This is part one of a three-part series on Judge Gorsuch.

In the confirmation questionnaire Judge Gorsuch submitted to the Senate Judiciary Committee, he listed Gutierrez-Brizuela v. Lynch first among the list of the ten most significant cases he has decided. It could hardly be otherwise. His concurring opinion in Gutierrez-Brizuela calling for the Court to overrule Chevron attracted more notice than anything else he’s ever written. Even his majority opinion in Gutierrez-Brizuela, which addressed the far more esoteric question of whether an agency interpretation that is held to abrogate circuit precedent under Brand X applies prospectively from once it is rendered or only once the circuit in question holds its precedent has been abrogated, has excited spirited debate on this blog and elsewhere. And any number of commentators have noted Gutierrez-Brizuela’s immigration context and hailed Judge Gorsuch’s willingness to check the immigration agency and seeming empathy for illegal immigrants caught in the gears of the administrative state.

In spite of all the talk about Gutierrez-Brizuela, however, surprisingly little of substance has been written about it. De Niz Robles v. Lynch, the Gorsuch opinion that Judge Gorsuch claimed controlled the outcome in Gutierrez-Brizuela, has barely been noted anywhere, in spite of its striking holding that a “Chevron step two/Brand X” agency adjudication is never entitled to retroactive effect because such adjudications are tantamount to legislation. Hardly anything has been written about the particular “Chevron step two/Brand X adjudication” to which Judge Gorsuch was so eager to deny retroactive effect in De Niz Robles, a measure of prospective effect in Gutierrez-Brizuela, and ultimately deference in the Gutierrez-Brizuela concurrence. And very little has been written about the reasons Judge Gorsuch gave there for overruling Chevron, or the lecture he gave months before that prefigured and expanded on critical moves in that argument while drawing constitutional morals from “the story of Alfonzo De Niz Robles.”

I have taken a great deal of time with these writings, and I find them disturbing, just as much for what they say about Judge Gorsuch the legal craftsman and judge as for what they say about Judge Gorsuch the administrative lawyer. They exhibit a remarkable carelessness about the basic facts and legal background of a case, and a willingness to substitute armchair theorizing for rudimentary empirical inquiry. The opinions’ treatment of precedent is less than serious at best and at times genuinely shocking; Supreme Court precedent is (among other things) openly “tamed,” turned on its head, caricatured, and frivolously distinguished, while circuit precedent is overruled sub silentio in one opinion and pronounced overruled in the next. Doctrines Judge Gorsuch doesn’t like are pared down with no evident regard for whether what’s left after the paring is workable, coherent, or even legal. And the argument against Chevron amounts to either a naïve denial of statutory indeterminacy, a proposal to cure the problem of unconstitutional delegations to agencies (that current doctrine doesn’t recognize as a problem) by pretending the delegations don’t exist and transferring the discretion they vest in agencies to courts, or both.

This story begins with De Niz Robles’s curious treatment of fact. De Niz Robles is a case about whether to retroactively apply an agency adjudication that reasonably interpreted an ambiguous statute to persons who relied on contrary circuit precedent that was subsequently overruled under Brand X. The specific ambiguity at issue in De Niz Robles was created by two supposedly “contradictory provisions of the Immigration Code,” as the Tenth Circuit described them in its first case on the question, Padilla-Caldera v. Gonzales, or Padilla-Caldera I. One of the two provisions, 8 U.S.C. § 1255(i), provides that an alien who entered the country illegally can apply for discretionary adjustment of status to that of a lawful resident, but goes on to say that such an alien can only receive adjustment if he “is admissible to the United States for permanent residence.” The other provision, 8 U.S.C. § 1182(a)(9)(C)(i), provides that an alien who has been unlawfully present in the United States for a year, has been removed from the country, and then illegally re-enters is inadmissible unless he leaves the country for at least ten years. These completely symbiotic provisions are the “two conflicting statutes” that we heard so much about from Judge Gorsuch in his confirmation-hearing commentary on Gutierrez-Brizuela.

