Phased Discovery: Undue Burden and Practical Proportionality

In a racial profiling case, the parties were in gridlock over custodians and date ranges to search of the electronically stored information.  Barrera v. Boughton, 2010 U.S. Dist. LEXIS 103491 (D. Conn. 2010).

After a meet and confer, the Plaintiffs proposed searching 40 custodians, with 80 key words from January 2004 to present.  Barrera at *8.

The Defendants objected, citing they had already produced over 900 email messages (none of which supported the Plaintiffs claims).  Barrera at *9.  Moreover, the Defendants’ estimated cost from their IT experts for conducting the Plaintiffs’ search would be over $60,000.  Barrera at *9. 

The Defendants proposed an initial search of three specific custodians from 2005 to 2007.  The Defendants estimated the cost to search at $13,000 to $15,000, which they argued the Plaintiff should pay.  Barrera at *9. 

The Plaintiffs wanted, at a minimum, another three custodians initially searched and maintained their position that 40 custodians needed to be searched.  Barrera at *9-10. 

The Court found that the electronically stored information from the 40 custodians was not reasonably accessible because of cost.  Barrera at *11-12, citing Fed. R. Civ. P. 26(b)(2)(B).  The Court also found that the Plaintiff had not demonstrated good cause in justifying the production of the “not reasonably accessible” ESI.  Barrera at *12. 

The Court ordered the parties to conduct a phased discovery approach, where the three custodians would be searched with the Plaintiff’s search terms from 2004 to 2007.  Barrera at *12.  The Court also denied any cost-shifting.  Barrera at *12. 

Bow Tie Thoughts

Proportionality is entering court opinions.  The Court’s order for phased discovery highlights a practical approach in searches of multiple custodians. 

However, the opinion is silent on whether the Defendants’ IT experts were in-house IT or a service provider. The opinion is also silent on how the data was being searched.  What technology were they using to conduct the searches?  How was the data stored?  While collecting data over six years might involve back-up tapes, searching three custodians with 80 keywords might be less than $15,000.

A colleague estimated the following workflow and cost:

The average person would have 5GB of email/data a year.  Provided there are three custodians and three years worth of data, there would be a total of 45GB.  The total time to cull down, index and run key words in DT Search would likely be twelve hours.  Assuming that the service provider was charging $200 an hour for the work (some service providers might be lower), the cost would be around $2,400.

Parties address e-Discovery challenges with the tools and resources they have on hand.  While the in-house IT expert might have the skills of a certified BMW mechanic on a company’s network, you would not ask that mechanic to tune-up a 747.  This is where the outside service provider is important, because they can use their tools, provide quality services and drive down costs.

  1. It’s good that courts are finally starting to enforce proportionality.

    I think, however, that your colleague’s estimate of cost omits some basic and very large components of the cost equation. Specifically, the workflow omits the pre-production review by Defendant’s Counsel. No matter how good your collection processes are, at some point Counsel must put eyes on the documents and review them for relevance, privilege, etc. (Yes, you can delay that expense with non-disclosure and other agreements with opposing Counsel but that’s only cost delay, not cost avoidance.)

    Unless the keywords are extremely tightly targeted, the search will return a high proportion of false positives – emails which include the keyword but in some completely innocent or irrelevant context. With a list of 80 keywords, I consider it unlikely to the point of impossibility that ever one of those keywords would be tightly targeted and uniquely used.

    That means that regardless of your collection time or tool efficiency, you will return many Gb of data. Advanced filtering tools could reduce the results set further but even the best tools will likely return tens of thousands of emails for a set of 80 keywords.

    We should all use technology as much as we can to reduce the volume but eventually you have to get to manual review. At 50 docs per hour times a lawyer’s hourly rate (plussed up for quality control and overhead), you are quickly going to dwarf the estimate of $2,400 of search and collection costs.

    Ideally, the Defendants should have some sample database that they can use to quickly test the effectiveness of proposed keywords. Failing that, I agree that a phased approach is best. It at least gives you some ability to test the keywords before exploding the cost across a wider target set.