Rethinking Relevance: A Call to Modify the Rules of Discovery

Electronic discovery is inherently expensive and burdensome because of the high volume of  information in the world today and because of the broad definition of relevance in our procedural rules. The high volume of ESI is a result of the information explosion, an inevitable byproduct of rapid advances in technology. These advances cause the amount and complexity of our writings to increase exponentially every year. This trend accelerates and stresses our ability to conduct discovery in a proportional manner. There is nothing we can do to put the technology-genie back in the bottle. But we can control the rules.

We can and should revise Rule 26(b)(1), Federal Rules of Civil Procedure, which defines the scope of permissible discovery, and related Rule 401, Federal Rules of Evidence, which defines relevance. These federal rules and their state counterparts permit very broad discovery. They should be revised to limit the scope of relevancy. Only in this way can the cardinal rule of civil procedure be followed, Rule 1, which requires the “just, speedy and inexpensive determination of every action and proceeding.”

The Rules Today Allow for the Discovery of Inadmissible Evidence of Only Marginal Relevance

Due to the unforeseen, rapid advances in technology, Rule 1 now stands in conflict with Rule 26(b)(1). It conflicts because 26(b)(1) allows for the discovery of information not sufficiently relevant or trustworthy to be admissible as evidence, just so long as it appears to be reasonably calculated to lead to the discovery of admissible evidence. That is the grey area shown in the wheel diagram below of discoverable information. This expansive outer circle of e-discovery should be eliminated. Legal actions cannot be determined in a just, speedy and inexpensive manner if discovery into the grey area is permitted. Only directly relevant and otherwise admissible electronic writings should be discoverable, the blue inner circle.

The way it now stands, discoverable “information need not be admissible at the trial if the discovery appears reasonably calculated to lead to the discovery of admissible evidence.” See e.g.Martin Properties, Inc. v. Florida Industries Investment Corp., 2003 WL 1877963, *2 (N.D.Ill. Apr. 14, 2003). Courts permit discovery if there is “any possibility” that the information sought may be relevant to the claim or defense of any party. See, e.g., Sheldon v. Vermonty, 204 F.R.D. 679, 689-90 (D.Kan. 2001).

As the Second Circuit explained in In re Surety Association of America, 388 F. 2d 412, 414 (2nd Cir. 1967)

The only restriction placed upon the matters which may be gone into upon discovery examinations is that they be relevant. Rule 26(b), Fed.R.Civ.Pro. Relevancy, in this context, is tested by a rather liberal standard. “Thus it is relevancy to the subject matter which is the test and subject matter is broader than the precise issues presented by the pleadings.” Kaiser-Frazer Corp. v. Otis & Co., 11 F.R.D. 50, 53 (S.D.N.Y.1951).

When the discovery sought appears relevant, the party resisting the discovery has the burden to establish the lack of relevance. This is done by establishing the requested discovery to be of such marginal relevance that the potential harm occasioned by discovery outweighs the ordinary presumption in favor of broad disclosure. Beach v. City of Olathe, Kansas, 203 F.R.D. 489, 496 (D.Kan. 2001).

The process of requiring the responsive party to seek a protective order, wherein they must prove potential harm and need for protection under Rule 26(b)(2)(C), can be expensive and uncertain. Trial judges and magistrates have broad discretion and protective orders are not lightly provided.

Under the rules as now written, the broad discoverability standards apply to both paper documents and ESI. As the District Court Judge pointed out in Mirbeau of Geneva Lake LLC  v. City of Lake Geneva, 2009 U.S. Dist. LEXIS 101104 (E.D. Wis. 2009):

The [*3] advisory notes to the Federal Rules of Civil Procedure indicate that the federal rules take an “expansive approach toward discovery of ESI and that “discovery of [ESI] stands on equal footing with discovery of paper documents.” Fed. R. Civ. P. 34(a), advisory notes.

The bottom line, as all litigators know, is that it is unduly difficult to defeat discovery requests on the grounds of irrelevance. Parties requesting ESI discovery know this. Some exploit this fact, particularly in disproportionate litigation where one side has many more computers and ESI than the other. They misuse e-discovery as a weapon. Over-broad discovery requests are common under the rules as they now stand. As a result, it is all too easy in discovery to run up a big bill chasing down and producing ESI that has little, if any, real value or impact upon the merits of the case.

