Supreme Court 2017 Statutory Term in Preview: Definitional "Ambiguity" in Digital Realty Trust, Inc. v. Somers, Part 1

The subject of the following series of posts will again be textualism, and what I intend to again persistently asseverate, to borrow a phrase from a passage of Henry James's dialogue (the character saying it is supposed to be rather simple!), is that textualism is a less constraining methodology than some of its advocates claim, but a more constraining one than some of its false friends want it to be.  The particular textualist problem I want to address here is statutory definitions that seem to make a peculiar meaning out of uses of the defined terms to which they apply.  

This problem is the subject of the statutory blockbuster (I speak unironically) of the upcoming Supreme Court term, Digital Realty Trust, Inc. v. Somers, as it was of the Court's decision in Bond three years ago, as it was in part, in a fashion that escaped almost everyone's attention, of the Court's decision one year ago in Puerto Rico v. Franklin California Tax-Free Trust.  I will propose a middle course between the definitional nihilism that insists a definition is a definition no matter what, on the one hand, and the definitional immolation of Bond on the other—the course that Justice Thomas steered for the Court in Franklin California Tax-Free Trust and that Justice Scalia steered for the Court in Utility Air.

That middle course, in sum, is this.  Assuming a typical definitional clause, a statute can never be ambiguous on whether a statutory definition applies to that statute's use of the defined term that definition defines.  So long as a statute says where its definition applies—"in this section," "in this chapter," "in this Act"—the definition unambiguously applies where the statute says it does.  "In this section/chapter/Act, the following definitions shall apply" is not ambiguous language, and such language is the only language relevant to deciding whether a statute's definition textually applies to a use of a defined term.  Uses of the defined term that don't seem to match the definition only tend to show that Congress may have made a scrivener's error in using a term elsewhere defined.

However, a statute's application of a definition to a particular use of a defined term may very well be a correctible scrivener's error.  Scrivener's errors of this sort happen often; they happen when drafters use a term and forget how that term was previously defined.  The error here, technically, is neglecting to use a different, non-defined term, or neglecting to revise the definitional clause to except certain sections of a bill from its application.  Particularly in longer statutes, and particularly in the instance of terms that have non-technical meanings and come easily to mind, it's quite unsurprising that drafters forget what a definitional section says and use terms in ways that clash with their statutory definitions.  

But while we shouldn't be surprised that these errors happen, before we correct them we shouldn't demand anything less than the certainty we demand to correct other kinds of scrivener's errors.  Definitions that don't fit uses of defined terms, again, are not a species of ambiguity; there is nothing ambiguous about saying that a definition shall apply in this section of this statute.  They are accidents of drafting that permit courts to depart from the meaning of otherwise clear texts, and before courts do so, they must be certain that an accident occurred.

Finally, because the two options in a case of possible definitional mismatch are either following the text's plain meaning, or correcting a plain error, cases of definitional mismatch are not cases where Chevron has any role to play, or where, for that matter, any other ambiguity-resolving canon has any role to play, including clear-statement rules or constitutional avoidance.  Statutes are never ambiguous on whether their definitions are mismatched; the only "ambiguity" to speak of in such cases is uncertainty about whether Congress intended to write what it wrote, not ambiguity in the meaning of what it wrote.  But under current doctrine, that sort of ambiguity resolves in favor of assuming Congress did intend to write what it wrote; the only ground for correcting drafting error is certainty that one happened.  

Further, uncertainty about whether Congress intended to enact the text it wrote or some other variant is particularly not a ground for Chevron deference.  Chevron assumes that Congress enacts text that is indeterminate on some question and leaves that question open for further policy choice, thereby delegating that question to an agency to decide (not interpret).  Ambiguities about what text Congress meant to write are not delegations to make interstitial law, but rather ambiguities about the content of the determinate policy choice Congress did make.

1.  The definitional problem in Digital Realty.

There is a section buried in the Dodd-Frank Act that provides "securities whistleblower[s]" a small but potent suite of "incentives and protection[s]," quoting the section's title.  The question in Digital Realty is whether that section's definition of "whistleblower" applies to the section's whistleblower-protection's use of the word "whistleblower."  Though section 922 of the Dodd-Frank Act takes a while explaining itself, what it gives securities whistleblowers is fairly simple.  