The straightforward upshot of these two provisions would seem to be that illegal aliens who illegally reenter the country are ineligible for adjustment of status unless they leave the country for ten years. In Padilla-Caldera I, however, the Tenth Circuit, apparently led astray by § 1255’s slightly misleading offer of application for adjustment of status to all illegal aliens, found §§ 1182 and 1255 contradictory, and with no authoritative pronouncement on the question from the responsible agency, the Board of Immigration Appeals (BIA), was free to resolve the contradiction. Relying on a savings clause in § 1182 providing that the classes of aliens it made inadmissible were so except as otherwise provided by immigration law, and bits of legislative history that vaguely suggested § 1255 applied to all manner of illegal entrants, it concluded that Congress intended illegal reentrants to be eligible for adjustment of status and remanded to the agency for further proceedings on Padilla-Caldera’s application for adjustment.

Shortly thereafter, Alfonzo De Niz Robles, an illegal reentrant residing in the Tenth Circuit, applied for adjustment of status in reliance on Padilla-Caldera I. Not long after, however, in a decision called In re Briones, the BIA resolved § 1255’s application to illegal reentrants, and for perhaps obvious reasons that I’ll discuss below, held they were ineligible for adjustment of status unless they left the country for ten years. Meanwhile, Padilla-Caldera’s application for adjustment of status was still pending. The government successfully argued to the BIA that Briones superseded the Tenth Circuit’s decision in Padilla-Caldera’s own case, under the Brand X doctrine whereby agency interpretations of statutory ambiguities receive Chevron deference notwithstanding prior contrary circuit precedent. Upon Padilla-Caldera’s petition for review in Padilla-Caldera II, the Tenth Circuit agreed, holding that Briones reasonably interpreted the interplay between §§ 1182 and 1255, that it abrogated the circuit’s precedent in Padilla-Caldera I, and that it was entitled to effect even in Padilla-Caldera’s own case notwithstanding the court’s mandate in Padilla-Caldera I.

Padilla-Caldera II seemed to hold that Briones would apply retroactively in the Tenth Circuit to reentrants who filed applications for adjustment before its decision, though Padilla-Caldera II had no occasion to address whether Briones would apply retroactively to illegal reentrants like De Niz Robles who applied for adjustment in reliance on Padilla-Caldera I. But under the Retail, Wholesale test for whether agency adjudications can apply retroactively—a D.C. Circuit refinement of the Supreme Court’s decision in Chenery II that most every circuit has followed, including the Tenth—it could easily be argued that Briones should not apply to reentrants in De Niz Robles’ situation. Among other factors, Retail, Wholesale considers whether a rule is an “abrupt departure” from settled law, whether a party relied on an old rule that preceded the new one, and the burden that retroactive application would impose. All these factors weigh heavily in favor of the class of reentrants in De Niz Robles’ position who applied for adjustment of status after Padilla-Caldera I and before Briones. Besides his reliance on a decision that Briones abrogated, applying Briones retroactively to De Niz Robles would have forced him to leave the country for ten years before reapplying for adjustment.

Judge Gorsuch could have written a simple opinion along these lines in De Niz Robles. In fact, the Ninth Circuit had already done just that in Acosta-Olivarria v. Lynch, a decision that rotely applied Retail, Wholesale and held Briones didn’t apply retroactively to applicants for adjustment who relied on the Ninth Circuit’s equivalent of Padilla-Caldera I, while distinguishing a prior decision that held Briones did apply retroactively to aliens who applied for adjustment before the Ninth Circuit had addressed the issue and hadn’t relied on circuit precedent. Judge Gorsuch, however, chose to write a much more ambitious opinion, one that would hold that agency adjudications that abrogate circuit precedent under Brand X are never entitled to retroactive application regardless of reliance interests, and suggest that the much larger set of agency adjudications that interpret ambiguities at Chevron Step Two are likely ineligible for retroactive application as well.

The argument goes like this. When Congress enacts law, that law only applies prospectively absent clear statement otherwise (subject to constitutional limits). In Bowen v. Georgetown Hospital, the Supreme Court, analogizing to the presumptive prospectivity of legislation, held that agencies’ legislative rules presumptively apply prospectively unless Congress expressly gives agencies the power to make retroactive legislative rules. And by further analogy, Judge Gorsuch reasoned, a legislative agency adjudication, or as he puts it, an adjudication where an agency “acts like a legislator,” should apply purely prospectively too.

