Knowledge is power: predictive coding success requires appropriate legal expertise


Posted by on

GCs rightly expect their law firms to deliver top quality results at a competitive price.  One technology that promises to enable firms to deliver on both counts simultaneously is predictive coding.  It is hardly surprising, therefore, that so many column inches have been dedicated to the High Court judgments in Pyrrho Investments Ltd v MWB Property Ltd [2016] EWHC 256 (Ch) and David Brown v BCA Trading.  However, many of these articles laud the court’s endorsement of the technology without addressing the extent to which it is reliant upon expert direction.

Predictive coding is no simple panacea.  Its efficacy is hugely dependent upon close collaboration between legal and technical experts, both of whom need to have a detailed appreciation of each other’s core competencies.  Law firms that treat technology in a “traditional” fashion - as merely an enabler of conventional legal service delivery – will struggle to leverage predictive coding technology.  So too will firms that attempt to outsource this work, literally distancing the legal expertise from the technical implementation and execution.  Effective utilisation of predictive coding will be a critical differentiator for clients instructing disputes lawyers over the next decade.

Why is this hybrid “LegalTech” expertise so important?  A good example lies in appreciating the fundamentals of the technology.  Predictive coding discards basic assumptions and intuitions such as identification and listing of synonyms, acronyms, and translations.  Asking whether a predictive coding engine can handle multiple languages, for example, misses the point; it doesn’t have access to dictionaries, thesauruses, or catalogues of industry terminology.  It doesn’t need to.  These linguistic frameworks and associations are at best irrelevant, and at worst could provide misleading bias. Predictive coding infers relevance by identifying word patterns, learned solely through the coding decisions made by the reviewer.  The reviewing lawyer must understand this methodology on a conceptual level, even if not on a deep “computer science” level.

Another good example of lawyers needing technical aptitude comes from the statistical insight provided during the early stages of a review.  One of the potential advantages of predictive coding is the ability to quickly identify a significant percentage of potentially relevant documents, using the “Active Learning” review methodology.  This can assist the client in making early case strategy and risk management decisions.  However, it will not work unless the lawyer truly understands how much potentially relevant material it has exposed, and how much remains uncovered.  Predictive coding technology expresses confidence in its results by reference to three key metrics: prevalence, recall and precision.  The lawyer needs to understand each of these metrics, and how they interrelate, in order to properly advise the client.  It is also critical in order to challenge demands by an opponent to spend significant sums prolonging a review exercise past a stage of proportionality, where returns are diminishing.

All of this goes far beyond a simple understanding of terminology and methodology.  It also plays a crucial role in strategic decision making.  For example, should the review be coded simply relevant/not relevant, in a party-neutral sense?  Or should the coder apply different codes for “Relevant (favourable)” and “Relevant (adverse)”?  Both approaches have their merits, and either might be more suitable depending upon the context, nature of the dispute, and the qualities of the data set.  It is critical to understand the technology intimately in order to make the optimal choices to keep client costs down and maximise the quality of output.  By the same token, a strong understanding of the legal and factual intricacies of the dispute can assist in making efficient choices when establishing data sets and otherwise prioritising the data for review.

Learning any new skill requires an adjustment to instinct and prior experience.  A snowboarder, for example, must learn to put their weight onto their front foot, overcoming their natural inclination to lean back uphill.  Embracing counterintuitive principles, and learning new skills, is necessary for all lawyers looking to capitalise on the potential of predictive coding.  Some law firms will struggle to adapt, but clients should seek out firms that master the technology, because it will be a significant differentiator, and genuinely improve their prospects of resolving disputes more affordably and achieving more advantageous outcomes.

If you would like to find out how Predictive Coding and other forms of LegalTech can deliver cheaper, more effective litigation outcomes, sign-up here for BLP’s panel discussion with Professor Richard Susskind OBE on 14 September at 8.30am.

Stay informed

Sign up to receive email alerts from our award winning Expert Insights team

Sign up now

See more insights by category

This site uses cookies to help us improve our services and your browsing experience. For further information about cookies, including about how to change your browser settings to no longer accept cookies, please view our privacy policy.