it can request ‘extended disclosure’. It can only do so on an issue-by-issue basis, and it must persuade the court that such disclosure is required. In response to a request, the court can make one of five orders: (i) no disclosure; (ii) basic disclosure (to the extent not already provided); (iii) disclosure, pursuant to a specific search-request, for “particular documents or narrow classes of documents”; (iv) disclosure, pursuant to a reasonable and proportionate search, of the documents likely to support or undermine the party’s case or that of another party; and (v) in exceptional cases, disclosure, pursuant to a reasonable and proportionate search, of documents likely to be supportive or adverse, or which “may lead to a train of inquiry which may then result in the identification of other documents for disclosure” (in essence the old “Peruvian Guano” test for disclosure).
The proposed Practice Direction also includes a duty upon the parties, and their representatives, to co-operate in relation to disclosure. This duty has significant requirements relevant to electronic documents. Where the sought-after disclosure requires searches, the parties must discuss and seek to agree refinements by reference, amongst other things, to the use of technology assisted review (“TAR”) software. The express inclusion of this tool in the core procedural rules signifies its growing importance in the litigation process.
The results of the co-operation are to be recorded by the parties in a ‘Disclosure Review Document’ (the “DRD”). Section 2(12) of the draft DRD provides that where the parties decide against TAR, they should set out their reasons. Significantly, paragraph 10.2 of the draft Practice Direction adds: “If a party fails to cooperate and constructively to engage in [the process of agreeing the DRD] the court may make appropriate orders at the case management conference, including the dismissal of any application for Extended Disclosure and/or the adjournment of the case management conference with an adverse order for costs.”
This raises the question of when a party’s approach to the use of TAR might be criticised.
Technology Assisted Review
TAR, or ‘predictive coding’, refers to computer programmes that categorise potentially disclosable documents. When deployed, it involves: (i) determining the pool of documents for ‘review’ by the software; (ii) human categorisation of a sample from the pool. This phase is intended to ‘educate’ the software; (iii) an initial categorisation, by the software, of all documents in the pool; (iv) manual review and, if necessary, re-categorisation of a sample of the documents analysed by the software. This phase is intended to ‘re-educate’ the programme; (v) application, by the software, of the ‘re-educated’ programme to the document pool; and (vi) repetition of steps (iv) and (v) until re-categorisations fall within a tolerance level.
Recent caselaw provides some guidance on when a court may order the use of TAR, and how it should be used.
Pyrrho Investments Limited & Anr -v- MWB Property Limited & Ors  EWHC 256 (Ch)
In this High Court case which approved TAR, the key features included: (i) an agreement, between the parties, as to its role; (ii) in excess of three million electronic documents; (iii) a cost of several million pounds to manually review the document pool; (iv) a claim value in the tens of millions; (v) upper cost estimates for deploying TAR of c. GBP 470,000; (vi) a window before trial affording “plenty of time to consider other disclosure methods if for any reason the predictive software route turned out to be unsatisfactory”; and (vii) the absence of “factors of any weight pointing in the opposite direction.” The Court also noted that TAR had been sanctioned in other jurisdictions, including the Irish case of Irish Bank Resolution Corporation Limited v Quinn  IEHC 175 where, unlike in the instant case, one party had objected to its involvement.
Triumph Controls UK Limited & Anr -v- Primus International Holding Co. & Ors  EWHC 176 (TCC)
In this more recent High Court case regarding the review of c. 220,000 electronic documents, the Court held that the claimants’ unilateral application of TAR to a sample of the disclosable documents was unsatisfactory. Finding that the work-product of that process should, effectively, be disregarded, it ordered the claimants to undertake a further, sample-review of the document pool. In doing so, the Court made several key points, including:
- the claimants did not provide enough information regarding:
- setup of the TAR or its method of operation;
- the sampling process; and
- the absence of a single, senior lawyer with oversight of the process raised a question as to whether the TAR software had been ‘educated’ on the review criteria as well as it might have been.
These two cases indicate the breadth of factors relevant to: (i) the merits of any position, agreed or otherwise, on the use of TAR; and (ii) whether a party has been sufficiently co-operative and constructive when engaging on the question.
By redefining the regime in terms of known-adverse, basic and extended disclosure, hopes abound for a wholesale cultural change epitomised by greater intervention on the part of the judiciary, and a reduction in the instances of disproportionate (in terms of time and cost) disclosure.
This change coincides with increased judicial emphasis on the cost savings that TAR can produce. In document-heavy cases under the pilot, judges are likely to apply greater scrutiny to the parties’ analysis of, and discussions on: (i) the size of the potentially disclosable electronic-document pool, and how it should be refined; (ii) the costs of TAR; (iii) the mechanics of the TAR process, including suitable tolerance levels for sample reviews; (iv) how the use of TAR would align with the trial timetable; and (v) safeguards, including human review, in the event that the process yields anomalous results. Parties will be doing so against the backdrop of an enhanced duty to engage co-operatively and constructively on issues of disclosure and, ultimately, against the real threat of potentially serious costs consequences for being found by the Court to have taken an unsuitable position.