Skip to main content

A Q&A with Chris Dale Regarding the UK Pyrrho Predictive Coding Judgment


  1. Why has predictive coding or technology assisted review (TAR) been accepted just now in the courts of England and Wales?

    I don’t think there is a particular reason why the beginning of 2016 should bring us this judgment. The Civil Procedure Rules have long required that the parties and the court consider the most proportionate way of managing disclosure, and proportionality was an express component of the rules even before it was inserted into the overriding objective in 2010.

    CPR Rule 31.5, introduced as part of the Jackson Reforms in 2010, requires parties to consider co-operatively between themselves, and then with the court, the scope, the proposed methods, and the cost of dealing with disclosure. In Pyrrho, the parties sensibly agreed what should be done, including the technology to be used, and presented comparative costs schedules is to help the court make a decision.

    The Goodale judgment, referred to extensively in Pyrrho, may not have been a specific validation of a particular technology, but it was a clear statement by Senior Master Whitaker that modern analytic technology should be considered when dealing with large volumes of data. It is also probable that there have been other court authorisations of technology like this in judgments which have not been published.

  2. Master Matthew’s description of predictive coding workflow is only one type, with many other options available to parties considering this use of technology. Regardless of the type of predictive coding software used, what are common process components that parties should seek to adopt?

    It may sound obvious, but the first step is to decide whether the use of TAR is appropriate for the case. As Master Matthews made clear at the end of Pyrrho, his judgment was made on a particular set of circumstances.

    The attributes of one particular type of TAR were baked into the judgment and this is perhaps a pity. There are many a number of different solutions falling under the TAR umbrella, each with their own attributes and terminology, and it should be part of a litigation lawyer’s skill set to know who the main players are and what differentiates one product from another.

    There are other decisions to make as well – are the lawyers intending to host and manage the process themselves or will they do better to engage consultants to do part of the process for them in a hosted environment? Conduent, for example, offers OmniX Assisted Review module (fairly self-service but with guidance from Conduent experts), Viewpoint Assisted Review (an integrated module within Viewpoint), CategoriX (a fully outsourced solution with review integrated into the OmniX interface, and Relativity

    There are a range of factors, including the lawyer’s own skills and the competing pressures on their resources which may point one way or the other.

    Regardless of the system used, the process should include at least:

    • Setting up a team with appropriate skills, whether in-house or external.
    • Having a system for measuring the quality of the results and ensuring statistical validity.
    • Maximising the capability of the chosen software to adapt to and learn from results.
    • Deciding what the technology is to be used for – simply for culling, for prioritisation or, in tandem with other tools, to make judgments about relevance.
    • Devising a system which is repeatable and audited so that decisions can be justified and, if necessary, varied as the understanding of the case develops and as different players, including opponents, have input into the result.

    Lastly, there must be cooperation with the other side. Whatever the arguments about this in the U.S., it is a requirement under the English rules that parties agree on the tools and on the method to be used and submit their agreement – and the areas of disagreement – to the court. This must be set up from the outset and be a continuing process right through to the giving of disclosure.

  3. What can UK legal teams learn from the U.S. now that predictive coding has been accepted?
    It is commonly asserted in the U.S. that U.S. discovery is in some sense “ahead” of that in the UK. It is certainly true that the pressures of the Federal Rules of Civil Procedure have driven the development of first-rate eDiscovery technology, the skills to use it, and relevant workflows and processes, which can be imported into other jurisdictions.

    It is also fair to say that Opinions from the likes of U.S. Magistrate Judge Andrew Peck give a good model for other jurisdictions to follow – as indeed the judge in Pyrrho did.

    UK lawyers have also been relatively slow to appreciate the value of distributing the functions of eDiscovery among those who can best perform them. U.S. lawyers seem more willing to accept the idea that expert advice from third-party consultants can produce better results while saving money in the long term; the lawyers in Pyrrho did just that.

    The U.S. leads also in the development of managed review services and other third-party ways of achieving the expensive review process at a lower cost.

    Beyond that, however, the U.S. example is not merely unhelpful but positively damaging to those of us who seek to encourage UK lawyers to manage eDiscovery proportionately. The fear (however illusory) of sanctions, the battles over Requests, privilege logs and other requirements seem to us to add expense without value, even before taking account of the combative posturing which seems to be expected of litigation lawyers.

  4. What can U.S. legal teams learn from this judgment?
    The U.S. has just been through a long and contentious process to change its eDiscovery rules. I hesitate to shout about it too much, but some of the changes, consciously or not, reflect those already made in the UK, including the removal of the right to demand what we in the UK call “train of enquiry” discovery, that is, documents brought in only because they may lead to other evidence; we got rid of that in 1999.

    Things like that, coupled with the much narrower definition of a discoverable document (we replaced “relevance” with a much narrower test in 1999) offer real opportunities to reduce the eDiscovery burden. A further difference, not one limited to discovery, is the cost-shifting default under which the loser pays the costs of the winner.

    None of this is directly derived from the Pyrrho judgment. What does appear which is helpful in other jurisdictions is the value of enforced cooperation based on openness and the duty of active management imposed on judges.

  5. Any other lessons here which relate to the increasingly important subject of cross-border discovery?

    Giving effect to U.S. discovery orders in other jurisdictions is always expensive. Even before the recent Schrems judgment, those seeking U.S. discovery were compelled to abide by EU laws relating to data protection and privacy. These require that personally identifiable information be flagged in-country and that the U.S. lawyers are enabled to go back to the U.S. court armed with information about sources, scope and expense together with a clear idea of what data will cause problems and what can be done to resolve them.

    The technology used in the Pyrrho case is ideal for this when coupled with processes and disciplines similar to those described in the judgment. It is also true to say that cross-border discovery disputes can be minimised by the kind of informed discussion and cooperation which was shown by the parties in Pyrrho.

You can view more of Chris Dale’s insights on Pyrrho at eDisclosure Information Project.

Rachel Teisch is Vice President, Marketing at Conduent.  She can be reached at