How AI could have changed the shape of the Enron investigations

25 November 2020 | Orianne Auger, Head of Luminance Discovery

In 2000, Enron was the hottest company in the world, boasting revenues of more than Goldman Sachs, Dell and Microsoft combined. Just one year later, a company once valued at $100 billion was revealed to be a financial con and disintegrated almost overnight. It was the fraud that changed everything, with the subsequent legal investigations taking five years to complete and legal costs running into the millions on both sides. In this blog, I draw on some of the insights provided by my colleague Luke Dailey, Senior Product Specialist at Luminance, at a recent Luminance webinar to look at how Luminance’s Discovery platform could have been used to investigate the Enron dataset. In particular, I take a look at how advanced AI-powered Early Case Assessment, Technology Assisted Review and powerful data production tools would have enabled investigators today to analyse the dataset with unmatched efficiency and insight.

More than two decades since the epic rise and fall of Enron, the Enron dataset remains a “staple” in the eDiscovery sphere for software demonstrations and testing, with Luke noting “it’s pretty much part of the furniture at this point.” However, whilst the dataset might be the same, the technology available to interrogate this information most certainly isn’t.

Get straight to the matter at hand with Luminance’s powerful searching and filtering

Traditionally, eDiscovery technologies relied on complex keyword search strings – in the Enron case, perhaps, investigators were searching for phrases like “electricity bid” or “energy schedule”– followed by an expensive and time-consuming manual review process in which lawyers would have to read through thousands of documents to determine which ones were pertinent to the investigation. However, not only was this approach incredibly time-consuming – and thus costly - but investigators also ran the risk of missing crucial information within the sheer volume of data they had to review.

Fast-forward to 2020, and Luminance has brought AI to the eDiscovery process, ensuring that investigators are able to gain total insight into their datasets. Powered by a unique blend of supervised and unsupervised machine learning, Luminance is able to instantly analyse huge quantities of documentation without the need for intensive machine pre-training, rule-creation or configuration. Moreover, if users are interested in the subject matter of a certain document, Luminance is able to immediately surface conceptually similar documentation to lawyers, allowing them to conduct the most rigorous and insightful review. This is further bolstered by Luminance’s instantaneous de-duplication and filtering functionalities, allowing the reviewer to narrow down the dataset in a few simple clicks in order to find the most important information to their review.

The ‘Deep Coding’ revolution

During the Enron investigation, technology assisted-review (TAR) was not commercially available. Nonetheless, as Luke explains, “there was a boom in eDiscovery technology not long after the Enron investigations due to a culture and fear of ‘what next’”. Earlier this year, Luminance introduced its own AI-powered TAR, known as ‘Deep Coding’. This feature allows the machine to actively learn from the lawyer’s expertise when identifying which documents are relevant to the investigation. Luminance can then plot on a graph how relevant it anticipates the remaining documents to be. As well as predicting results, Luminance highlights discrepancies between coding decisions on documents, including those documents which it believes are relevant despite having been coded as irrelevant, and thus could have been miscoded by the reviewer. This allows lawyers to quickly check these documents at any point, in this way acting as an effective audit tool when multiple users are reviewing a set of documents

Users can then choose to extend Luminance’s understanding of relevancy across the dataset to expedite the review through auto-coding. To ensure that the lawyer remains fully in control of the legal process, users are able to set inclusivity thresholds which can be modified at any stage. Further, Luminance’s relevancy coding is auditable via a simple search.

By forming a conceptual understanding of the documents in this way, Luminance is able to cluster documents based on their similarity on an interactive visual heatmap. Each tile on the map represents a conceptually alike area, and each tile is colour-coded according to review progress. This enables litigators to see at a glance which areas of the dataset have been under-investigated and get a general understanding of the breadth and thoroughness of their review, a tool described by Luke as “incredibly innovative and flexible” and thus would have been “invaluable” during the Enron investigations.

Data production with Luminance Discovery

Another crucial feature of Luminance’s Discovery platform which would have been used with “great effect” during the Enron investigations is Luminance’s robust production capabilities. Indeed, one would imagine that during the Enron investigations, tight, court-established deadlines would mean that the investigation team would be scrambling to produce relevant documents, with daily document productions as multiple levels of review are completed. Previously, this would have been a very manual process - teams would have to spend an enormous amount of time exporting and importing documents, and they were not always able to accommodate last-minute changes or modifications to the production set.

Yet, as Luke notes, “productions are extremely easy to run within Luminance”, with the platform giving users the freedom and flexibility to control how they produce data along with the confidence that they are doing so defensibly. A number of features help expedite the production process. For example, the platform’s synchronisation with Microsoft Word also allows legal professionals to automatically hyperlink all production references in Word to their original document. This ‘live’ Word document can be sent on and opened by colleagues or opposing counsel, who can click on the hyperlink and instantly be taken to the correct document in Luminance. What’s more, by offering highly customisable production options, Luminance can also be used to create, modify and finalise electronic court bundles ahead of trial hearings, ensuring that litigators are more ‘court-ready’ than ever before.

Luminance Discovery: changing the scope of eDiscovery

As eDiscovery’s influence and prevalence has grown over the years, the technologies, economics, processes and expertise around it have in many ways become unrecognisable to what they were in the early 2000s during the Enron investigations. Almost twenty years later, Luminance, the leading AI tool for lawyers, is now allowing litigators to truly expedite their review and comprehend their datasets, transforming the legal industry in the process.

For more information about upcoming Luminance webinars, please click on this link.