Computer-assisted review: technology says yes

12 Feb 2014Feature

The technology and the evidence are there, now all litigators are waiting for the first test case, say Chris Jackson is a partner and Lloyd Nail

Most commercial disputes solicitors are now used to the idea of using an electronic disclosure platform to categorise and review documents in at least some of their caseload. Parties and judges expect to discuss data capture, search terms and electronic document questionnaires at CMCs and solicitors are becoming ever more knowledgeable about the electronic disclosure process.

New and improved software and platforms for electronic document review are becoming cheaper, faster and more powerful. Smaller cases are now capable of being handled economically through on-screen review. However, a real revolution in data management and review is happening at the other end of the spectrum.

Computer-assisted review (CAR), which has been promised for at least a decade, is already established in the US and is on the verge of the mainstream in the UK. It is championed by certain judges and, as electronic data exponentially increases, it will become a necessary tool for the effective disclosure in the post-Jackson age - particularly where 'standard disclosure' is no longer automatic).

Nonetheless, even the big law firms are waiting on the sideline shuffling their feet and watching for others to use it first. It is not a matter of doubt - senior lawyers in most major firms sing its praises. Everyone is waiting for the right test-case.

Caution rules

At its most basic, assisted review is a process by which the computer is 'trained' to recognise relevant documents from a wider dataset by 'learning' from a succession of manually reviewed batches of documents. It is an iterative process - as further batches of documents are reviewed the review tool hones its ability to categorise documents. Eventually, a threshold of stability is reached and the review tool can apply what it has learned to the rest of the dataset, thus allowing the review team to focus their efforts on the documents categorised as relevant. This reduces the distraction and wasted cost, time and effort of reviewing irrelevant documents so huge cost savings are, in theory, possible.

Perhaps more surprisingly, the statistics show that assisted review is more accurate than manual review. With this in mind it seems inevitable that once everyone understands computer-assisted review and trusts it, it will be used. But that is why caution currently reigns. Lawyers are inherently reluctant to believe that there is nothing relevant in a huge pile of documentation which no-one has looked at, despite what the evidence shows. Until there is a proven track record, the parties will doubt whether the court will trust a computer to say that the bulk of the data recovered should not be reviewed. One disclosure provider cites a case where the court got cold feet and asked the claimant to review a further 20 per cent of documents which the computer had identified as not relevant - an extreme example of how a court might require a further validation process.

Even if the court is minded to allow it, the other side may object unless they are equally familiar with the approach. One of the inherent difficulties in employing a review tool which depends on statistics and algorithms is translating the maths of probability into court- and client-friendly language. Solicitors and the courts will have to become comfortable with the language of assisted review - the terminology of confidence levels, overturn rates, precision, recall and acceptable margins of error (and how these interrelate) - before they will recommend or accept its use.

Gen Y factor

There is also a generational factor in play. Generation Y have grown up with technological solutions and generally have no issues with seeing them deployed. However trial preparation approaches are traditionally set by Generation X who, while they have adapted to technology, tend to adopt it more incrementally.

CAR also threatens business models based on large outsourced review facilities deploying large teams of junior lawyers. Some of those review facilities have only recently been set up as being 'the future'.

Ultimately, your ability to use assisted review (even if appropriate cases) may be dependent on your ability to convince the other side or the court that it is an appropriate tool in your case i.e. that it can reduce the burden and cost of electronic disclosure without compromising the standard of disclosure provided.

Nonetheless, the Jackson Reforms and the Practice Direction on Electronic Disclosure (PD31B) have introduced strong drivers for considering the use of assisted review and doing so at a much earlier stage. The requirement to produce a disclosure report, the encouragement of the use of EDQs in appropriate cases, and the need to accurately budget for disclosure as a result of the strictures of Precedent H, work together to create a strong case for formulating a disclosure proposal at least several weeks in advance of the CMC, if not before. In appropriate cases, consideration of the use of assisted review should form part of that process.

The stark reality is that 90 per cent of the data in the world came into existence in the last 24 months. Technology has created an issue that traditional electronic case review cannot begin to deal with. In parallel the courts, after Jackson, require greater efficiency and streamlined approaches. Those factors dictate that the future of electronic disclosure is obvious. Computer-assisted review is not only important but in certain cases is currently the only effective solution. Now it needs an appropriate test case.

Chris Jackson (pictured) is a partner and Lloyd Nail a senior associate at Burges Salmon