Fact Versus Opinion Work Product

Rule 26(b)(3) sorts a lawyer's file into two tiers, and the tier a document lands in decides what a requester must show to get it. Subsection (A)(ii) opens the first tier to a party that shows substantial need for the materials to prepare its case and an inability, without undue hardship, to obtain their substantial equivalent by other means.¹ Subsection (B) guards the second: if the court orders discovery of those materials, it must protect against disclosure of the mental impressions, conclusions, opinions, or legal theories of a party's attorney or other representative concerning the litigation.² Courts call the first tier fact (or ordinary) work product, meaning tangible material that includes facts but not an attorney's mental impressions. They call the second opinion (or core) work product, meaning material, tangible or intangible, that includes or reflects an attorney's thoughts, such as impressions, theories, and conclusions. Fact work product yields to the (A)(ii) showing, while opinion work product is almost wholly immune from discovery.³

Because the burdens differ so sharply, the label is usually the real fight. The resisting party calls a contested document opinion work product, and the requester calls it fact work product. Courts settle the dispute by asking what disclosure would reveal. Subsection (B) protects the thinking of “a party's attorney or other representative,” so a document's author does not decide its tier.⁴ Its content does: a document that shows counsel's evaluation belongs in the second tier, and a document that records facts belongs in the first.

The second tier cannot grow to cover every document that hints at strategy. In In re San Juan Dupont Plaza Hotel, the First Circuit refused to treat every item that may reveal some inkling of a lawyer's mental impressions, conclusions, opinions, or legal theories as opinion work product, because that rule would let the exception swallow the rule. The heightened protection applies when disclosure creates a “real, nonspeculative danger” of revealing the lawyer's thoughts. An answer to a complaint shows the point. It discloses its drafter's mental impressions with some appreciable clarity, yet no one seriously suggests an answer could be hidden as work product, because pleadings are drawn knowing they will be served on the other parties. The same holds for complaints, counterclaims, third-party pleadings, requests for admissions, interrogatories, and motions. The rules do not compel disclosure of matters that would likely remain inviolate in the bosom of the lodge, and they merely accelerate disclosure of matters the other side will see in due course, which carry only a relatively low expectation of privacy.⁵ That limit frames the two problems that recur in practice: the records lawyers make of witness interviews, and the compilations they assemble from documents the other side can often obtain anyway.

Interview Notes and Memoranda

A lawyer's record of a witness interview is the clearest case of opinion work product, and Upjohn Co. v. United States explains why. The Government there sought notes and memoranda of company lawyers' interviews with employees, and the Court said that forcing an attorney to disclose notes and memoranda of witnesses' oral statements is “particularly disfavored because it tends to reveal the attorney's mental processes.”⁶ The reason lies in what a note-taker does. A witness talks at length, and the lawyer writes down part of it, in the lawyer's words, in the lawyer's order, with the lawyer's emphasis. The Court pointed to that feature when it quoted Hickman on “what he saw fit to write down regarding witnesses' remarks,” and it cited Justice Jackson's observation that the statement would be the attorney's language, “permeated with his inferences.”⁷ Reading such notes lets an opponent read the lawyer's theory of the case through the witness's words.

The Eighth Circuit's decision in Green places that protection inside the rule's two tiers. First, work prepared in anticipation of litigation by an attorney or the attorney's agent is discoverable only on a showing of need and hardship. Second, “core” or “opinion” work product, which encompasses the mental impressions, conclusions, opinions, or legal theories of an attorney or other representative of a party concerning the litigation, is “generally afforded near absolute protection from discovery.”⁸ Green adds that the protection reaches intangible work product as well as tangible, and because intangible work product includes counsel's thoughts and recollections, it is often eligible for the special protection owed opinion work product.⁹ A lawyer's interview notes fall in the second tier for the reason Upjohn gave, so a requester who has shown substantial need and undue hardship under (A)(ii) still meets the stronger protection.¹⁰

The label follows what a document reveals, so a record tied to a litigation investigation can still sit in the lower tier. In United States ex rel. Bibby v. Wells Fargo Bank, N.A., two short bullet points in meeting minutes summarized the result of an internal investigation prompted by litigation. The court called them fact work product: they stated a factual determination, evaluated no liability, revealed no legal strategy, and did not show the weeding out of unimportant facts that would expose the company's theory of the case. Fact work product is still protected, though. The court granted a protective order because the relators had not yet carried their heavy burden of showing substantial need and undue hardship, and the underlying facts remained open to them through other discovery.¹¹

When fact and opinion share one document, courts look for the seam. In In re Intel Corp. Microprocessor Antitrust Litigation, the Special Master reviewed interview materials in camera and found both core and non-core work product in them. The core portions were not so intertwined with the facts that each whole interview record had to be treated as core, because the facts could be easily separated. Had counsel's impressions been so thoroughly intertwined with the facts that an entire memorandum counted as opinion work product, the memorandum could not have been produced.¹² The shield protects thinking, and facts that share a page with it are reachable only when they can be lifted out without exposing it.

