Who decides what laws passed by Congress mean in practice? As almost everyone knows, usually the courts fulfill this role, citing Marbury v. Madison – “It is emphatically the province and duty of the Judicial Department to say what the law is.” Readers of Notice and Comment are undoubtedly familiar with the large modern exception to this rule created by Chevron, under which courts defer to administrative agencies’ interpretations of statutes in a wide range of cases. For Cass Sunstein, Chevron is a “counter-Marbury for the administrative state”.

But if Chevron is an exception to Marbury, there are further exceptions nested within Chevron, matryoshka-like. The boldest and most naked such exception is that for “major questions” – the idea that in certain “extraordinary cases” judicial deference to agency interpretations of law should not be granted. This doctrine has been heavily criticized and was believed by many until recently to have been dead, or at least dormant. In a recent paper (forthcoming next month in the Connecticut Law Review) I defend the doctrine, arguing that it is a valuable shield or safety valve for Chevron itself.

The major questions doctrine was established more than a decade after Chevron in two cases, MCI v. AT&T and FDA v. Brown & Williamson. In each case, the Supreme Court denied deference to the relevant agency; in MCI the court refused to defer to the FCC in its attempt to exempt certain long-distance carriers from ratemaking procedures, and in Brown & Williamson, the court more famously refused to defer to the FDA in its attempt to regulate tobacco as a drug. Formally, both cases were decided based on the text of the relevant statute, guided by traditional interpretive tools. That is, they were decided at Chevron step one, making deference unavailable to the agencies in question.

But the textual arguments in both cases are quite weak, especially in Brown & Williamson. Recognizing this, the Court added an additional factor to the Chevron analysis – a higher bar for deference (or, conversely, a lower bar for finding clear Congressional intent in the statute) in major or extraordinary cases. What, exactly, makes a case extraordinary is never defined, though a review of the few cases invoking the doctrine suggests it is some combination of economic significance, political significance or controversy, a shift in agency position, and (possibly) short statutory text.

After Brown & Williamson in 2000, the major questions doctrine appeared to fall dormant. Some even declared it dead after the Court failed to even mention it in 2008’s Massachusetts v. EPA, though as I discuss in the paper this case can’t be taken to have killed the doctrine. 2014’s King v. Burwell confirmed that the doctrine was alive and well, however, with the Court denying deference to the IRS regarding its interpretation of tax provisions of the Affordable Care Act. Some likely near-future cases, including judicial review of EPA’s Clean Power Plan (assuming the Trump administration is not able to withdraw that rule before review) appear to be strong candidates for “extraordinary case” status. The major questions doctrine lives.

Is that a good thing? Traditionally, the answer has depended on one’s view of Chevron. If one is skeptical of judicial deference to agencies, then exceptions to that deference would likely be welcomed. Most of Chevron’s defenders (those who think it is an accurate reflection of congressional intent, is likely to lead to better regulatory outcomes, or both) have sharply criticized the major questions doctrine, noting some serious problems with it. It lacks strong theoretical justification in the cases. It suffers from fundamental line-drawing problems leading to a danger of arbitrary application (as Sunstein cheekily notes, Chevron itself could very well have qualified as an extraordinary case). Moreover, there is no reason to think that either the implied delegation or pro-technocratic justifications for Chevron would be less applicable in major cases – if anything, deference would seem more justified there.

These are valid criticisms, but as I argue in the paper I think they are secondary to an important virtue of the major questions doctrine – it protects Chevron from the violence courts might otherwise do to it in “extraordinary” cases. Judges are likely to want to decide major cases – they may have strong policy preferences, or they may simply feel (more strongly than in other cases) that deciding questions of law is their proper role. They may also feel pressure from other political actors or the public to decide a case (Burwell may be an example of this). This pressure makes it appealing to find clear congressional intent in the text of the statute and thereby decide a case at Chevron step one without deferring to the agency. But doing so carries a risk – it creates Chevron step one precedent that shifts doctrinal boundaries in the far greater number of interstitial Chevron cases.

One solution, of course, would be for judges to resist the temptation to decide major cases. But if you think they will nevertheless (or for other reasons should nevertheless) deny deference in major cases, then separating these cases into a different category, as the major questions doctrine does, limits the damage they can do to interstitial cases. To make this concrete, consider that Brown & Williamson appears to have had little effect on Chevron step one statutory interpretation. It still looks today like an ambitious reading of the Food and Drug Act, out of line with other Chevron cases. This is because (I argue), the major questions doctrine has limited its influence. If you are suing an agency and the best citation in your brief is Brown & Williamson, your chances probably are not good.

The major questions doctrine is not pretty. It is a hack and a kludge, bolted on to Chevron with almost no theoretical or practical support. It is likely to be frustrating in its practical application. But so long as it continues its past trend of being applied in only one or two “extraordinary cases” per decade, it has a valuable role to play as a shield and safety valve for Chevron. Without it, the boundaries of Chevron step one could easily shift, perhaps enough to effectively overrule it. That would have effects far beyond extraordinary cases. This means that both Chevron’s critics and its defenders can find something to like in the major questions doctrine. It therefore appears safe for now, even if the Trump administration prompts some academics to shift their previous positions on the wisdom of deference to agencies.

Nathan Richardson is an Assistant Professor at the University of South Carolina Law School and a Visiting Fellow at Resources for the Future