As readers of this blog know well, the major questions doctrine (MQD) has become a staple of contemporary administrative law practice. For the past two Supreme Court terms, the MQD has been invoked to vacate administrative actions canceling student loan debt, imposing mortgage foreclosure moratoria, requiringvaccination against COVID-19, and controlling carbon dioxide pollution from power plants. At the same time, the MQD continues to ripple through the lower courts, fundamentally changing how challenges to regulatory authority are framed and even whether they occur at all. What is remarkable about all of this is that the MQD remains deeply undertheorized: we are, in a sense, building the ship as we sail it. 

For most observers, and indeed for at least some of the Justices (most notably, Justice Gorsuch), the MQD seems to be a textbook example of a substantive canon of statutory interpretation. Substantive canons are typically framed as allowing judges to depart from the plain meaning of statutes in order to promote some normative or constitutional value. However, textualists and non-textualists alike have begun to raise concerns about the provenance of substantive canons, and the MQD’s branding as a substantive canon is no exception. As Justice Barrett recently put it, “some articulations of the major questions doctrine on offer—most notably, that the doctrine is a substantive canon—should give a textualist pause.”

That’s why it was so significant that Justice Barrett authored a concurrence in Biden v. Nebraska to articulate why she could join a decision deploying the MQD despite her textualist commitments. As others have noted on this blog, Barrett sought to re-brand the MQD as a “common sense” linguistic doctrine that helps readers find the semantic meaning of statutes rather than a substantive one that gives judges license to depart from it. In other words, Barrett articulated an argument that the MQD is linguistic in nature, and in doing so she joined other textualists who have similarly sought to pivot to safer ground for those who have committed to the primacy of textual meaning over judicial policymaking.

In an article we have recently posted on SSRN, we empirically assess the case for a “linguistic MQD.” And, we should add, we do so because, although it might at first sound fanciful, it is entirely plausible that an interpretive canon might have both a substantive and linguistic basis. Perhaps the MQD is one. Textualists increasingly anchor the legitimacy of linguistic canons to a hypothetical “ordinary reader” of statutes. On this theory, when linguistic canons track the tendencies of ordinary readers in a given linguistic context, they have legitimacy. In Duguid v. Facebook, Justice Alito recognized this kind of claim for what it is—an empirical hypothesis—and called for greater attention to whether linguistic canons like the series qualifier canon are accurate representations of ordinary readers’ thinking about language.

Ultimately, Justice Barrett and other proponents of the linguistic MQD make just such empirical claims about how ordinary readers approach language. These claims are almost entirely untested. In Biden v. Nebraska, Justice Barrett offered assertions about the ordinary reader’s “common sense” about general delegations. Barrett’s primary example is the babysitter hypothetical: 

Consider a parent who hires a babysitter to watch her young children over the weekend. As she walks out the door, the parent hands the babysitter her credit card and says: “Make sure the kids have fun.” Emboldened, the babysitter takes the kids on a road trip to an amusement park, where they spend two days on rollercoasters and one night in a hotel. Was the babysitter’s trip consistent with the parent’s instruction? Maybe in a literal sense, because the instruction was open-ended. But was the trip consistent with a reasonable understanding of the parent’s instruction? Highly doubtful. In the normal course, permission to spend money on fun authorizes a babysitter to take children to the local ice cream parlor or movie theater, not on a multiday excursion to an out-of-town amusement park. If a parent were willing to greenlight a trip that big, we would expect much more clarity than a general instruction to “make sure the kids have fun.”

To be sure, this recognition that context can influence interpretation is a constructive and welcome statement, especially coming from a textualist Justice. However, it is critical to distinguish between this more general claim (context can matter) and the more specific and tendentious claim that ordinary linguistic practice includes a contextual restriction of delegations as strong as the MQD. On one reading of the babysitter argument, it is simply a “motte” and “bailey” fallacy, conflating one proposition that is easy to defend (the motte) with one much harder to defend (the bailey). The bailey here is much stronger than some generalization about the importance of context; it is something like: “Ordinary people understand general delegations to X to be limited to only the most reasonable ways to X, absent further contextual support for X.” But do they? This is, as far as we are aware, an entirely untested theory about ordinary readers.

The other major version of the linguistic defense of the MQD comes from Ilan Wurman, who has argued that the MQD simply generalizes how ordinary readers take account of the stakes of interpretation. Although Wurman’s argument is complex, for our purposes it relies on two empirical premises: 1) that the ordinary reader’s knowledge is sensitive to high stakes, and 2) that the ordinary reader’s understanding of textual clarity is sensitive to high stakes. Specifically, higher stakes leads to (1) less knowledge and (2) less textual clarity. As we discuss in our paper, the first claim is the subject of a large literature in epistemology, where philosophers have debated whether knowledge is lost when the stakes of knowledge are heightened. For instance, imagine that a person has a check to deposit at the bank on Friday, and they can either go to the bank that day or the day after (Saturday). Stakes sensitivity would mean that a person would “know” it is open Saturday when there are low stakes (nothing bad will happen if the bank is closed and the check is not deposited until Monday), but not when there are higher stakes. Imagine a higher stakes situation: The check has to be deposited before Monday, to cover a large debit on the account. In that high-stakes case, philosophers have proposed, the person does not know the bank is open on Saturday. 

The legal literature on high stakes interpretation relies on reactions from academic philosophers to examples like the bank cases. But the linguistic MQD depends on a claim about the ordinary reader’s knowledge.  Experimental philosophy studies have surveyed ordinary people, and the record is decidedly mixed about whether stakes impact ordinary knowledge. And, critically, Wurman’s second claim—that judgments about the clarity of legal text are a form of knowledge that are driven by the same stakes sensitivity as any knowledge claim—is entirely untested. 

In our article, we conduct two original survey experiments that fill the empirical gaps left by Barrett’s and Wurman’s linguistic defenses of the MQD. The results are bad news if you are a believer in the linguistic MQD. Consistent with some prior experimental philosophy literature, we find a small stakes effect on ordinary knowledge claims. But such a small effect is hard to interpret in favor of the MQD. In the “low stakes” bank case, 95% of people attributed knowledge, while in the “high stakes” case 86% did. So does stakes impact knowledge for the “ordinary reader”? Not for the vast majority of ordinary participants. It would be more accurate to say that the vast majority of ordinary participants did not factor high stakes into their assessment of what was known.

More importantly, we find that the second claim about clarity judgments and interpretive stakes finds no support in ordinary readers. Higher-stakes don’t make rules less clear. A central premise of the “high stakes” MQD defense thus reflects an unordinary epistemology.

Turning to Barrett’s argument, we gave survey respondents the babysitter hypothetical with five experimental treatments designed to assess whether ordinary readers implicitly limit a general delegation of authority to the babysitter to “use this credit card to make sure the kids have fun” to only the limited set of the most reasonablepossible forms of fun. Ordinary people did not share Barrett’s “common sense” about the trip to the amusement park (only 8% of percent of respondents believed that the babysitter broke the rule by taking the kids to an amusement park). And even more extreme actions, such as bringing an animal entertainer and live alligator into the house for a showing, were also overwhelmingly judged to be within the rule. Moreover, ordinary readers made these judgments about the text of the delegation despite believing that both of these actions were not as reasonable as other actions, such as renting a movie for the kids—so, in ordinary understanding the authorization of delegations is not limited to the most reasonable actions. Importantly, it was not as if ordinary readers did not see anything as inconsistent with the rule. They judged that the babysitter was not authorized to simply rent a movie for himself.

These results undermine the empirical claims at the heart of the linguistic MQD. To be sure, it might be possible to articulate different versions of the linguistic defense. It might also be possible to build a case for a linguistic canon that is informed by generalizations about legislative intent in delegating authority to agencies (indeed, Wurman offers this as an alternative, and Barrett nods to it), although that tact would presumably not appeal much to the textualists who have generally rejected evidence of legislative intent that is not objectified in some way in the text itself. But, as it is, the linguistic defenses on the table appear indefensible in light of ordinary people’s actual “common sense”. 

Our research has broader implications beyond the MQD. Some of our evidence suggests that if the goal is to make administrative law doctrine congruent with how ordinary readers think about delegations, something like an anti-MQD might make more sense. Specifically, we find that ordinary readers are likely to say that the babysitter violates the instruction when the babysitter fails to use the credit card at all (simply playing card games with the kids to have fun). In other words, ordinary readers appear to be much more skeptical when agents under-use their authority than when agents take actions that fall under the instruction’s meaning but may not be among the set of most reasonable reactions. 

These ordinary intuitions about delegations have a lot more in common with the now largely defunct rule that “remedial” statutes should be given a liberal construction than they have in common with an anti-regulatory MQD. They also would seem to call into question administrative law’s many doctrines that insulate agency inaction and nonenforcement from judicial scrutiny. Although these questions are for another day (and we are not committed to outsourcing administrative law to laypeople), it is important to note that taking the ordinary reader seriously may lead in a very different direction than the textualist Justices behind the MQD have supposed.

Writing in response to the amicus briefs in Loper Bright Enterprises v. Raimondo, Nicholas Bednar noted that “[i]t is one thing to lack evidence and speculate about the potential consequences of a proposed doctrinal change; it is quite another to recite unsupported empirical claims as certain in furtherance of that change.” We agree with Bednar’s call for greater skepticism of “loose empirical claims” as a motivator for doctrinal change. As of right now, there is no evidence to support a “linguistic MQD,” and we do not see how one could claim to be a textualist and support this rebranding of the doctrine. We acknowledge that textualists may ultimately choose the MQD over their commitments to the primacy of statutory text, but there is value in laying that choice bare.

Kevin Tobia is an Associate Professor of Law at the Georgetown University Law Center. Daniel E. Walters is an Associate Professor of Law at the Texas A&M University School of Law. Brian Slocum is the Stearns Weaver Miller Professor at the Florida State University College of Law.