First, if a securities whistleblower gives the SEC, and the SEC alone, the goods on someone—as the statute puts it, if "1 or more whistleblowers . . . voluntarily provided original information to the Commission that led to the successful enforcement of [a] covered judicial or administrative action"—he will get an award of between ten to thirty percent of the monetary sanctions the SEC recovers.  Section 922 proceeds to lay out the details of this award scheme, and the money to pay for it, in seven subsections, (b) through (g) and (i).  Subsection (j), the last subsection of the statute, is a grant of rulemaking authority that doesn't mention "whistleblowers," subsection (a) contains the section's definitions, to which we'll get, and subsection (h) is the section's eponymous "protection," an unusually generous anti-retaliation and confidentiality provision at the heart of Digital Realty Trust.

Subsection (h), entitled "Protection of whistleblowers," does a few things.  First, it prohibits three types of retaliation against "a whistleblower."  Employers may not retaliate against whistleblowers for any "lawful act done by the whistleblower" in providing the SEC information, assisting or testifying in an SEC investigation, or for "making disclosures that are required or protected under the Sarbanes-Oxley Act of 2002 [cross-references omitted] and any other law, rule, or regulation subject to the jurisdiction of the Commission."  Sarbanes-Oxley, crucially, protects disclosures of violations of securities law to non-SEC federal agencies, Congress, or anyone with supervisory authority over the employee.  Does that mean that 922(h) will offer retaliation protections to anyone who discloses violations of securities law to internal auditors?  Put a pin in that and read on.  

The balance of subsection (h) does three things.  First, it provides victims of the proscribed sorts of anti-whistleblower retaliation with a generous cause of action, with a six-year statute of limitations and a three-year discovery rule in case of late discoveries, double back pay, reinstatement, and fee-shifting.  Second, it prohibits the SEC from disclosing information that could be "reasonably  . . . expected to reveal the identity of a whistleblower," though certain inter- and intra-governmental disclosures are permitted provided that the recipients of the disclosures maintain the same level of confidentiality.  Third, the last sentence of subsection (h) is a sort of Ninth Amendment for "whistleblower[s]," preserving all rights they have under all other sources of law.

I've belabored the various things the statute does for whistleblowers for a reason, but long before I say what that reason is, I want to discuss the definition of "whistleblowers."  And before I say what the definition is, I want to say why the definition is needed.  Though this might not be immediately obvious, the repeated usage of "whistleblower" in this statute is quite eccentric statutory drafting and wouldn't work at all without a definition.  The word whistleblower, in an ordinary-language sense, is a non-technical term of extraordinary vagueness, encompassing a vast array of types of whistleblowing on a vast array of subjects to a vast array of people; no competent statutory drafter would use the word without giving it a technical, non-ordinary definition.**

** The Solicitor General, arguing that "whistleblower" keeps its ordinary meaning in subparagraph (h)(1)(A) of section 922, claims that "numerous federal statutes use 'whistleblower' in the ordinary sense."  Not so.  His sole citations for that claim are: (1) a findings and purpose section of the Whistleblower Protection Act of 1989 that uses the word twice, and in the first instance defines whistleblowers as "those individuals who make disclosures described in such section 2302(b)(8)"; (2) the title of an operative section of the Act, section 5; (3) section 4 of the Act, which never uses the word and seems to have been cited by mistake.  Indeed, it's rather telling that even the Whistleblower Protection Act never uses the word "whistleblower" in the text of any of its operative sections.

As for other federal statutes, some thirty-three statutes (I've checked) use the word "whistleblower."  Of the statutes that really refer to whistleblowers in some operative sense (rather than generically referring to "whistleblower protections"), virtually all of them define whistleblowers either formally (and almost always in terms of disclosures to a particular agency) or informally, by describing the whistleblowers to be protected with great specificity.  I am aware of just one exception to this rule, in the Inspector General Act of 1978, and even that one's debatable, as the provision strongly implies that the whistleblowers Congress is talking about are whistleblowers who talk to Inspectors General.  More typical of congressional drafters' sense of the propriety of using "whistleblower" sans definition in statutes is the Internal Revenue Code's reference to "any provision of federal law (popularly known as whistleblower protection provisions)" that affords certain protections which the Code proceeds to define in more technical terms. [End of footnote.]

To be sure, perhaps you could get by without a definition if you were very clear on what sorts of whistleblowing a whistleblower would get an award for, and what sorts of whistleblowing he'd be protected from retaliation against, and what a whistleblower had to say to entitle himself to confidentiality.  But then, if the meaning of "whistleblower" were to be derived exclusively from context, you wouldn't have to use "whistleblower," would you?  Instead you'd say that a "person" would get an award for providing certain information to the SEC, and an "employee" would be protected from retaliation for disclosing certain information to certain people, and a "person" would have his identity kept confidential if he disclosed this or that—not that a "whistleblower" would be protected if he blew the whistle in very specific ways.  

That would be a somewhat harder statute to draft, though, and in fact a close reading of the statute reveals that as presently drafted it would collapse without a definition of "whistleblower."  The awards subsection doesn't say quite what a whistleblower provides the SEC—information that leads to a successful enforcement, yes, but about what?—the retaliation subsection doesn't even say that much about what information the whistleblower has to give the SEC, just that it's got to be "information [provided] in accordance with this section," and the confidentiality subsection doesn't say a word about what earns a "whistleblower" confidentiality.  So the statute truly hinges upon the definition of "whistleblower" in all operative respects.

Unsurprisingly, given the SEC-focus of most of the operative provisions in the statute, it turns out that section 922's definition of "whistleblower" is exclusively about SEC, securities-law whistleblowing.  A "'whistleblower' means any individual who provides, or 2 or more individuals acting jointly who provide, information relating to a violation of the securities laws to the Commission, in a manner established, by rule or regulation, by the Commission."  That clears up what sort of information the whistleblowers have to provide: information of securities-law violations.  Where does this definition apply?  Just in this section, actually, in the context of the small handful of incentives and protections it provides: "In this section the following definitions shall apply."  And so it turns out that only SEC, securities-law whistleblowers get the benefit of section 922's generous retaliation provision, of its awards, and of its confidentiality protections.

But wait, you ask:  doesn't this definition make one third of the anti-retaliation provision meaningless?  Weren't "whistleblowers" to be protected from retaliation against Sarbanes-Oxley-protected disclosures, which included disclosures to mere internal auditors?  Could Congress really have meant that only SEC whistleblowers would be protected from whistleblowing to auditors or other agencies?  And if it had, wouldn't that make that part of the anti-retaliation provision surplusage?  After all, Congress had already offered "whistleblowers" protection from retaliation against SEC whistleblowing, so what work does it do to protect SEC whistleblowers from retaliation against internal reporting?  Is there any SEC whistlebower who would be fired because of his internal report and not because of his more damaging SEC whistleblowing?  And given all these concerns, doesn't it seem as if section 922's SEC-specific definition of "whistleblower" can't apply to 992(h)(1)(A)'s prohibition of retaliation against whistleblowers who blow internal whistles?  That's the question presented by Digital Realty Trust.

2.  The scope of the definition in Digital Realty (and most any definition) is unambiguous.

All fair questions, though all, as we'll see, with answers, but for now I want to ask a different question:  do these questions actually make whether 922's whistleblower definition applies to 922(h)(1)(A) ambiguous?  Now, a small group of very smart people, including Judge Kavanaugh, have begun to argue that this question should have no bearing on interpretation; if legislative history, policy concerns, or even inferences of intent from surplusage are relevant to interpretation, they should be relevant even when a statute's text is clear, but if they're not so relevant, we shouldn't be bothering with them even when a statute's text is ambiguous.  I think this is all wrong and quite radical, but whether these people are right or not, they're not describing the law as it exists, as they of course readily admit.  Under current law, whether 922's definition's application to 922(h)(1)(A) is ambiguous all but decides the case, in the following four ways.

First, whether 922's definition's application to 922(h)(1)(A) is ambiguous determines, under current law, whether the SEC's interpretation of "whistleblower" in the anti-retaliation provision to mean anyone who does any of the acts it protects will receive Chevron deference.  

Second, whether the definition's application to 922(h)(1)(A) is ambiguous determines whether the surplusage argument I've sketched does any work; if a statute unambiguously contains surpulsage, arguments against surplusage are unavailing, but if it's ambiguous whether it does or not, courts may be influenced by anti-surplusage presumptions to avoid reading the statute to contain surplusage.  

Third, whether the definition's application is ambiguous determines whether policy concerns, or legislative history, do any work; in our half- or pseudo-textualist world, if the statute's ambiguous, legislative history can be used to decide what Congress meant, but if it's not, legislative history can't be used.  

And finally, if 922(h)(1)(A) unambiguously incorporates section 922's definition of "whistleblower," the non-SEC whistleblower in Digital Realty Trust suing his employer under Dodd-Frank for an internal disclosure is going to lose unless he can show that the definition's application 922(h)(1)(A) is (a) unmistakably a drafting error or (b) absurd (and even that probably overstates things—he probably has to show that it's an unmistakable drafting error because it's absurd).

So is it ambiguous whether 922's definition of "whistleblower" applies to the use of "whistleblower" in 922(h)(1)(A)'s retaliation provision?  No, not at all, and in fact, the only sort of question about whether a definition applies to a use of its defined term is whether the definition's application to that term is a scrivener's error.  An "ambiguity" about whether a definition applies to a use of a defined term is really just a doubt about whether the drafter accidentally forgot to except that use from the scope of the definition.  And because that doubt is a doubt about whether there's a drafting error—the existence of which doctrine (correctly, in my view) requires certainty before courts can correct it—such "ambiguities" always must be resolved in favor of applying the definition.  The only definitions that don't apply to uses of the terms they define, on this view, are the ones that were undoubtedly applied to those uses in error.

Why am I so sure of all this?  Consider section 922's definition of "whistleblower" as an example.  As Judge Dennis Jacobs correctly points out in his dissent on this question, the relevant language to whether that definition applies to the anti-retaliation provision in 922(h)(1)(A) is not the anti-retaliation provision, but what the definition itself says about its scope.  And what it says is that "[i]n this section the following definitions shall apply."  Among the definitions that follow is the whistleblower definition, and 922(h)(1)(A) is in "this section," section 922. 

What this unambiguously means is that the meaning of "whistleblower" in 922(h)(1)(A) is that provided by the whistleblower definition in 922(a)(6).   There is no alternate reading.  "In this section the following definitions shall apply" can't be read to mean "in this section the following definitions may apply," or, "in this section less subsection (h)(1)(A), the following definitions shall apply."  These are not meanings that any reasonable speaker of English can give to "shall" or "in this section."  They are not, crucially, things that anyone would ever intentionally use the sentence "in this section the following definitions shall apply" to say—a critical test of ambiguity.  Anyone trying to communicate the message that the whistleblower definition would apply throughout section 922 less section 922(h)(1)(A) would say that.  No one who wanted "whistleblower" to mean something different in 922(h)(1)(A) and realized that she had written that the whistleblower definition "shall apply" "in this section" would hope for courts to figure out what she wanted from the fact that she protected non-SEC whistleblowing from retaliation in (h)(1)(A)(iii).  Rather, she would correct the definition.  

The definition's statement of its scope is not, to be clear, merely unambiguous in a literal or semantic sense; if that were all, that would not exhaust interpretation.  It's also unambiguous in the sense that no one could intentionally use the statement to imply something narrower than what it literally means, and unambiguous in the further sense that no one could use the statement to mean something narrower on the mistaken belief that it has a narrower meaning.  

While we can imagine, for example, people using phrases like "convicted in any court" to impliedly mean any court in the United States, or "Pick up every object on your floor" to impliedly mean something more like "pick up every object but the furniture," given that people use "any" and "every" all the time in reference to some impliedly limited domain, there is no similar phenomenon of people referring to "this section" and implicitly meaning "this section less its most important provision."  And while Congress will sometimes use a word in a way that suggests it might be mistaken about its meaning and would have used a different word absent its mistake (which I have contended is a form of ambiguity), like the time it gave "defendants" special protections against impeachment in the Rules of Evidence and seemed to mistakenly believe that "defendants" meant criminal defendants, there is no conceivable mix-up of that sort about the meaning of "this section" or "shall apply."  

Again, a good litmus test for ambiguity that embraces far more than semantic ambiguity is to ask whether a person trying to say something could intentionally use the language at issue to say it.  If they couldn't—if the only way you can imagine someone using the language at issue to say something is their accidentally writing the wrong words, or intentionally writing what they wrote, changing their mind about what they wanted to say, and forgetting to correct what they wrote—then the language is unambiguous and the only way to escape its meaning, if you believe Congress meant to convey that point, is to invoke drafting error.  And here, the only way we can imagine a drafter using "in this section, the following definitions shall apply" to mean that the whistleblower definition wouldn't apply to the core of "this section" is a sort of drafting error, probably the sort where Congress, after writing the definition, used "whistleblower" more broadly in the anti-retaliation provision and forgot that the definition existed.

3.  The Court's confusion of ambiguous intentions with ambiguous meaning in Bond.

The Court had a very different view of these matters in Bond.  There, the Court thought it ambiguous whether the extraordinarily broad definition of "chemical weapon" in the Chemical Weapons Convention Implementation Act, which included any toxic chemical, applied to the operative provision of that act.  Claiming that "dissonance" between the ordinary meaning of "chemical weapon" and its statutory definition made it ambiguous whether the term's ordinary meaning or its definition controlled, the Court resolved the ambiguity, on the basis of ambiguity-resolving federalism canons, by functionally axing the definition from the entire statute.  

To be precise on how radically sweeping this judicial repeal of the act's definition of its key term was, a point which I don't think has ever been fully appreciated, the Court technically held that the statute's definition of chemical weapons did not apply to section 229 of the statute, which prohibited the use, possession, ownership, or production of chemical weapons; in its place, the Court interpolated an ordinary-meaning definition.  Now, there are only six other sections in the statute, and all of their usages of chemical weapon are completely parasitic on section 229's, section 229 being the only truly operative provision in the statute.  Going through those sections:

Section 229A is a penalties section for violating section 229 that doesn't mention chemical weapons.

Section 229B concerns criminal forfeiture of chemical weapons in the event of a conviction under section 229 for owning or possessing them; if the definition doesn't apply to section 229, it can't apply to section 229B.   

Section 229C is a savings clause for pepper sprays that would be completely unnecessary if not for the statute's broad definition of chemical weapon; it also doesn't mention chemical weapons.  

Section 229D, also not mentioning chemical weapons, gives the United States a cause of action to seek an injunction against the conduct prohibited in section 229; it is completely parasitic on section 229's use of "chemical weapon" and the meaning it has there.

Section 229E says that the Attorney General can ask the Department of Defense for assistance in enforcing section 229 "in an emergency situation involving a chemical weapon."  The predicate for that authority is enforcing section 229, so it too is completely parasitic on section 229's use of chemical weapon; the Attorney General couldn't seek assistance in an emergency involving a chemical weapon that satisfied the statute's definition but fell outside the Court's reading of chemical weapon in section 229, or there would be no section 229 enforcement to assist.  

Finally, section 229F is the statute's definitions section.  It uses "chemical weapon" (outside of its definition of "chemical weapon") only in defining "purposes not prohibited under this chapter," a phrase that's a component part of and carve-out from 229F's definition of "chemical weapon."  

The following, then, is a non-hyperbolic statement; after the Court's ostensibly interpretive holding in Bond that 229F's definition of chemical weapon doesn't apply to section 229, there is no use of chemical weapon in the entire statute to which it does apply.  It is simply as if Congress had never enacted the definition at all.

Now, was there some sort of ambiguity as to whether the definition of chemical weapon in the Chemical Weapons Convention Implementation Act actually applied to the Chemical Weapons Convention Implementation Act?  The question is like asking whether the term "endangered species" in the Endangered Species Act means what Congress said it did in the ESA's definitional section, or some other preferred meaning of your imagination.  The Chemical Weapons Convention Implementation Act is codified in chapter 11B of title 18.  Section 229F begins, "In this chapter:  The term 'chemical weapon' means the following . . . ."  Can "[i]n this chapter: The term 'chemical weapon' means the following" mean "In this chapter: The term 'chemical weapon' doesn't mean the following, but instead means whatever it means ordinarily"?  Not so much.  

What Chief Justice Roberts described as an ambiguity of statutory meaning is really an ambiguity in congressional intention.  And while there was no ambiguity of the former variety, the ambiguity of the latter variety was very real; the Court's mistake wasn't in finding some kind of ambiguity at all, but in confusing unactionable doubt over whether Congress really meant to write what it said with doubt as to the meaning of what Congress did say.  As Heather Gerken, hardly an unforgiving textualist, has written, "[t]he statute's language was crystalline"; the Court "thought the statute was ambiguous" because "it couldn't bring itself to believe that Congress had, in fact, passed a statute broad enough to reach Bond's conduct."   

Did Congress really make it a federal crime to use (with exceptions for peaceful purposes) any "toxic chemical" capable of causing permanent harm or temporary incapacitation to human or animal life?  Absolutely.  Did Congress really want or consciously intend to make any use of a toxic chemical a violation of the Chemical Weapons Convention Implementation Act?  Hard to say, and perhaps hard to believe; most likely, Congress implemented the titular Chemical Weapons Convention by copying its definition of chemical weapon without entire awareness of its breadth or the consequences.  And would Congress have really wanted to make it a violation of the Chemical Weapons Act to smear mild-skin-burn-inducing chemicals on someone's mailbox, were the question put to them?  Very possibly not.  But these uncertainties about what Congress wanted to do, or what it would have wanted to do had it thought harder about a court's present problem, aren't ambiguities—not for a textualist anyway.  They only matter, really, to a sort of purposivist that's no longer supposed to exist.

Two final comments on Bond.  First, what of Roberts's argument that it isn't "unusual to consider the ordinary meaning of a defined term, particularly when there is dissonance between that ordinary meaning and the reach of the definition"?  Is there nothing to that?  As Justice Scalia said in his separate concurrence, it is most unusual to consider a defined term's ordinary meaning for the Court's purposes; prior to Bond the Court had only considered the ordinary meaning of defined terms to shed light on the meaning of ambiguous definitions, not to delete them.  

But more fundamentally, the argument is nonsense.  To say that there's dissonance between the ordinary meaning of chemical weapon and what the statute says it "means," and that it's therefore ambiguous if "chemical weapon" in that statute means what the statute says it does, is like saying that a child's answer on his homework that "3 X 7 = 37" is ambiguous because there's profound dissonance between "37" and the ordinary meaning of "3 X 7."  One could talk of scrivener's error in some similar cases; if the child wrote that "3 X 7 = 10," you might wonder if he meant to write a plus sign, or maybe even if he thought the multiplication sign is the addition sign.  But the meaning of the equation would still be unambiguous and the only way you could dissolve the dissonance between "3 X 7" = 10 would be to read the use of the multiplication symbol as an accident or mistake.  

Similarly, imagine Congress defined a vegetable, in the school-lunch statute, to include any food or beverage containing a food syrup derived from a vegetable—e.g., a soda containing corn syrup.  The dissonance between what vegetable means in any ordinary sense and how Congress defined it would be profound, much more so than the dissonance that worried the Court in Bond, and the policy, at least to my mind, would be vastly more offensive.  But would there be any ambiguity about whether "vegetables" in the statute included soda?  And if a parent sued a school for providing soda-vegetables as vegetables on the theory that sodas weren't statutory vegetables because they weren't ordinary-language vegetables, the "dissonant" definition notwithstanding, wouldn't (and shouldn't) their lawyer be sanctioned for frivolously asking the court to rewrite the school-lunch law?  What distinguishes Bond?

Finally, why didn't Roberts suggest that equating chemical weapons with any harmful toxic chemical was likely some sort of mistake, rather like "3 X 7 = 10" might be, instead of claiming that the statute was ambiguous as written?  Because making that argument, I think, would be impossible.  

How, after all, in legal drafting, do definitions get mistakenly (technically, accidentally—linguistic mistakes are confusions about meaning, not drafting errors) applied to defined terms?  Usually, you define a term, use it to mean its natural meaning, and forget you already defined it to mean something rather different.  That account of definitional accident, was unavailable in Bond, because it's unthinkable, at least extremely improbable, that Congress or its drafters carefully defined "chemical weapon" in section 229F, forgot how it had been defined, and then wrote each of sections 229, 229B, and 229E, the small handful of adjacent sections that mention chemical weapons, on the mistaken assumption that their central, hardly self-defining term either wasn't defined or had been defined in a very different way.  To impute that sort of accident to Congress requires an assumption of staggering congressional incompetence or collective drunkenness.

In my next post I'll address when definitions can be corrected, Chevron's relationship to claims of definitional error, and hopefully get around to how Digital Realty itself should be decided.