An Expansive Concept of Discovery is the Product of a Bygone Era

The expansive concepts of discovery were developed in prior centuries, long before the computer was invented or mankind could even imagine a world of terabytes of information and instantaneous, global writings.  As the Supreme Court noted in Hickman v. Taylor, 329 U.S. 495, 515 (1947):

‘Discovery’… traces back to the equity bill of discovery in English Chancery practice and seems to have had a forerunner in Continental practice. See Ragland, Discovery Before Trial (1932) 13-16.

The revisions to the Federal Rules of Civil Procedure at the end of the last century accelerated and expanded broad discovery. This expansion made sense in the paper world where these discovery concepts were formed. It was not inherently burdensome in the pre-digital age to allow for discovery of inadmissible writings, so long as they were likely to lead to relevant evidence. There were a limited, manageable number of paper records. But in today’s world, the expansive concept of discoverable writings makes little sense. We can no longer afford the discovery of ESI that lacks sufficient probative value and trustworthiness to be admissible as evidence.

We Must Rethink Relevance for Purposes of ESI Discovery

If ESI is not relevant or trustworthy enough to be admissible evidence, it should not be discoverable. Relevant ESI is in itself already voluminous. The additional grey areas of ESI that appears to someone as reasonably calculated to lead to discovery of admissible evidence, is inherently excessive and burdensome. It is a luxury we can no longer afford.

The rules must change for the system to function. Discovery of irrelevant and otherwise inadmissible ESI evidence should not be permitted. It makes speedy, inexpensive determinations impossible in many, if not most cases involving high volumes of ESI.

We need to tighten our concepts of relevancy for purposes of discovery. Modification of the state and federal rules on this key point should make it easier to curb the abuses of  disproportional discovery. Other rule changes may also be necessary, but this one seems to be obvious.

Change Procedure Rule 26(b)(1)

In the federal courts we need to change the provision in Rule 26(b)(1), Federal Rules of Civil Procedure, which reads as follows:

Rule 26. Duty to Disclose; General Provisions Governing Discovery. (b) Discovery Scope and Limits. (1) Scope in General. Unless otherwise limited by court order, the scope of discovery is as follows: Parties may obtain discovery regarding any nonprivileged matter that is relevant to any party’s claim or defense … Relevant information need not be admissible at the trial if the discovery appears reasonably calculated to lead to the discovery of admissible evidence. (italics added)

That last part should be modified by eliminating its application to the discovery of electronic information, at least where large-volume ESI discovery is involved, say anything involving over 100,000 files. This is just one possible way to do it. I am sure others can think of better ways to tie it down so that discovery is not unduly limited. The broad reasonably calculated standard of expansive discovery should remain for purposes of depositions, interrogatories, and requests for admissions, but not for e-discovery.

This call for a change in the rules is directed to both the federal and state courts, as all states have adopted this expansive concept of discovery relevance. As the Supreme Court explained in Seattle Times Co. v. Rhinehart, 467 U.S. 20, 30 (1984):

Most States, including Washington, have adopted discovery provisions modeled on Rules 26 through 37 of the Federal Rules of Civil Procedure. F. James & G. Hazard, Civil Procedure 179 (1977). Rule 26(b)(1) provides that a party “may obtain discovery regarding any matter, not privileged, which is relevant to the subject matter involved in the pending action.” It further provides that discovery is not limited to matters that will be admissible at trial so long as the information sought “appears reasonably calculated to lead to the discovery of admissible evidence.” Wash. Super. Ct. Civ. Rule 26(b)(1); Trust Fund Services v. Aro Glass Co., 89 Wash. 2d 758, 763, 575 P. 2d 716, 719 (1978); cf. 8 C. Wright & A. Miller, Federal Practice and Procedure § 2008 (1970).