Now, as readers of this blog will know, legislative rulemaking is an administrative law term of art; indeed, agencies openly distinguish between their interpretive rules and their legislative rules, the distinction being engrained in the Administrative Procedure Act. But agencies are not in the habit of describing themselves as engaged in “legislative adjudication”; in fact, the Supreme Court has held the APA forbids “making rules in the course of adjudicatory proceedings.” Unsurprisingly then, in justifying its retroactive application of Briones to De Niz Robles’s case, the BIA explained that its decision in Briones did not “change the law, but rather explain[ed] what the law had always meant.” So one might fairly ask how Judge Gorsuch’s rule is supposed to work. How is a court to tell when an agency adjudicator that claims to be interpreting the law, as agency adjudicators often will, is really acting like a legislator?

Here Judge Gorsuch had a very clever insight. Chevron and Brand X say, or so he claims, that ambiguous statutes delegate power to agencies to make policy choices on what the law should be, and that courts should defer to agency interpretations of ambiguous statutes because they are delegated policy choices. So, if an agency adjudicator finds a statute ambiguous and proceeds to interpret the ambiguity, what it must be doing is “avowedly and self-consciously . . . exploit[ing] the law’s ambiguity . . . to write a new rule of general applicability according to its vision of the law as it should be.” Hence, Judge Gorsuch concludes, any “new agency rule announced in a Chevron step two/Brand X adjudication” is legislative and cannot apply retroactively.

As to Briones’s status under this rule, that followed easily. Since Briones received Chevron/Brand X deference in Padilla-Caldera II, the BIA’s claim that Briones just explained what the law always meant had to be wrong. Instead, Briones must have been a “quasi-legislative policy choice.” And as a policy choice, it could only apply prospectively. Finally, in a section of his opinion filled with complaints about the imprecision and incommensurability of the balancing tests in Chenery II and Retail, Wholesale that traditionally controlled agency adjudications’ retroactive effect, Judge Gorsuch found that Briones didn’t apply retroactively under them either.

The key move in Judge Gorsuch’s argument is assuming that agency adjudicators interpreting ambiguous statutes are invariably making policy choices because Chevron and Brand X say or seem to assume they are. Is that right? If it is, it isn’t because of Chevron or Brand X. Chevron and Brand X are cases about judicial review. Neither purported to tell agencies how to interpret statutes; rather, they made claims or assumptions about how agencies do interpret statutes in support of their arguments for deference. Chevron just doesn’t dictate that agencies encountering ambiguities in their organic statutes make policy choices about how to resolve them. Nor did Chevron even begin to purport to make an empirical claim about agency interpretive practice. Its assumption that agency interpretations of ambiguities amount to deference-worthy policy choices was just that—an assumption. So inferences from what Chevron says about why courts ought to defer to agency interpretations of ambiguous statutes to what agencies really do with ambiguous statutes are quite wobbly.

Now, there are some counterarguments available to Judge Gorsuch. He might argue that even if Chevron didn’t tell agencies they must make policy choices upon finding ambiguity, Chevron freed agencies to do so and that agencies must take advantage of that freedom. After all, why, if agencies can get deference to doing whatever they reasonably want to do, would they bother to figure out what an ambiguous statute “really” means? Alternatively, he might argue that if a court defers to an agency under Chevron or Brand X, it must have found that the agency was making a policy choice, such that at least any rule that receives Chevron deference must be one.

How agencies actually interpret ambiguous statutes or when courts defer to them are empirical questions, so we should have some examples. Of course, the only example that matters for the correctness of De Niz Robles’s specific holding is Briones itself, the putative subject of Judge Gorsuch’s essay on retroactivity. What we find if we look at Briones is that Judge Gorsuch’s claims about the invariably legislative nature of so-called “Chevron step two/Brand X adjudications” are completely undermined by the facts of the case actually before him.

