By | 27th June 2013

Vilified in death to such an extent that his bones have been moved on a number of occasions to prevent attempts at desecration, Hernando (Fernando) Cortes was one of the most successful conquistadors of the 16th century.

The man who conquered the Aztecs in Mexico winning vast tracts of that country for the Spanish empire, once said to the Spanish Emperor who had demanded to know who he was that ” I am a man who has given you more provinces than your ancestors left you cities.”

What is less well known is that he introduced vanilla to the Old World where it had previously been unknown.

Produced from a Mexican species of orchid, the word vanilla comes from the Spanish word vaina meaning little pod.

Nothing particularly special there, you may think, and of course there isn’t but its ordinariness has given rise to the secondary definition of vanilla to mean without frills.

This links to a recent report by Rational Retention on the subject of predictive coding and a section on vanilla computer assisted review.

I came across this report because I am a member of a LinkedIn Group concerning Computer Assisted Review where I saw recently an announcement about a paper prepared by experts at Rational Retention which seeks to set out how machune learning can be applied to a variety of other activities including attorney review.

The announcement reads in full:

“Machine learning-based text categorization tools and techniques have been recently introduced to the Legal industry under the guise of Predictive Coding or Technology Assisted Review (TAR). Although currently aimed at the automation of low value attorney review during eDiscovery, the same machine learning technologies may be applied to solve a number of other real world challenges across a variety of legal domains. Because machine learning for the legal industry is in its nascent stages, most legal and business professionals have a limited understanding of its capabilities and limitations and, more importantly, misconceptions are widespread.

To address these misconception, the machine learning experts at Rational Retention have authored a paper to explain machine learning, text classification, and predictive coding and outline how it may be applied to solve various real word challenges, including automating attorney review.

This paper is not “Predictive Coding for “Dummies.” Replicating human decision-making is a complicated endeavor and as such requires sophisticated solutions. This paper shows that with the right technologies and the right data experts, it is possible today for organizations to effectively cope with the growing volume of data and the associated costs, risks, and opportunities.

Please contact Charles Skamser at to receive a copy of this paper.”

Anyone interested in TAR should be interested in this report and should email Charles for a copy as specified above, not least because the report is anything but vanilla!

Including notes, the report runs to 79 pages and I want to draw attention to just one of the many issues discussed by the authors. Readers of this blog and a myriad of other sources will be familiar with the vanilla uses of predictive coding.

Beyond the vanilla, is section 3 of the report which deals with the other strategic uses to which predictive coding may be put, where examples include the fields of litigation readiness and information governance.

The authors summarise this section as follows:

  • Predictive coding tools can review documents produced by third parties, even if those parties do not use predictive coding to review their documents.
  • Predictive Coding experts can also serve to highlight the inadequacy of opposing party’s technology and results.
  • A huge opportunity lies in classifying and controlling documents in the enterprise, long before litigation strikes. Enterprise content can, for the first time, be reliably examined, classified, and defensibly pared down.
  • It is foreseeable that predictive coding providers will be able to function as neutral party (or magistrate of document production), so that neither side can game the system.
  • Government may be one of the biggest beneficiaries of predictive coding technology.

Food for thought!