What can predictive coding do for you?

Last month I wrote about the techno-logical aspects of predictive coding. This month I want to expand that into its potential effects on the profession. Now that predictive coding has been ratified for use in the discovery process, its potential is fairly huge in the world of litigation. It could also lend itself to non-litigious areas of work, such as reviews of large cases, judicial review, large scale enquiries, and more besides. When you consider the expansion of data generally, it is likely to become a reasonably commonplace exercise to use this sort of technique to extract and distil information. And while most of this is likely to come with the scope of litigious work, it could easily be adapted to a broader audience that is any process involving huge electronic document stores.

Cost reduction

There is no doubt in the minds of most industry experts that predictive coding will take off and become a standard feature of casework for many lawyers. The ability to sift through vast volumes of documentation and obtain accurate and relevant results is likely to revolutionise life for both client and lawyer alike. As with all technologies though, it will more than likely have to develop to cover more than its current remit. The ability to recognise non-text file types such as graphics and sound files will become more important over time, and within that, the ability to analyse the content of various types of files will make this kind of system more useful still.

In my piece on Big Data (Solicitors Journal 156/37, 2 October 2012), I wrote briefly about the rate at which information is being stored, and how that is increasing exponentially over time. It’s not just the quantum of data but also the types, and with it the relative size, that is changing. More and more of us rely on increasing data storage levels for more and more things. And the advent of Facebook, Google, and many other internet enterprises means a mix of information, and the generation of more types of information such as nested metadata and derived data.

It’s likely that at some point in time the latter type of information will become as useful to legal processes as the original data itself, and thus there is scope for Big Data techniques and predictive coding to merge together to provide a multifaceted querying structure, one that can blend results from a multitude of types of system.

The impact on lawyers, and clients of predictive coding should not be underestimated. Taking a process such as discovery, and producing useful results in a fraction of the time it has traditionally taken should mean better and more accurate case handling. Attacking vast data stores that would previously have remained untouched should widen the scope for actions being processed. It is not difficult to conclude that more cases will end up going to trial due to the expanded ability to produce discovery material. The question that some may proffer is whether this facility will skew cases in favour of the more well-heeled client.

While there are as yet no accurate measures of the difference in cost of predictive coding versus traditional manual discovery, it seems clear that it will be a fraction of the cost, due to the potential reduction in required, skilled, man hours. This in turn should allow clients to instruct on matters where either the cost would previously have been prohibitive, or not in proportion to the case value.

Effect on fees

Much of the commentary on predictive coding suggests that the result of its widespread adoption will be that lawyers will be used less and less in the process, which may have an effect on their ability to charge clients, as well as a change in the charging method. I’m not convinced about this. To my mind it may well have the opposite effect in the short term, where more cases are taken further down the legal process as a result of the access to additional information. The longer term is anyone’s guess, but in my time I have not seen technology particularly erode the lawyers margin.

One thing to bear in mind with predictive coding is that it requires a legal expert to train the system to find results in a given data set. The same expertise must then be applied to the results to ensure that they are accurate enough. This cannot be done without legal expertise, and I would venture that this type of work will simply replace the traditional discovery process for lawyers. Of course the end result should be more information and increased accuracy, but not necessarily at the expense of human input. Providers of predictive coding services are at pains to point out that it does not replace the human element of document review.

The use of predictive coding is still very much in its infancy in the legal world. The charging mechanism for the use or purchase of the software is likely to go through a settling period as more suppliers enter the market during the next couple of years. This should mean a lowering as well as a levelling of prices. One thing that is for sure is that it cannot be ignored by the legal profession, and as with many new technologies, early adoption could garner some competitive advantages, at least in the short term.