Change Evidence Rule 401

One additional, complementary rule change would be to tighten the definition of relevant evidence itself, so that the blue inner circle of the diagram is also reduced, at least for voluminous ESI discovery. Thus, for instance, you could reduce the scope of relevance for purposes of ESI discovery to ESI reasonably likely to have high probative value. This is shown in the inner red circle below.

Relevancy is defined in state and federal rules of evidence, not in rules of procedure. The definitions are all very broad in scope and do not distinguish between types of evidence. For instance, the Federal Rules of Evidence define relevance as follows:

Rule 401. Definition of “Relevant Evidence.” “Relevant evidence” means evidence having any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence.

The Rule’s standard of relevance is said to be a liberal standard allowing consideration of evidence. Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).

The evidence rules could be revised to add a weight factor for purposes of ESI or some other limiting criteria. For instance, the words “any tendency” could be revised for purposes of ESI only, to something like “a significant tendency,” or “a reasonable likelihood.” Again, smarter minds than mine could collectively parse out the specific language that would work best to rein in excessive ESI discovery. The idea is to limit discovery of ESI to information reasonably likely to have strong probative value.

An additional possible change is to limit relevance for purposes of discovery to facts that are disputed. The definition of relevance in the Evidence Code and case-law does not contain any such limitation.

These changes, and perhaps the others suggested here as well, could probably be implemented in procedural rules alone, without revising the related evidence rules. This would require careful study and consideration by the rules committee.

Again, all of these proposed changes to relevance would only apply to ESI evidence, not to testimony and other forms of evidence. It could also be limited to purposes of discovery, with the same broad-scope relevancy definitions used for purposes of trail (the 2% of cases that actually make it that far).

More Discussion on Video

For further discussion on the subject of over-broad discovery relevance, how it is being misused as a mischievous e-discovery weapon, and the pros and cons of a rule change, see the video below. Here Bill Hamilton and I are wearing our Law Professor hats and discussing these issues with students in the Advanced e-Discovery course at the University of Florida Levin College of Law. Turn up the volume to hear Professor Hamilton and turn HD (high-definition) to on.


It has only been a few years since the last federal rule changes went into effect on December 1, 2006. The rules were changed to try to help litigants, attorneys, and judges cope with electronic discovery. The conventional wisdom is to wait a few more years before making any more revisions. That is what I used to think, until just recently, and in the past it was certainly prudent to proceed slowly. But I have come to realize that what made sense in the last century no longer makes sense now. The pace of change has accelerated. We must as a profession keep up with the rapid advances. We can no longer afford to wait. Society is changing fast and so should the law. To wait is to condone further violations of the prime directive of Rule 1.

Changes to Rule 26(b)(1), Federal Rules of Civil Procedure, and Rule 401, Federal Rules of Evidence, are an obvious next step in what must be a continuous evolution of the law. We should no longer tolerate the discovery of information that only has a vague appearance of relevance. We need to redefine relevance for purposes of electronic discovery and we need to do it today, not tomorrow.

I welcome your comments on this proposal.

10 Responses to Rethinking Relevance: A Call to Modify the Rules of Discovery

  1. Bill Potter says:

    Hi Ralph,

    I enjoyed this post, particularly the legal granularity.

    As you may know, The Cowen Group has been commissioned by the eDiscovery Institute to study the costs and burdens of corporate preservation in eDiscovery. The results will be presented by Tom Allman at the 2010 Litigation Review Conference in May.

    If you’re interested, I’d love to talk to you about our initiative and learn more about your perspective.

    All the best,
    Bill Potter

  2. Hi Ralph,
    As a technical professional, as opposed to a legal professional, I am not sure I see the explanations provided as compelling as you do.

    Basically, while much more data is being produced, and thus discoverable than every before – so too is the ability to process that data. What part of the cost curve is the real issue? A million times the volume of data, that can be searched, copied, and distributed a million times faster would seem to just be keeping parity. Is it more an issue related to possible clawback, or protection of the inadmissible evidence from release – in that the defense has a more difficult job determining anything that might be mined from the data that would be useful outside the direct litigation? Have the costs of the defense really escalated more than that of the review of discovered documents?

