So, you’re deep in the weeds of your latest e-discovery project for an important client and things are going smoothly. Protective orders are in place, the scope of discovery requests agreed to, custodians properly identified, search terms satisfactorily negotiated, the review team exhaustively trained and plugging away, QC processes set up and running successfully and timely production of your documents seems viable. You’re even under budget and ahead of schedule! (If all that seems farfetched, don’t worry, just accept it as a hypothetical.) An email now arrives from the review team flagged as urgent, and with the tag line, “Take a look at this document, it’s got some pretty sensitive info – that isn’t relevant – and I’m pretty sure the client won’t be happy sending this out.”

You pull up the document and, “Yikes – that sure is sensitive!” Yes, the document has some marginal reference in a line or two that makes it responsive to the discovery requests, but it also has page after page of sensitive information on subjects completely unrelated to the litigation in question. For our purposes it doesn’t really matter what the information is – trade secrets, strategy documents that could affect the client’s stock price, internal investigations that could open the client up to unrelated liability, information that could lead to public scorn and PR nightmares… the point is, even with a protective order in place, here is something you just don’t want to let out of the building. Even worse, a search through the universe of documents identified as responsive reveals dozens, perhaps even hundreds, of documents with similar information. Fortunately, the information is indisputably not relevant. No worries, right? Time to tee up the redactions team! (Isn’t it lucky we’re under budget and ahead of schedule?) But hold on a sec - before you start drawing black boxes all over your documents, did you know a trial court in Wisconsin recently said you can’t do that? What will your court say?

In IDC Financial Publishing, Inc. v. BondDesk Group, LLC, No. 15-cv-1085-pp, 2017 WL 4863202 (E.D. Wis. Oct. 26, 2017) [you can find the case here], the Defendants produced approximately 6,000 documents and redacted approximately 600 of those documents on the basis that they contained non-relevant information on contractual and business relationships that were not at issue. The Court ordered the Defendants to produce unredacted versions of the challenged documents and held that the protective orders in place in the litigation were sufficient protection for the defendants with regard to any irrelevant information. Id. The particular reasoning of the Court may not seem particularly troubling to you, given its lack of precedential weight, but it does demonstrate how unpredictable the trial level outcome of a fight over redactions may be. And remember, trial courts have broad discretion over discovery matters. Roberts v. Shawnee Mission Ford, Inc., 352 F.3d 358, 360 (8th Cir. 2003) (citations omitted).

A key to the Court’s reasoning appears to be that the basic discoverable unit in the discovery process is a document: “[the Defendants] have blocked out large chunks of information on documents that, by virtue of producing them, they admit are discoverable.” IDC. Although it is common practice in e-discovery to discuss production in terms of documents, it is not clear that Federal Rule of Civil Procedure 26(b)(1) contemplates or mandates production by (whole) document. Remember that Rule 26(b)(1) and its various state analogs say that “Parties may obtain discovery regarding any nonprivileged matter that is relevant to any party's claim or defense…” (Fed. R. Civ. P. 26(b)(1)) (emphasis added). Relevant matter, not relevant documents. There is no inherent reason that non-relevant matter should be discoverable under 26(b)(1) simply because of proximity to relevant matter, for instance because it appears in the same document.[1] Note also the Court’s fascinating intellectual legerdemain in the above statement “by virtue of producing [the documents], [the Defendants] admit they are discoverable.” IDC. In actuality, the Defendants, by redacting material for non-relevance are clearly making the explicit claim that the redacted material is not discoverable. The Court flips that on its head and turns it into an admission from the defendants that the material they have redacted is actually discoverable. Id. Certainly, the Court likely has discretion to compel production of whole documents under its broad oversight of discovery, but it’s quite a stretch to argue that the Defendants have conceded that whole documents are inherently discoverable by the act of producing them with redactions.

We probably don’t want to get too hung up on a discussion of documents as the essential discoverable unit[2], but I find that argument so interesting that I just have to spend a bit more time with it. Have you ever seen spreadsheets with a hundred thousand or more lines in your discovery practice? (Yeah, me too!). So let me travel way back to college philosophy and engage in a little reductio ad absurdium here. Suppose you have a 100,000 line spreadsheet with 12 columns. It’s probably fair to describe that spreadsheet as having 1.2 million pieces of information. Suppose further that one line of the spreadsheet is relevant in your dispute. Are we really going to compel the production of nearly 1.2 million pieces of irrelevant information simply because they are in the same spreadsheet as 12 pieces of relevant information? If you think that’s absurd (reductio ad absurdium, remember) and that no court would compel you to turn over such spreadsheets without being able to redact or otherwise account for all that non-relevant information, it turns out that this is exactly what the IDC Court is compelling, albeit possibly not to such an extreme extent. The Court notes that “the defendants have redacted nearly all the text in ‘dozens of emails, contracts, and spreadsheets’” (internal citation omitted) and that “[the Defendants] have blocked out large chunks of information on documents, that, by virtue of producing them, they admit are discoverable.” IDC. Presumably, the spreadsheets in question are of some more typical size and not 100,000 line behemoths, but it’s not clear that the principle would vary based on spreadsheet or document size.

Other points made by the Court include: 1) the Plaintiffs are expected to take the Defendants’ word that the redacted material is not relevant; 2) context necessary for understanding non-redacted material may be hidden; and 3) parties may be incentivized to push the line on redactions and hide as “much as they dare.” Id. Note this third point is basically the first point with the hint of impropriety thrown in. Naturally, no impropriety on the part of Defendants is alleged in IDC.

Let’s take those three points in order:

1) So the Plaintiffs have to take the Defendants’ word on relevance? Aren’t they already doing that regarding all the documents that did not get produced? After all, the Defendants turned over 6,000 documents. How many did they review to get to that number? 20,000? 40,000? 100,000? How many documents did they not even bother to review because the files did not hit on search terms or were from uninvolved custodians? Millions? The whole edifice of discovery is based on the premise that one party takes the other party’s word that the documents they do not turn over are not relevant or otherwise non-discoverable. Why should this premise be challenged in the case of redactions, particularly in the absence of any allegations of bad faith? In IDC, the Defendants actually proffered descriptions of the challenged redactions to ameliorate that very concern.

2) The Court says that context necessary for understanding non-redacted material may be lost due to redactions. Arguably, if the context were indeed necessary to understand the unredacted portions of a document, the redacted material is actually not irrelevant and should be produced. In any case, it seems likely that in most instances properly redacted material would not be essential to understanding the produced material, and lack of necessary context would be an exception, rather than the rule. A narrowly tailored remedy would seem to be more appropriate here, to the extent there is any issue (though there does not seem to be any allegation that necessary context has actually been redacted out in IDC).

3) Parties might try to push the envelope on redactions? Again, absent an allegation of bad faith, I’m not sure how concerned should we be with a party trying to hide relevant material through redactions when we aren’t concerned with them trying to hide the ball with unproduced documents.

Finally, the Court indicates that the protective order in place in IDC is all the protection the Defendants need for the non-relevant material in their produced documents, no matter how sensitive. IDC. Indeed, a protective order covering produced confidential material is pretty much mandatory in any e-discovery proceeding. According to the Sedona Principles, “To address appropriate protection for trade secrets and other confidential commercial information in the United States, the parties and counsel should discuss … a suitable protective order..” (The Sedona Principles, Third Edition, 19 Sedona Conf. J.1, Cmt. 10.j. 163) (forthcoming 2018). (The Sedona Principles is the treatise put out by the Sedona Conference Working Group on Electronic Document Retention and Production which for over fifteen years has been developing and codifying best practices in e-discovery and can be downloaded here). However, the purpose of such a protective order is to reduce the burden of production and to protect the producing party’s information. Redacting out all non-relevant material, or even just all confidential and sensitive non-relevant material[3] is typically expensive, time consuming, and impractical and thus entirely suitable to be covered by a protective order rather than through redactions. However, should this shield for the protection of producing parties then be used as cover to force the production of material so sensitive that the party was willing to take the burden, expense, and time of redacting? Remember, what we’re discussing here is fundamentally non-relevant material. No matter how robust the protective order, once information is out of your hands, you have no control over what happens to it and it may, through inadvertent error, carelessness, the action of some other court, or otherwise, get handled inconsistently with the protective order. As noted by the Sedona Principles with respect to a Rule 502(d) protective order (dealing with non-waiver of privilege) “Rule 502 cannot erase inadvertently disclosed information from the mind of the requesting party” Id. at 152.

What is the takeaway for your discovery practice? The fact that the IDC Court found that Defendants could not redact confidential non-relevant material from documents they were producing is concerning, even though we can find trial courts that come out on the other side of this issue. So, what to do? What is the best practice here?

One approach would simply be to go ahead and redact particularly sensitive non-relevant information. After all, the material is not relevant and the other side may never make an issue out of it. The better practice, however, may be to address the issue head on at the earliest stages of litigation – an approach supported by the Sedona Principles. The Sedona Principles operate on the basic assumption of openness and cooperation in discovery issues: “parties should address discovery issues cooperatively” (Sedona Principles, 77), “counsel should always attempt to resolve potential disputes where feasible” (Id.), and “if both requesting and responding parties cooperate to evaluate the appropriate procedures, methodologies and technologies to be employed in a case, both may potentially achieve significant monetary savings and non-monetary efficiencies.” Id. at 78.

If you anticipate you may need to redact non-relevant confidential information, be prepared to discuss it at any 26(f)[4] conferences you may have or even in informal discovery conversations in earlier stages of the dispute. Ideally, you can reach an agreement or get an early court ruling now, before the litigation gets contentious, or any more contentious. If the issue is raised at this point, counsel that is dubious about agreeing to redactions may, for instance, be more open now than later to compromises regarding redactions. If redactions are going to be an issue of contention, it will be helpful to identify that early, as it may inform how you conduct the discovery review or what you try to include in a protective order.

I do have one discovery practice tip before we depart. In your responses to the opposing party’s requests for production, you undoubtedly have some boilerplate language that you use when agreeing to produce material for the specific requests. Maybe it reads like this: “Non-privileged documents responsive to this request and not subject to an objection will be produced at a mutually convenient date and time to be determined among counsel.” Are you agreeing to produce “documents” in your responses? If so, don’t. Likely, in your general objections, you are already objecting to producing documents that contain proprietary, commercially sensitive or otherwise confidential information. Feel free to go ahead and include an objection to producing portions of documents that are not relevant and confidential.

Now – happy redacting everybody!

[1] Considering a document to be the basic discoverable unit (without, of course, mooting the discoverability of other types of units: sound files, photos, movies, very large rocks, etc. in the appropriate cases) may be a defensible position both considering industry practice of discussing discovery in terms of documents and the use of the term “documents” in other parts of the Federal Rules governing discovery. The author notes that the Court here did not defend or establish that position, however.

[2] Omitted from this discussion is the potential difficulty in defining what should or should not be considered a “document” in the age of electronic communications.

[3] Protective orders, obviously, are also designed to protect relevant confidential material that cannot be redacted out at all because it is relevant.

[4] Fed. R. Civ. P. 26(f)(1). Conference Timing. Except in a proceeding exempted from initial disclosure under Rule 26(a)(1)(B) or when the court orders otherwise, the parties must confer as soon as practicable—and in any event at least 21 days before a scheduling conference is to be held or a scheduling order is due under Rule 16(b).