Besides opining on the qualities of “Chevron step two/Brand X adjudications” in general, Judge Gorsuch has said quite a lot, both judicially and extra-judicially, about how the BIA decided Briones in particular, and how the Tenth Circuit previously described that reasoning in Padilla-Caldera II. In De Niz Robles, he claimed that Padilla-Caldera II said the ambiguity in §§ 1182 and 1255 “entitled the BIA to make a reasonable quasi-legislative policy choice . . . and that Briones qualified as just such a judgment.” Despite the BIA’s claim that Briones merely “explained what the law has always meant,” Judge Gorsuch would go on to say that Briones’ adjudicative “form” couldn’t “obscure the fact” that the BIA was really acting “like a legislative actor making new policy.” In Gutierrez-Brizuela, Judge Gorsuch would again claim that in Briones, “the BIA offered its view that—as a matter of policy discretion—the statutory tension” between §§ 1182 and 1255 should be resolved in favor of barring illegal reentrants from adjustment of status. And in a published lecture he gave after De Niz Robles, Judge Gorsuch would claim with some asperity, again citing and appearing to quote Padilla-Caldera II’s account of Briones, that in Briones “the BIA argued . . . that it enjoyed the right to exercise its own ‘delegated legislative judgment,’ [and] that as a matter of policy it preferred a different approach” than the holding of Padilla-Caldera I.

Who was right about Briones—the BIA or Judge Gorsuch? And if Judge Gorsuch was wrong, can his reading of Briones be attributed to Padilla-Caldera II, as he repeatedly has? The answer is that the BIA was right and that Padilla-Caldera II emphatically supports the BIA’s reading of Briones. Indeed, it is simply impossible for any lawyer, whatever his fluency in immigration or administrative law, to read Briones or Padilla-Caldera II and conclude the BIA had justified its interpretation as a policy choice, much less a “delegated legislative judgment.”

Briones turns out to be a surprisingly textualist exegesis of some fairly complex immigration law, written by a career DOJ attorney who has been a BIA member for sixteen years. Far from “avowedly and self-consciously” using its power under Chevron to “exercise delegated legislative judgment” and “overthrow judicial interpretations to effect its new vision,” it only cited Chevron’s Step One, as a “see also” for the proposition that “Congress says in a statute what it means and means in a statute what it says,” and mentioned Brand X only by way of saying it would not decide whether to apply its interpretation in circuits that interpreted § 1255(i) differently.

To show how purely interpretive Briones was in face of some fairly definitive statements from Judge Gorsuch to the contrary, I’ll have to go into some detail. Briones’s essential argument goes as follows. When § 1255(i), the adjustment-of-status provision, was enacted, it unambiguously and unproblematically required applicants for adjustment to be admissible for permanent residence. Things became complicated, however, when after § 1255(i) was enacted, first-time illegal entry became a ground of inadmissibility, not just deportability. Because § 1255(i)’s remedy is particularly available to unadmitted aliens by its terms, making first-time illegal entry a ground of inadmissibility rendered § 1255(i) self-contradictory and largely superfluous.

Addressing this contradiction, the BIA concluded that literally applying § 1255(i)’s admissibility requirement to illegal entrants would produce absurd results that Congress clearly didn’t intend. As to illegal reentrants, however, the BIA saw no contradiction or superfluity. It didn’t follow from onetime illegal entrants’ eligibility for adjustment that illegal reentrants were eligible for adjustment, nor would barring them from adjustment make anything in § 1255(i) superfluous. As for the savings clause in § 1182, which Padilla-Caldera I relied on and which says that the classes of aliens § 1182 makes inadmissible are so “except as otherwise provided,” the BIA pointed out that elsewhere Congress had invoked the savings clause by either expressly negating, or providing for discretionary waiver of, the reentry inadmissibility bar as it applied to particular classes of aliens. Ultimately, the BIA declined to “do violence to the plain language of [§ 1255(i)’s] admissibility requirement . . . by simply declaring it inapplicable” to illegal reentrants, and concluded that “implementing [it] in accordance with its plain language . . . seems perfectly consonant with the language, structure, and purpose of the Act, taken as a whole.” In short, the BIA could hardly have written a more purely interpretive decision.