A lawyer's work can also hide inside a witness's signed words. When counsel interviews a nonparty witness and drafts an affidavit for the witness to sign, district courts divide on whether the signed affidavit is work product. In SFEG Corp. v. Blendtec, Inc. the court noted that it had no controlling circuit authority and described one line of cases, said to be the majority, that treats a signed affidavit as a statement of facts within the witness's personal knowledge and no longer an expression of counsel's opinion.¹³ A second line holds that an affidavit prepared from a lawyer's interview almost certainly reveals counsel's mental impressions, because counsel chose the topics, took the notes, and included the facts counsel considered significant to the legal theories in the case. The magistrate judge found the second line more persuasive. The judge added that even courts denying protection to signed affidavits generally protect the drafts, the communications with counsel about them, and counsel's drafting notes, so the character of the contents does not change at the moment of signature.¹⁴ Neither line controls, and an exam answer should say that either can be argued while naming the one the facts support. Classification did not end the dispute in SFEG either. The court denied the motion to compel because the requester had shown no attempt to interview the former employee or depose him, and a party that wants to know what a witness knows can do that work itself.¹⁵

Across all of these records, the protection attaches to the document and to what it reveals about counsel's thinking, while the facts the witness knows stay available through other discovery, as the relators in Bibby learned.¹⁶ A requester may therefore ask the witness what the witness saw. The requester ordinarily may not demand the lawyer's account of the interview.

Exam Analysis Steps

Because the tier depends on what a record would reveal about counsel's thinking, work an interview-record question in this order:

Identify the record: a lawyer's notes or memorandum of an interview, a short statement of an investigation's result, or a counsel-drafted affidavit or draft.
Ask whether disclosure would reveal what counsel chose to write down, emphasize, or leave out. That selection is what makes interview notes opinion work product.¹⁷
Where fact and opinion share the record, ask whether the facts can be separated from the impressions. If they can, the facts are reachable and the impressions protected. If the two are thoroughly intertwined, the whole is opinion work product.¹⁸
Apply each tier's test. Require the requester to show substantial need and undue hardship for the fact portions, leaving the underlying facts open to other discovery, and apply the subsection (B) shield to the opinion portions.¹⁹

Selection and Compilation

Rule 26(b)(3)(B) requires the court to protect counsel's mental impressions, conclusions, opinions, and legal theories even after other work product has been ordered produced,²¹ and a lawyer's choices about which documents to gather and how to arrange them can reveal exactly that kind of thinking. The difficulty is that the documents in a compilation are usually ordinary materials that parties and nonparties can be made to produce, so a demand for the compilation is a demand to see counsel's filter. SEC v. Collins & Aikman summarized the Second Circuit's position: selection and compilation may fall within work-product protection despite the general availability of the documents, but the protection is a “narrow exception” aimed at requests with “the precise goal of learning what the opposing attorney's thinking or strategy may be.”²² The cases below show what makes a selection revealing, what makes it harmless, and when a protected selection still yields.

Protection is strongest when a few documents are picked from a very large set. In James Julian, Inc. v. Raytheon Co. plaintiff's counsel assembled a binder to prepare current principals, officers, and employees for their depositions. It held selected documents the plaintiff had obtained from an opposing party in discovery, public records obtained through a Freedom of Information Act request, and documents the plaintiff had prepared itself.²³ The plaintiff did not resist producing the documents, and many had already been turned over. It resisted producing the binder, arguing that the selection and ordering reflected counsel's opinions and mental impressions. The court called the threshold question easy. The binder held a small percentage of the extensive documents counsel had reviewed, so selecting and ordering a few out of thousands “could not help but reveal important aspects of his understanding of the case,” and in a document-heavy case that process “is often more critical than pure legal research.” The binder was therefore protected work product, at least in the first instance.²⁴

The selection counts as the lawyer's thinking because of what the opponent already has. The opponent could obtain every document in the binder, so the binder adds only a ranking of which few documents counsel judged important enough to put in front of a witness. Producing it would let the opponent learn counsel's view of the case without reviewing the thousands of documents itself. Collins & Aikman observed a pattern consistent with that reasoning. Decisions protecting compilations typically involved ones prepared to assist witnesses, which expose counsel's thought process about witness preparation and strategy, while decisions rejecting protection involved compilations of documents not otherwise available or very large compilations.²⁵

