An article called The Red Herring of Defensibility and Predictive Coding by Craig Carpenter of Recommind on the Inforiskawareness site draws attention again to the technology generally known as “Predictive Coding”. Craig can fight his own corner as to the merits of Recommind’s technology versus, say, Equivio>Relevance – anyone interested in speeding up document review should look at least two implementations of prioritising software.
I say that because whilst such technology is not necessarily easy to describe to those unfamiliar with it, its benefits are instantly obvious when you see it. One might add that anybody not interested in at least considering one or more of these solutions might usefully try and articulate the reasons for ignoring it – I will be happy to host a debate.
From my experience of talking about it (and I moderated a panel in Singapore recently with a star panel from Recommind, Epiq Systems, Nuix and Kroll OnTrack which covered various aspects of automated review) the chief concern is the one which Craig identifies and which is generally labelled “defensibility”. I prefer to think of this as “Am I doing my job properly by using such technology?”
The focus when you look at such things is not so much what the algorithms are doing, still less how they are doing it, but whether you can follow in their footsteps, for example by sampling, to satisfy yourself both as to what has been included and as to what has been left out by the process.
This is by now a mature technology, and one which is available for users far less exalted than those identified in Craig’s article. How big does a case have to be before there is a value in having a means of prioritising them, that is, putting the most important documents at the top of the pile, so that the best-qualified lawyers can look at them first? Not necessarily enormous, I would say.
There is much more to write about this subject (I am in fact doing just that at the moment), but Craig’s article gives you the flavour of the subject.