What about Padilla-Caldera II, which Judge Gorsuch has repeatedly claimed to have described (and even appeared to quote as describing) Briones as a “delegated legislative judgment”? Here too, it would be impossible to actually read Padilla-Caldera II and find support for the claim that Briones was a policy choice. Padilla-Caldera II’s discussion of Briones takes up five pages of the Federal Reporter, and in those five pages the Tenth Circuit said exactly what I’ve said about Briones in considerably greater detail. The court, for example, managed to quote all three of Briones’ descriptions of §§ 1182 and 1255 as “clear” in various respects, as well as its conclusion that its result was faithful to “the language, structure, and purposes of the Act, taken as a whole.” Not only is Briones never described as a “legislative judgment” or policy choice, not only is any sense of Briones as a policy choice missing from the court’s extensive description of Briones, it’s even absent from the court’s account of Chevron and Brand X, which simply says that courts must defer to and overrule their precedent in favor of reasonable agency interpretations of ambiguous statutes­—not that they defer to agency interpretations like Briones because they represent delegated policy choices.

As it’s impossible to read Briones or Padilla-Caldera II and believe what Judge Gorsuch has repeatedly said about them, it’s impossible for me to believe that Judge Gorsuch has read them—unless he believes that Chevron doctrine somehow can transform a plainly interpretive agency decision into a policy choice as a matter of law, which would make sense of Gorsuch’s claims, but be so eccentric a view as to raise serious concerns of its own. Rather, I can only conclude that Judge Gorsuch was misled by Chevron’s picture of agency statutory interpretation into assuming that when the BIA claimed Briones just explained what the law meant, it misunderstood its own decision—and that he then built a sweeping rule of retroactivity on the back of that assumption without seeing if his rule made sense on the facts of the case before him.

It speaks rather poorly for Judge Gorsuch’s deliberative process that he would theorize his way to rejecting an agency’s characterization of its own decision without checking to see if the agency was right, and then base two of the most important decisions of his career on his fact-free guess. But the truth about Briones also doesn’t speak well for his anti-retroactivity rule more generally. De Niz Robles and Gutierrez-Brizuela are based on a premise that agency adjudications on the meaning of ambiguous statutes are all policy choice, all the time. But are they even policy choices a lot of the time? In one of the few moments in De Niz Robles that isn’t completely disconnected from how agency adjudications work, Judge Gorsuch acknowledges that the BIA decided Briones “in a quasi-judicial proceeding with lawyers and administrative law judges and briefs and arguments and many of the other usual trappings of a judicial proceeding.” These aren’t just trappings, though. By regulation, BIA members are lawyers by trade (often immigration specialists), not bureaucrats. The briefs they read and arguments they hear are by and large arguments about the meaning of immigration law, not invitations to make pro- or anti- immigrant policy. When a bunch of lawyers get together to decide disputes and listen to legal arguments from other lawyers on how they should do it, they tend to give legal reasons for their decisions.

Unsurprisingly then, if you read the BIA’s opinions, even ones on immigration statutes sufficiently ambiguous to fracture the Supreme Court, what you find over and over is purely interpretive reasoning that barely even verges on purposivism. Like courts, the BIA occasionally engages in openly purposive choice if a provision is terribly vague; the BIA’s decision (which was the subject of a recent Supreme Court argument) that statutory rape of a late adolescent should be deemed “sexual abuse of a minor” if the perpetrator is meaningfully older than the victim is such a case. But such cases are rare. All this is quite relevant. The BIA is easily the most judicially reviewed agency adjudicator in federal court (a Westlaw search indicates the Courts of Appeals have decided 1,915 petitions from BIA decisions in the last twelve months, as compared to the 65 or so NLRB cases the Courts of Appeals decide a year); today, any doctrine on agency adjudication is overwhelmingly a doctrine about the BIA. Moreover, the BIA is hardly unique among agency adjudicators in its interpretive tendencies. The NLRB, another agency adjudicative body comprised of specialist lawyers, is said to “approach[] the task of statutory construction like a court” and often “rely on statutory interpretation tools alone to defend its reformulations” of labor law.

Judge Gorsuch might well celebrate this if he knew about it, though I’m not so sure that even these good tidings would allay his concerns about agency interpretation; a lot of administrative-law scholars would greet these findings less sanguinely and propose to deny the BIA and NLRB Chevron deference in favor of a weaker form of deference like Skidmore that turns on persuasive exegesis and agency expertise. That day hasn’t come though, and perhaps the best defense of Judge Gorsuch’s anti-retroactivity rule is that we do routinely defer to agency interpretations as if they were policy choices, not just persuasive interpretations of Congress’s. So long as agencies get the benefit of that fictional presupposition, shouldn’t they have to pay the costs?

It’s undeniable that agency adjudicators benefit from receiving Chevron instead of Skidmore deference, with win rates in the courts of appeals over twenty points higher when they’re reviewed under the former instead of the latter. But it’s not so clear that the reason they’re getting Chevron deference is an assumption that their interpretations of statutory ambiguities are unfettered policy choices. If it were, why would courts continue to defer to agency adjudicators even when nothing in their reasoning, which courts review at Chevron step two, suggests they’re really exercising raw policy discretion? Indeed, the judges who claim Chevron is motivated by deference to raw agency policy choice tend to either be Chevron’s critics, like Judge Gorsuch or Justice Thomas, or thoroughgoing pragmatists like Judge Posner. The Court and Chevron’s mainstream proponents, on the other hand, usually claim that Chevron is rooted in a presumption of delegation, without much further specification on how agencies are presumed to use it. And a growing body of scholarship argues that Chevron doesn’t liberate agencies from pursuing the best interpretation of the ambiguous statutes they administer, albeit in a purposivist way that draws on their superior understanding of what rules would best further Congress’s goals.

Even if agency adjudicators really were reaping the benefits of a presupposition that their interpretations of ambiguous statutes consist of raw policy choice, count me skeptical that denying agency adjudications retroactive effect whenever they interpret ambiguous statutes is an acceptable price to pay for a thoroughly fictitious presupposition. Taken to its logical conclusions, denying every agency adjudication that “reach[es] Chevron step two” (a formulation that obscures that agencies don’t apply Chevron to their own interpretations) retroactive effect would be tantamount to banning agency adjudications of ambiguous statutory questions altogether. If agency adjudicators couldn’t apply their decisions to the parties before them, why would they adjudicate anything, or be allowed to?

As I’ll address in a later post, Judge Gorsuch is rather obscure on whether his rule applies to all Chevron step two adjudications or just the ones that abrogate judicial precedent under Brand X, though his argument from Chevron and Bowen doesn’t permit any distinction between the two. But even if his rule only applies to Brand X, it’s hard to see why agencies must pay for Brand X with non-retroactivity. In a world where agencies are treated as superior interpreters of their organic statutes than courts, there’s nothing in particular about agency decisions that abrogate judicial precedent that warrants prospectivity—except for reliance, which isn’t a condition on Judge Gorsuch’s rule.

Ultimately, then, I don’t think Judge Gorsuch’s rule can be defended in terms of legal fiction, anymore than it can be defended in terms of how agency adjudicators actually interpret statutesse. It only makes sense if we share the strikingly dark, if not paranoid view of agency adjudicators that pervades the less doctrinally interesting but rhetorically telling second half of the opinion—as “decisionmaker[s] driven by partisan politics,” who announce retroactive rules in “full view of [who] the winners and losers” will be, and may use that foresight to “single out disfavored persons and groups and punish them for past conduct they cannot now alter.” This vision would warrant prospectivity. But whether it actually describes the principal agency adjudicator in the federal system—a mixed body of immigration lawyers, serving in some instances since the Clinton administration, that aren’t, by any conceivable stretch of the imagination, denying relief to illegal reentrants or other classes of immigrants because of a desire to target foreseen “losers”—hardly seems to have crossed Judge Gorsuch’s mind.

Asher Steinberg writes The Narrowest Grounds, a law blog. He can be reached at asher.steinberg@outlook.com.