    For example, pure data mining techniques – ignoring any actual specifics of the data, or even just working with encrypted data, knowing just dates and volume of data – could be used to show that particular executives generated and archived unusually low amounts of data over certain time intervals. Unless there is some reason for this to have occurred, it is good circumstantial information that perhaps they are hiding data, wiped data, or were using different computer systems whose data was not supplied, that could lead to the discovery of meaningful information.

    Note that I am not advocating the release of everything (although escrow might be a different matter), but if the data is not stored, it cannot be retrieved later if it does turn out to be relevant. What part of the costs are the real problem? Would restricting the scope like this address the problem, or just one of the more visible symptoms?


  3. If it takes as long to revise the rules as it did for the Academy to even admit there were e-discovery issues, we can all expect to wait a long, long time for this one.

  4. Dave Degnan says:

    I applaud your efforts to address this important issue. The fix you propose is novel and interesting.

    However, I think it would be better to focus on FRE 403 and give that rule more teeth. Just as FRCP 26(b)(2)(B) & (C) limits 26(b)(1), so too should FRE 403 be modified to limit FRE 401 for purposes of ESI.

  5. Ethan says:

    Ralph: no. Please: no. Treating ESI like its own category of discovery is wrong-headed; we’ve just gotten away from that notion after 20 (30? 40?) years. Much better to directly address the concern at the core of your post: costs.

    In theory, ESI should be cheap. The cost of production media for ten GBs is essentially the same for a hundred GBs — buy 20 more DVDs. Excluding 26(b)(2)(B) sources, computers can write that much data cheaply as well.

    Of course, the high costs are from reviewing outgoing documents. It is unavoidable that a producing party might need to review ESI to determine whether it is responsive to a subjective request (“all documents related to…”). If request is for objective criteria (“all documents with the following keywords”), though, the search for responsive documents can be automated (i.e., cheap). Savvy requesting parties will want to primarily use objective criteria, precisely so they don’t have to trust their opponent’s judgment in determining whether damaging evidence are is responsive. See, e.g., Qualcomm. And if producing parties document that they designed their keyword searches in a good faith effort to locate documents responsive to subjective criteria, they can go a long way towards cheap subjective criteria searches. Cf., e.g., In re Seroquel.

    Keyword searches still need to be reviewed for privilege. But we have the new Rule of Evidence 502 precisely to allow us to find ways of reducing the costs of privilege review.

    We don’t need to change the scope of what is discoverable. Ultimately, if the discovering party requests’ are overinclusive, but the producing partys’ costs are controlled, doesn’t the overinclusiveness work to shift the burden of review back to the discovering party? What we need is judicial development on two fronts: the scope of responding parties’ duties with respect to automated searches for subjective criteria, and new ways to reduce privilege review costs under Rule 502.

  6. Judge Andrew J. Peck says:

    Your most recent blog is interesting as always. But isn’t the problem the search itself? Until companies really archive email in a reasonable fashion, the cost is in reviewing the thousands or millions of emails to see what has to be produced. If a party produced the 100 key smoking gun emails that the opposing party would use as trial exhibits, most counsel would be happy to forego the “outer circle.” But putting aside from hiding the ball, how will the producing party find those key emails? Search the millions of other emails? So revising the discovery rules to limit inadmisible email is not likely to be the solution.

    Also, under your proposal, does an email that says “Watson told me he did X” where X is relevant but the “watson told me” makes it inadmissible hearsay, why should not that have to be produced so that I know to depose Watson?

    While it may be true we need to “do something,” I don’t know if your proposal is the correct “something.”

    Judge Andrew Peck

    • Ralph Losey says:

      Thanks for allowing me to publicly post your comment. Here is my reply.

      You are right search is the key, but search for what? The scope of search now allowed under the rules is by design very broad and that exasperates the problem. We should only have to search for the red inner circle in my diagram above. The barely relevant stuff is a big waste of time to find and review. A common rule of thumb is developing that 2% or less of the “relevant” documents found, using the current rule definitions of relevance, are actually used.

      We have got to figure out a way to both focus the search and improve the search methods and technologies. We need to try multiple approaches to make discovery more affordable, and I certainly would not rely on rule changes alone. But I no longer want to rule-out rule changes, and the very broad definition of relevance used in discovery seems to me like a good place to start.


