E-discovery and Predictive Coding: FUD Fight!

FOOD FIGHT!!!!!!

The blogosphere has manufactured a lot of controversy this year arguing over predictive coding. Whether it’s transparent enough. Whether it’s defensible. Whether lawyers can use it. Whether we’ll even need lawyers anymore once we start.

Maybe it's not so surprising given that it’s an election year. But in IT we call this FUD (fear, uncertainty, doubt). Because predictive coding is already being used. Vendors have been selling it in products and services for years. Service providers use it. Law firms use it. Legal staffing firms are using it. General counsel use it directly for internal investigations, among other things. Yes, there have been some thorny cases this year in the press, but I’d say these are more the growing pains of mainstream adoption than the cautionary tales of the bleeding edge martyrs.

If my testimony doesn’t convince you that demand – and usage – exists already, consider the evidence that the vendor R&D race/”land grab” phase has begun. Every review tool that didn’t have predictive coding at this year’s LegalTech will damn sure have it by 2013.

This means A) continued, unrelenting buzzworthiness and B) We may start to see more competitive pricing. Cost has been a big barrier to adoption – both in hard costs (licenses, per-GB fees, data plans, services engagements) and soft costs (retraining time, assigning senior attorneys to build the model, figuring out whether the case justifies it, determining whether the data supports it in scale and responsiveness). Once the hard costs drop because every vendor and their brother offers it, the soft costs to adoption will be a lot easier to stomach. We’ve seen this happen before with ECA.

And we already see hard costs dropping – or becoming more accessible, at least. Recommind, an incumbent in the product space, has introduced new flat pricing models for its subscription hosted version with Axcelerate Unlimited. kCura has changed pricing for Relativity Assisted Review (RAR) as well, from per-GB to a data plan model (read more about this in my new ESG brief); its growth in partner sales has some service providers using analytics indexing on every case (Fios) and others offering RAR free for special promotions (Applied Discovery at ILTA). Meanwhile, Relativity’s emerging license business is spurring RAR growth even more quickly as an add-on for direct sales to corporations and law firms. Likewise, Clearwell's entry with transparent predictive coding will also bring the functionality to an already-strong base of enterprise clients - especially if Symantec cross-sells the capability to its archiving customers. [Update: Clearwell's TPC is also no addt'l charge to new or current clients using the review/production module]

The bevy of new options and price drops also means we’re going to have to figure out what’s what. There are real considerations going into choosing a tool, particularly when the market is being flooded with them and pricing gets cutthroat. Is the workflow defensible? What’s powering the underlying categorization engine, and how does it work? What’s the QC methodology; how does it use sampling to validate results?

Two vendors have released Predictive Coding for Dummies in the last few months –but as helpful as these primers are, it’s worth considering a non-vendor perspective. To this end, I’m appearing on a panel at the upcoming IQPC Information Governance and E-discovery Strategy Exchange in Munich, Germany, on November 27 on a panel of experts to discuss predictive coding in-depth. Join us for a lively discussion,“This House Believes that Manual Review Is Not the Holy Grail: The Pros and Cons of Technology Assisted Review."

At the end of the day, predictive coding is one more tool in the litigator’s box. But is it the right one for the job? First you need to know when to choose it and how to use it for optimal results.