A selection made by someone other than a lawyer sits in the middle of the range. In Coburn Group v. Whitecap Advisors LLC, a nonlawyer at the defendant company collected, selected, and organized information about the company's dealings with the requester, including where certain investors lived, in order to answer questions relayed from the company's attorneys. The court held the resulting email was work product but not opinion work product, because it disclosed no attorney's mental impressions, conclusions, opinions, or theories.²⁶ The investors' residences stayed discoverable facts, while the employee's work in gathering and organizing them was protected, so the email could be reached only on waiver or on the substantial-need showing of Rule 26(b)(3)(A)(ii). Selection and compilation can therefore support two levels of protection, depending on whether the arrangement exposes counsel's own evaluation or only the work of collecting facts.

A grouping that follows facts already pleaded reveals nothing new. In Collins & Aikman the SEC had produced about ten million pages, and an agency attorney had prepared 175 folders of documents supporting the factual allegations of the complaint. The defendant asked the agency to identify them.²⁷ The court held that the selection was not core work product. The highest protection reaches a compilation only when it is organized by legal theory or strategy, and a selection made by reference to pleaded facts reveals no more than the complaint did, since Rule 11 requires evidentiary support for a pleading's factual contentions. The court also rejected the agency's position that every document an attorney reviews is core work product. That theory leaves nothing to surround the core, and an attorney's assessment of relevance is the first step in answering any document request.²⁸

What moved the answer between the binder and the folders is what the selection reveals. In both settings the opponent could in theory reach the underlying documents. The Julian binder displayed priorities counsel had never announced, while the Collins folders displayed only what the complaint had already announced. Protection also does not end the analysis. Having found the folders unprotected as core work product, the court held in the alternative that even a protected compilation would yield, because the defendant plainly needed documents he had to review to prepare his defense, and a page-by-page review of ten million pages is undue hardship by any definition. Equity pointed the same way, since the court found it patently inequitable to make a party search ten million pages for documents its adversary had already identified as supporting the complaint.²⁹ The tier sets the weight of the requester's burden. The showing must be highly persuasive when core work product is sought,³⁰ and the ordinary showing under Rule 26(b)(3)(A)(ii) controls when the compilation is not core.³¹

A compilation can also lose its force when the other side will learn its contents anyway. In In re San Juan Dupont Plaza Hotel, a case-management protocol required a party to identify in advance the documents it would use as exhibits at depositions, and the party objected that the lists were opinion work product.³² The First Circuit held that the lists could not claim that status. The order decided when counsel's document selection would be disclosed, not whether, because once the deposition begins the questioner's choice of documents, and the stratagems it reveals, become obvious to everyone.³³ The court acknowledged that choosing deposition documents can be more crucial than legal research, but the same information would be gained at the deposition in any event. Like requiring pleadings, answers to contention interrogatories, pretrial exhibit and witness lists, and trial memoranda, the protocol merely adjusts the timing of disclosure.³⁴ The same court stated the governing contrast elsewhere: the key factor separating such material from protected opinion work product is that the information will probably come to light during the course of trial, if not before.³⁵

The San Juan Dupont lists and the Julian binder both involve documents chosen for depositions, and the difference between them follows the timing point. A binder used to prepare one's own witnesses stays in counsel's files unless someone orders it produced, while exhibits an examiner will mark must be shown to the witness and so to the other side. The First Circuit was reviewing a case-management order and not a request for production, so its reasoning serves best by analogy: a compelled disclosure that only accelerates what the proceeding will reveal intrudes little on counsel's privacy.

Exam Analysis Steps

Because a compilation dispute turns on what the arrangement discloses and on how hard the requester can reach the same result alone, work it in this order:

Identify what the requester wants, the underlying documents or counsel's selection and arrangement of them. Documents that ordinary discovery will produce leave only the filter at stake.
Ask whether the selection reveals counsel's understanding of the case or strategy that would otherwise stay private, as with a few documents chosen from thousands to prepare witnesses, or only repeats pleaded facts or an exhibit list that the deposition will expose anyway.³⁶
Place any protected compilation in its tier. Treat one organized by legal theory or strategy as core work product, and one that reflects only the collection of facts as ordinary work product.³⁷
Apply the matching showing of need and undue hardship, weighing the equities when the alternative is a search through a vast set that the adversary has already culled.³⁸