  7. David Stanton says:

    This is a thoughtful article, but is it missing a piece of the puzzle?

    Since the 2000 amendments, Rule 26(b)(1) limits the scope of discovery to matters “relevant to any party’s claim or defense,” and requires parties to get a court order based on a showing of good cause before probing into more peripheral matters merely “relevant to the subject matter involved in the action.” Thus, while discovery can still be had of inadmissible information, the scope of relevance is already constrained by the pleadings and the probative contents of the sought-after materials.

    It seems to me that the article sets up something of a straw man by citing In re Surety Association and Seattle Times to explicate the breath of discovery under the FRCP. These cases were decided before codification of the distinction between “claims and defenses” relevance and “subject matter” relevance.

    When this dichotomy under 26(b)(1) is coupled with the proportionality limitations of 26(b)(2)(C), don’t we already have three concentric rings for discovery: (1) relevant to claims/defenses and cost-justified; (2) relevant to subject matter and cost-justified; and (3) relevant by not cost justified?

    I don’t see what an attempt to define “relevance” according to “admissibility” would add to this.

    Moreover, “admissibility” is something that largely depends on non-textual qualities of documents (e.g. their status as routine business records, their usefulness for impeachment). Designing an automated search based such extraneous criteria rather than the actual contents strikes me as a monumental task. Seems that assessments of admissibility are so much more complex than relevance that they would require more human involvement and legal acumen, substantially increasing the costs of the e-discovery process.

  8. Jeff Reed says:

    Thanks once again, Ralph, for making me think outside the current box and try to figure out if we can and should be doing something different and better.

    Several solutions come to mind:

    The problem for business usually isn’t in the harvesting or hosting of the data – it’s in the legal hold and the review. So let’s work on releasing the legal hold within a relatively short time of harvesting – but only if we also come up with a way of shortening the review stage so that the receiving party has a chance to examine the production and raise any red flags. How do you shorten the time to production? Bar any use of any item that ultimately proves to be privileged. Embellish that rule with additional penalties (like disqualification of the lawyers and firm) if the receiving counsel fails to ask the producing party if an interesting document that looks/acts/smells like a privileged doc isn’t in fact a privileged doc. Right now a number of states and the FRCivP require a lawyer who receives a privileged doc and knows or should know that it is privileged to call up opposing counsel and report it. Why not give that Rule some teeth? Why not make the privilege real and actually protect this stuff rather than play the gotcha game with it?

    Then review isn’t necessary – or at least can be done more quickly through targeted searches to excise the bulk of it relatively quickly – and we can get on with the case.

    That leaves us with Judge Peck’s and your observation that “search is the thing.” Indeed it is – but in two ways not one. First – find the material that fits today’s loose definition of discoverable material. Second – artfully search through the mass of material produced in order to find what you really want. It’s the second search that lawyers need to master in order to master the data and their cases. Much like doing legal research, searching through email, word processing documents and spreadsheets is kind of an art form. If lawyers can master headnotes and Shephards, why can’t they also manage to master searching document databases?

    All the best.

  9. Hi Jeff:

    Read the whole post and the reply, so pardon me if I somehow misunderstood you wrong.

    first off, the ideas you set forth are sound and worth looking into…

    “How do you shorten the time to production? Bar any use of any item that ultimately proves to be privileged. Embellish that rule with additional penalties (like disqualification of the lawyers and firm) if the receiving counsel fails to ask the producing party if an interesting document that looks/acts/smells like a privileged doc isn’t in fact a privileged doc. Right now a number of states and the FRCivP require a lawyer who receives a privileged doc and knows or should know that it is privileged to call up opposing counsel and report it. Why not give that Rule some teeth? Why not make the privilege real and actually protect this stuff rather than play the gotcha game with it?”

    This is basically easier said and done. From what I know most states do not even define “privileged” docs and have no set time for disclosure. Even so, if it is really a “priviledged” doc, often times, you wouldn’t have to admit to having it, unless you have to brandish it in court.

%d bloggers like this: