A Tale of Two Vote Switches

Article - Online Edition - Volume 100

I. Introduction

Supreme Court Justices publicly switching votes in a case—voting for a case outcome that is the opposite of their opinion’s reasoning—is exceedingly rare. However, over the course of two weeks in June 2021, two cases featured vote switches: Justice Thomas switched his vote in Borden v. United States, and Justices Breyer, Sotomayor, and Kagan switched their votes in United States v. Arthrex.

In this Article, I break down what happened in both cases and explain why exactly these four Justices switched their votes. I then argue that Justices Breyer, Sotomayor, and Kagan’s switch in Arthrex is much more defensible than Justice Thomas’s in Borden. In fact, even though Justice Thomas defended his switch in Borden as necessary while Justices Breyer, Sotomayor, and Kagan said nothing about theirs in Arthrex, the Arthrex switch was the necessary one, while Borden could have been resolved without Justice Thomas switching votes. The Article concludes with some thoughts about whether these two switches in a short period of time might be a sign of things to come on the new Roberts Court.

II. The Two Vote Switches

Justices changing their votes privately happens all the time. They may feel persuaded at oral argument to abandon the position they believed they supported, be convinced to switch to a colleague’s position during the Justices’ conference on the case, or sign onto an opinion during the writing process that takes a different position than they had originally voted for at conference. These vote changes happen for a variety of reasons, but they all happen behind the closed doors of the Supreme Court with the public learning about them, if they ever do, long after the fact, such as from a leak to the media or when Justices’ internal papers become public.[1]

Both of the cases analyzed here involve something different—Justices publicly changing their votes. In other words, what makes these cases unusual is that the Justices wrote (or signed onto) an opinion that would logically produce one vote on the outcome in the case, but then ultimately concluded by voting for a different outcome. This section breaks down the two June 2021 vote switch cases and explains the vote switch in each.

A. Borden

Borden v. United States[2] presented the Court with the question of whether reckless criminal behavior qualified for an enhanced sentence under the Armed Career Criminal Act (ACCA). ACCA is a federal law that increases the sentence for illegal possession of a gun if the defendant has three prior violent felony convictions. As Justice Kagan framed the legal issue in the introduction to her opinion, “The question here is whether a criminal offense can count as a ‘violent felony’ if it requires only a mens rea of recklessness—a less culpable mental state than purpose or knowledge.”[3]

Answering this question turned on the definition of “violent felony” in the statute. ACCA defines “violent felony” as any crime punishable by imprisonment for a term exceeding one year that:

(i) has as an element the use, attempted use, or threatened use of

physical force against the person of another; or

(ii) is burglary, arson, or extortion, involves use of explosives, or

otherwise involves conduct that presents a serious potential risk of physical injury to another.[4]

In 2015, the Court ruled in Johnson v. United States[5] that the “otherwise involves” clause of subsection (ii), known as the “residual clause,” was unconstitutionally vague.[6] Thus, the question presented to the Court in Borden concerned subsection (i), the “elements clause.”[7]

The Court ultimately split 5–4 on the question, holding that reckless crimes did not qualify under ACCA’s violent felony definition.[8] The five Justice majority was divided between two factions—Justice Kagan’s plurality opinion for herself and Justices Breyer, Sotomayor, and Gorsuch, and Justice Thomas’s solo concurrence in the judgment. Justice Kagan’s opinion focused exclusively on the elements clause. As she analyzed it, that clause has two parts to it:

a) “use of physical force”

b) “against the person of another”[9]

Based on precedent in the context of a different domestic violence statute that uses similar language, Justice Kagan acknowledged that the “use” clause (part a) by itself does cover reckless crimes.[10] However, when that clause is coupled with the “against the person of another” clause (part b), something that is not a part of the domestic violence statute, the provision as a whole excludes reckless crimes.[11] She explained that the “against the person of another” language “demands that the perpetrator direct his action at, or target, another individual. Reckless conduct is not aimed in that prescribed manner.”[12]

Skipping to the dissent for reasons that will become clear, Justice Kavanaugh wrote for himself, Chief Justice Roberts, and Justices Alito and Barrett. Justice Kavanaugh’s opinion argued that the “against the person of another” language does not change anything with regard to the necessary mens rea for the covered crimes and that the “use of physical force” language alone covers recklessness.[13] Moreover, even if “against the person of another” mattered, that phrase in the context of the rest of the provision includes recklessness because the ordinary meaning of the term is broad enough to include bodily harm caused by disregarding a substantial and unjustifiable risk.[14]

With a 4–4 split on whether reckless crimes were covered by ACCA’s “violent felony” clause, Justice Thomas was the deciding vote. Justice Thomas holds previously expressed idiosyncratic views on ACCA’s language, and he stuck to those views in his concurrence. He argued that he continues to believe that the domestic violence case was wrongly decided and that the language “use of physical force” by itself does not include reckless criminal behavior.[15] ACCA’s extra language modifying the “use” provision—“against the person of another”—did not change Justice Thomas’s view; thus, he concluded that the elements clause does not cover reckless criminal behavior.[16]

As the majority and dissent analyzed the case, this conclusion should end the matter because Justice Thomas would provide the fifth vote, joining Justice Kagan’s bloc, against including reckless crimes under the ACCA. However, Justice Thomas went on to analyze whether ACCA’s residual clause covered reckless crimes. The plurality and dissenting opinions ignored the residual clause because Johnson had previously found it unconstitutionally vague. Justice Thomas, though, disagreed with that decision at the time[17] and, in his Borden concurrence, once again argued against the holding.[18] Without Johnson voiding the clause, Justice Thomas argued that the residual clause would cover the reckless crime underlying Borden.[19] “I hesitate,” Justice Thomas concluded on the merits, “to give petitioner the benefit of Johnson, because his crime is a ‘violent felony’ as Congress defined the term.”[20]

Justice Thomas’s separate opinion—analyzing both the elements clause and the residual clause—would seem to give a fifth vote to the government here, as he believes that reckless crimes qualify as a “violent felony” under ACCA, just as the four dissenters do, though for a different reason (dissenters under the elements clause; Thomas under the residual clause). Nonetheless, he finished his opinion by announcing his vote switch:

I reluctantly conclude that I must accept Johnson in this case because to do otherwise would create further confusion and division about whether state laws prohibiting reckless assault satisfy the elements clause. I therefore concur in the judgment.[21]

Thus, even though the rest of his opinion indicates that he believed that the conduct here was covered by ACCA, Justice Thomas switched his vote at the end of his opinion, providing the fifth vote in favor of a judgment for Borden and against the government.

B. Arthrex

United States v. Arthrex[22] is a separation-of-powers case presenting the Court with the question of whether Administrative Patent Judges (APJs) are inferior or principal officers pursuant to the Constitution’s Appointments Clause. There are over 200 APJs that, in panels of three called Patent Trial and Appeal Boards (PTAB), hear different types of patent disputes.[23] At issue in Arthrex was a process called inter partes review, when two parties battle over whether an existing patent satisfies the novelty and nonobviousness requirements for inventions.[24]

In an inter partes review proceeding over a patent related to orthopedic surgery, the PTAB ruled against Arthrex and invalidated its patent.[25] On appeal to the Federal Circuit, Arthrex raised a constitutional challenge to the PTAB, claiming that APJs were principal officers and should have been appointed by the President (and confirmed by the Senate) rather than just by the Secretary of Commerce, an appointments process that is allowed for inferior officers.[26] The Federal Circuit agreed that APJs were principal officers, vacated the PTAB’s decision in the case, and to cure the constitutional violation, remanded for a new hearing before APJs that would be stripped of the tenure protections they previously had.[27]

On appeal, a majority of the Supreme Court agreed with Arthrex’s constitutional argument. In an opinion written by Chief Justice Roberts and joined by Justices Alito, Gorsuch, Kavanaugh, and Barrett, the Court held that APJs are principal officers because their decisions are unreviewable by a superior executive officer.[28] As principal officers, the Appointments Clause requires they be appointed by the President with advice and consent of the Senate, not appointed by the Secretary of Commerce alone.[29] The Court concluded that “the unreviewable authority wielded by APJs during inter partes review is incompatible with their appointment by the Secretary to an inferior office.”[30]

Four Justices dissented on this point. Justice Thomas wrote the principal dissent, for himself and Justices Breyer, Sotomayor, and Kagan, arguing that APJs are “at the bottom of an organizational chart, nestled under at least two levels of authority” and thus are inferior officers that can be appointed by the head of a department.[31] Accordingly, the resolution of the substantive constitutional issue was a straightforward 5–4 split among the Justices.

The vote switch occurred, not on the constitutional issue, but rather regarding the remedy. Recall the Federal Circuit vacated the APJ decision and remanded for a new hearing before APJs without tenure protection, believing that this change in the structure of the PTAB would make the APJs inferior officers correctly appointed under the Constitution. None of the Justices adopted this position. Rather, in a separate part of his opinion, the Chief Justice wrote for himself and only three others, Justices Alito, Kavanaugh, and Barrett, that the proper remedy was to remand the case to the Director of the Patent and Trademark Office to review the decision of the PTAB to determine if a rehearing was warranted.[32] This review was necessary, according to the plurality, because “[d]ecisions by [inferior officer] APJs must be subject to review by the Director.”[33] The plurality’s remedy would cure the constitutional problem because the Director is a principal officer appointed by the President who has the authority to issue binding decisions.

Justice Gorsuch, who had agreed with the Chief Justice’s opinion about constitutionality, disagreed about the remedy. He wrote separately and charged the Chief Justice’s plurality with usurping Congress’s authority by rewriting the statutory scheme and inventing a new procedure out of whole cloth.[34] Rather than doing so, Justice Gorsuch’s proposed remedy was to “decline[] to enforce the statute” in the case and “‘set[] aside’ the PTAB decision in this case.”[35] In other words, Justice Gorsuch proposed finding the entire PTAB review process unconstitutional.

That leaves the four Justices who dissented on the issue of constitutionality. Justice Thomas’s opinion flatly rejects the Chief Justice’s proposed remedy as unsupported by the Constitution or statute.[36] Although he does not say so explicitly, his conclusion that the APJs were appointed consistent with the Constitution logically means his remedy would be to reverse the Federal Circuit’s decision and remand with an order to review the PTAB decision purely on the merits of the patent dispute.

The other three Justices who joined Justice Thomas on the issue of constitutionality did not join him on the issue of remedy. After giving additional support for Justice Thomas’s conclusion that the APJs are constitutional, Justice Breyer’s separate opinion for himself and Justices Sotomayor and Kagan tacked on a concluding paragraph about the remedy:

For the reasons I have set forth above, I do not agree with the Court’s basic constitutional determination. For purposes of determining a remedy, however, I recognize that a majority of the Court has reached a contrary conclusion. On this score, I believe that any remedy should be tailored to the constitutional violation. Under the Court’s new test, the current statutory scheme is defective only because the APJ’s decisions are not reviewable by the Director alone. The Court’s remedy addresses that specific problem, and for that reason I agree with its remedial holding.[37]

Although less explicit than Justice Thomas’s vote switch in Borden, this paragraph nonetheless publicly signals a vote switch for the three Justices. The rest of Justice Breyer’s opinion (along with the part of Justice Thomas’s opinion that he and the other two Justices signed onto) argues for the constitutionality of the APJ appointment process.[38] This substantive discussion would, like Justice Thomas’s position, mean that these three Justices would vote to reverse the Federal Circuit and remand for a determination on the merits. However, in this paragraph, they switch positions and vote for an outcome consistent with the opposite conclusion—that the APJ appointment process is unconstitutional. Thus, even though they wrote that they believed the APJ appointment process was constitutional, they ultimately voted in favor of an outcome inconsistent with that reasoning: the Chief Justice’s remedy to remand to the Director for a review of the PTAB decision.

III. Assessing the Vote Switches

Public vote switches in the Supreme Court are rare, but they do happen.[39] The background assumption necessary to understand all such switches is the norm among Justices to cast their votes in a case based on outcome.[40] Thus, in conference and throughout the opinion writing process, the Justices form voting blocs based on what they believe the Court should ultimately do with the case’s judgment—for instance, affirm, reverse, remand, or some other mandate. Voting based on issues, the alternative to outcome voting, would mean that Justices would break the case down into discrete legal issues, vote on the resolution of those issues, and then determine the outcome of the case based on the separate resolution of each issue.[41] Changing to issue voting has been supported by a small number of academics but is not, for a variety of reasons, how the Justices actually vote in cases.[42]

Public vote switches occur when a Justice writes (or signs onto) an opinion resolving an issue in a manner that logically leads to a particular outcome but ultimately votes for a different (usually the opposite) outcome. The following chart represents the vote switch:

Normal voting Vote switch
Resolution of issues within an opinion Analysis logically
entails Outcome A
Analysis logically
entails Outcome A
Vote on outcome at end of opinion Outcome A Outcome B


Switching a vote to an outcome different than the analysis in the opinion contravenes the Supreme Court norms of outcome voting and of opinions that are internally consistent.[43] However, some vote switches are defensible—even necessary—while others are not. This section argues that Justices Breyer, Sotomayor, and Kagan’s vote switch in Arthrex falls in that first category, while Justice Thomas’s in Borden falls in the second.

A. Arthrex Is Defensible

Justice Breyer’s opinion announcing the vote switch for himself and Justices Sotomayor and Kagan did so in a nonobvious way. There is no outward recognition that the Justices have switched their votes. The only indication that the opinion changes its vote is hidden within these two sentences: “For purposes of determining a remedy, however, I recognize that a majority of the Court has reached a contrary conclusion. . . . The Court’s remedy addresses that specific problem, and for that reason I agree with its remedial holding.”[44] Since these sentences indicate that these three Justices are voting in favor of the Chief Justice’s outcome (and thus providing the votes to make it “the Court’s remedy”), despite the rest of their separate opinion arguing that APJs are constitutional, which would necessitate a different remedy, this is the section of Justice Breyer’s Arthrex opinion that announces the vote switch.

Even more hidden is the reason for the vote switch. That is contained in the clause “For purposes of determining a remedy.” What Justice Breyer says in this clause is that, if he does not switch his vote on the outcome, there would be no remedy in the case. To understand why, here is the breakdown of the case without a vote switch from Justices Breyer, Sotomayor, and Kagan:

Voting Bloc Merits Determination About APJ Appointment Process How to Fix PTAB Outcome Vote
Chief Justice Roberts and Justices Alito, Kavanaugh, Barrett (4)




Director review of PTAB decision so APJs are now inferior officers


Remand for Director review of PTAB decision


Justice Gorsuch (1)




Cannot be fixed by the Court


Remand to set aside PTAB decision


Justice Thomas (1)




System is constitutional as is


Reverse and remand for appellate decision on merits


Justices Breyer, Sotomayor, and Kagan (3)




System is constitutional as is


Reverse and remand for appellate decision on merits


Simplifying the decision by grouping it based on how the Justices would have voted on the outcome (again, without the Breyer bloc’s vote switch) results in this tally:

Outcome     Number of Justices
Remand for Director review     4 (Roberts, Alito, Kavanaugh, Barrett)
Remand to set aside     1 (Gorsuch)
Reverse and remand     4 (Thomas, Breyer, Sotomayor, Kagan)


The result? Three different outcomes, none with enough votes to constitute a majority.

Without a vote switch, the Court would fail to accomplish the most basic goal of hearing a case—producing a judgment in the case before it. This then is the reason Justice Breyer joined Justice Thomas’s opinion arguing for the constitutionality of APJs and wrote his own opinion further bolstering that position yet, inconsistent with that position on the merits, voted to remand the case to the Director for review of the PTAB decision. In other words, Justices Breyer, Sotomayor, and Kagan switched their votes on the outcome of the case to avoid having no case outcome supported by a majority of the Court. With this vote switch, this chart captures the new tally:

Outcome     Number of Justices
Remand for Director review     7 (Roberts, Alito, Kavanaugh, Barrett and Breyer, Sotomayor, Kagan)
Remand to set aside     1 (Gorsuch)
Reverse and remand     1 (Thomas)


With this switch—and only with this switch—the Court produced a judgment in the case before it.

Justice Breyer explained none of this in his paragraph about the remedy, but other Justices in the past have more explicitly justified their similar vote switches. For instance, Justice Stevens explained a switch as follows:

It has been our practice in a case coming to us from a lower federal court to enter a judgment commanding that court to conduct any further proceedings pursuant to a specific mandate. That prior practice has, on occasion, made it necessary for Justices to join a judgment that did not conform to their own views.[45]

Justice Rutledge went into even further detail about a similar change:

Accordingly, I would affirm the judgment.

My convictions are as I have stated them. Were it possible for me to adhere to them in my vote, and for the Court at the same time to dispose of the cause, I would act accordingly. The Court, however, is divided in opinion. If each member accords his vote to his belief, the case cannot have disposition. Stalemate should not prevail for any reason, however compelling, in a criminal cause or, if avoidable, in any other. My views concerning appropriate disposition are more nearly in accord with those stated by Mr. Justice Douglas, in which three other members of the Court concur, than they are with the views of my dissenting brethren who favor outright reversal. Accordingly, in order that disposition may be made of this case, my vote has been cast to reverse the decision of the Court of Appeals and remand the cause to the District Court for further proceedings in accordance with the disposition required by the opinion of Mr. Justice Douglas.[46]

The explanation for this type of vote switch is quite simple. In order to produce a judgment supported by a majority of the Justices, something the Court aims to do in order to fulfill its basic job of deciding the case before it, Justices occasionally must switch their votes on the outcome to provide that majority.[47] That is what Justices Breyer, Sotomayor, and Kagan did in Arthrex, and that is why their vote switch is defensible—as necessary to produce an outcome in the case before the Court.

B. Borden Is Not Defensible

Unlike Justice Breyer’s opinion in Arthrex, Justice Thomas’s opinion in Borden not only explicitly states that he is changing his vote but also gives an explanation why. He explained that “to do otherwise would create further confusion and division about whether state laws prohibiting reckless assault satisfy the elements clause.”[48] Justice Thomas supported this statement with two different rationales—one contained in a citation; the other in a footnote. As this section argues, neither the citation nor the footnote justifies Justice Thomas’s vote switch.

First, the citation. Justice Thomas justified his vote switch by citing to his concurring opinion in Vance v. Ball State University.[49] In that case, the Court held that an employer was not vicariously liable under Title VII for workplace harassment perpetrated by an employee who did not have the power to take tangible employment action (fire, hire, discipline, etc.) against the victim.[50] Justice Alito wrote the majority opinion for a group of five that included Justice Thomas. Justice Thomas’s separate concurrence was a two-sentence regular concurrence (not a concurrence in the judgment, like in Borden) that noted he continued to find the Court’s controlling cases about employer liability wrong, but that he nonetheless joined Justice Alito’s majority opinion “because it provides the narrowest and most workable rule” under that (wrongly decided) precedent.[51]

While Vance bears some resemblance to Borden, it does not do the work Justice Thomas wants it to. Both Borden and Vance feature Justice Thomas indicating that he continues to believe positions he had previously expressed in dissent but that in this particular case, he would acquiesce to the position he opposed. The similarity ends there though. In Vance, acquiescing to the opposing view did not change Justice Thomas’s vote on the outcome in the case. Under Justice Thomas’s previously stated position that would more strictly circumscribe employer liability, he would reach the same result as Justice Alito’s opinion—affirm the lower court’s rejection of employer liability. Thus, although he indicated he was agreeing with a position that he continues to believe is wrong, doing so does not change either Justice Thomas’s position on the case outcome or the majority’s position—the employee loses either way.

The difference in Borden is significant—Justice Thomas’s giving in to precedent with which he continues to disagree necessitates him changing his vote on the outcome of the case. Earlier in the opinion, he wrote that he continues to believe that Johnson is incorrect and that the residual clause is not unconstitutionally vague.[52] Accordingly, to Justice Thomas, reckless crimes are covered by the (constitutional-in-his-view) residual clause, and he would vote against the criminal defendant and for the government, affirming the decision below.[53] But at the end of his concurrence, Justice Thomas “reluctantly conclude[s] that [he] must accept Johnson in this case.”[54] Unlike in Vance, this change in position means that he must change his outcome vote. Now that he has agreed to follow Johnson, there is no residual clause within which to fit reckless crimes, and his vote changes to be in favor of the criminal defendant and against the government, a reversal of the lower court’s decision. This change in vote on the outcome of the case is not present in Vance.

Second, the footnote. Justice Thomas explained the confusion he feared not switching his vote would create:

Voting to affirm petitioner’s sentence here would lead to a 5 to 4 judgment that petitioner’s sentence is correct even though five Justices conclude that Tennessee reckless aggravated assault does not satisfy the elements clause. That kind of fractured reasoning would be difficult for lower courts to apply.[55]

Justice Thomas’s explanation in this footnote is incomplete. In order to fully understand what he is referring to here, you have to consider not only the elements clause, the only part of the ACCA’s definitional section Justice Thomas mentions in the footnote, but also the residual clause.

Recall that under the statute as written by Congress, a crime fits within ACCA’s definition of violent felony if it fits within either clause. Based on the explicit and implicit arguments in the various opinions, and before Justice Thomas’s vote switch, five Justices believed the elements clause did not include reckless crimes; eight Justices believed the residual clause did not include reckless crimes (because they operated on the premise that Johnson controlled and the clause was unconstitutionally vague); yet five Justices believed ACCA covered reckless crimes. This is a classic voting paradox that the following chart captures:

Opinion author and total number of Justices joining Does the ACCA elements clause include reckless crimes? Does the ACCA residual clause include reckless crimes? Does ACCA include reckless crimes?
Kagan (4) No (4) No (4) No (4)
Thomas (1) No (1) Yes (1) Yes (1)[56]
Kavanaugh (4) Yes (4) No (4) Yes (4)
Total No (5–4) No (8–1) Yes (5–4)


The last line of the chart captures the voting paradox. A majority of the Court thinks that the elements clause does not include reckless crimes. A different majority of the Court thinks the residual clause does not include reckless crimes (because the clause is unconstitutional). However, even though both avenues for reckless crimes being covered under the ACCA definitional provision fail to garner a majority of the Court, a different majority believes that the ACCA does cover reckless crimes.

Although Justice Thomas does not spell the problem out in this much detail in his footnote, it is this voting paradox to which he is referring. He recognizes the challenge in a Supreme Court decision that says one thing on separate issues but then the opposite on the ultimate question. Because of this problem, he says that he must change his vote. He does so here by saying he will acquiesce in Johnson and provide the fifth vote against the ACCA including reckless crimes. Note that this new chart of the opinion, with his vote switch, does not contain a voting paradox:

Opinion author and total number of Justices joining Does the ACCA elements clause include reckless crimes? Does the ACCA residual clause include reckless crimes? Does ACCA include reckless crimes?
Kagan (4) No (4) No (4) No (4)
Thomas (1) No (1) No (1) No (1)
Kavanaugh (4) Yes (4) No (4) Yes (4)
Total No (5–4) [57] No (9–0) No (5–4)


Now, with the vote switch, the bottom line is completely logical. A majority is against reckless crimes being covered by the elements clause; a unanimous Court is against being covered by the unconstitutional residual clause; thus, the ACCA does not cover reckless crimes.

Even though he did not use the word “paradox” and his explanation was somewhat incomplete, the footnote indicates that Justice Thomas switched his vote to avoid this voting paradox. However, doing so was both unnecessary and inconsistent with his voting behavior in past similar cases. As I and other scholars have written previously, voting paradoxes are a rare but unavoidable fact of life for the Supreme Court.[58] If the Court were to switch to voting based on separately identified issues rather than voting for particular outcomes in cases, the possibility of the voting paradox would be diminished.[59] However, for sound reasons, the Court continues to vote by outcome, which means voting paradoxes will always be a possibility in certain cases.

Switching votes to avoid a voting paradox is unnecessary when the legal system can function in the face of a voting paradox. Scholars have identified numerous voting paradoxes from the Supreme Court.[60] It is true that each of those cases can present a problem for future courts in interpreting precedent, but those problems are not that different than the general challenge of interpreting precedent in new cases.

Take McDonald v. City of Chicago[61] as an example.[62] The case is known for a 5–4 Court establishing that the Second Amendment right to own a handgun for self-defense is incorporated against state and local governments.[63] But, underlying that important proposition of constitutional law is the fact that McDonald is a classic voting paradox case. Although five Justices agreed that the Second Amendment is incorporated, they disagreed about how, and in doing so, rejected the others’ view as to incorporation. Thus, the case can be summarized in the following chart:[64]

Opinion author (number of Justices joining opinion) Does the Due Process Clause incorporate? Does the Privileges or Immunities Clause incorporate? Is the Second Amendment incorporated?
Alito (4) Yes (4) No (4) Yes (4)
Thomas (1) No (1) Yes (1) Yes (1)
Stevens (1) & Breyer (3) No (4) No (4) No (4)
Total No (5–4) No (8–1) Yes (5–4)


A 5–4 majority of Justices rejected Due Process incorporation while a different 8–1 majority rejected Privileges or Immunities incorporation. Nonetheless, a 5–4 majority voted in favor of incorporation of the Second Amendment.

McDonald’s voting paradox is, in form, identical to the voting paradox Justice Thomas wanted to avoid in Borden.[65] In fact, Justice Thomas could have written the exact same explanation to avoid the paradox in McDonald that he wrote in Borden:

Voting to [incorporate the Second Amendment] here would lead to a 5 to 4 judgment that [the Second Amendment is incorporated] even though five Justices conclude that [the Due Process Clause] does not [incorporate the Second Amendment]. That kind of fractured reasoning would be difficult for lower courts to apply.[66]

Yet, that predicted difficulty has not been true since McDonald has been decided. In the vast majority of cases, lower courts are able to apply the precedent of McDonald without any concern because most cases present the same issue that is covered by the McDonald holding—that the Second Amendment applies against state and local governments when citizens’ rights are at stake.[67] Where McDonald has shakier clarity—on the question of non-citizens’ rights[68]—it is true that lower courts are divided on how to apply the case, [69] but that is not uncommon with lower court interpretation of non-paradoxical Supreme Court precedent. Either the Supreme Court will step in eventually to resolve the issue or the lower courts will continue to decide cases amidst unclear Supreme Court precedent. This is standard operating procedure for the United States’ judicial system.

Had Justice Thomas not switched his vote in Borden, the lower courts would probably have had an even easier time dealing with the voting paradox than they have with McDonald. Yes, as Justice Thomas pointed out, had he not switched his vote, a Court majority would have believed that the elements clause did not cover reckless crimes while a different majority found the ACCA did. But, what the precedent is (as opposed to why the precedent is what it is) would have been crystal clear on the question presented to the Justices and the main issue that lower courts will face—that ACCA’s penalty enhancement for violent felonies does include crimes committed with a mens rea of recklessness. Even if a majority of the Justices believe the elements clause does not, just like a majority believed the Due Process Clause did not incorporate the Second Amendment, the lower courts would easily apply the general holding about reckless crimes under ACCA, just like lower courts have easily applied McDonald’s general holding that the Second Amendment is incorporated. And, unlike with the tricky issue of non-citizens under McDonald, it is hard to imagine a similarly confounding issue presenting itself under the Borden paradox. Thus, there was nothing necessary about Justice Thomas’s vote switch, as lower courts could have easily applied a contrary ruling despite the voting paradox.

Moreover, that Justice Thomas switched his vote in this voting paradox case conflicts with his own voting pattern in past voting paradox cases. In those cases, he played a central role in creating the paradox and yet did not switch his vote. McDonald v. City of Chicago demonstrates this point. In that case, it was, like in Borden, Justice Thomas’s idiosyncratic and precedent-overturning position that caused the voting paradox. In McDonald, it was his views on incorporation; in Borden, his views on constitutional vagueness and the ACCA residual clause. As discussed above, had Justice Thomas shown the same concern for lower court confusion in McDonald as he showed in Borden, he could have switched his vote to avoid the problem.[70] The same is true for other voting paradox cases he has participated in.[71]

That Justice Thomas creates and/or accepts voting paradoxes in some cases, while switching his vote in Borden because of the threat of a voting paradox, indicates that he is inconsistently switching his vote. It would be easy to claim that he is doing so when it helps him to reach results that he prefers, such as sticking with the voting paradox in McDonald because incorporating the Second Amendment is the more conservative result. However, the outcome in Borden belies this kneejerk explanation. He states as much in his conclusion: “I hesitate to give petitioner the benefit of Johnson, because his crime is a ‘violent felony’ as Congress defined the term.”[72] Ruling for the criminal defendant in Borden is not what is thought of as the more typically conservative result (as evidenced by four of the Court’s other five usual conservatives being in dissent in the case).[73] Nonetheless, that he chose to avoid the voting paradox in Borden while not doing so in several past cases presenting a similar paradox creates the impression that he is inconsistently deviating from the Court’s standard practice of outcome voting and using issue voting when it suits his preferences, whatever those are.[74]

IV. Conclusion: A Sign of Things to Come?

Public vote switches are rare among Supreme Court Justices. That the two cases analyzed here were decided within two weeks of each other raises the question whether conditions on the current Supreme Court are ripe for producing even more. It is certainly possible that these two cases decided so close to one another are just coincidences, meaning nothing other than that randomness, even with small numbers, sometimes comes in bunches.

But there are a few particular conditions on the current Court that could indicate we might see an uptick of vote switches in the future. First, vote switches, either the justified kind in Arthrex or the unnecessary kind in Borden, occur when Justices hold idiosyncratic views that they refuse to change. Justice Gorsuch’s unique views on courts’ ability to remedy problems with statutes caused the conditions resulting in the Arthrex vote switch[75]; Justice Thomas’s unique views on vagueness and the ACCA residual clause caused the conditions resulting in his Borden vote switch.

On the current Court, uncompromising idiosyncrasy is not uncommon. Justice Thomas is known for having unique views that he will not compromise. Justice Gorsuch may be following in his footsteps, though on a different set of issues. While Justice Thomas often refuses to compromise on his sense of originalism and feels less bound by precedent than the other Justices,[76] Justice Gorsuch has strong and often solo views on statutory interpretation, severability, and remedy that he appears unwilling to compromise.[77] Justice Alito also has his own peccadillos about religion, anti-discrimination law, and conservative persecution about which he repeatedly writes lengthy separate opinions.[78]

This list includes only the Court’s conservative Justices because they are the ones who are more likely to be essential parts of a Court majority on the current Roberts Court. If they (or any other Justice with an uncompromising, unique view on an issue) are needed to provide a fifth vote for a majority but stick to their unique position on the matter, the conditions for a vote switch are possible. Thus, with more uncompromising Justices who hold unique positions on issues that are needed for a conservative Court to reach its results, it is possible more vote switches will be coming.

Second, it appears, at least from their public writings, that members of the Court do not fully understand the complexities of the voting patterns giving rise to the vote switches. As discussed earlier, Justice Breyer failed to explain his vote switch, as other Justices have done in the past with clarity, and Justice Thomas’s explanation for his vote switch was not well supported or clearly explained.

Moreover, Justices Kagan and Kavanaugh engaged in a footnote battle over Justice Thomas’s vote switch in Borden, with neither indicating they fully understood what was happening. Justice Kagan said that “there is nothing particularly unusual about today’s line-up,”[79] indicating that she fails to understand the rarity and complexity of the voting paradox or vote switch. And Justice Kavanaugh seems to understand that the Borden voting pattern is rare, calling it “an unusual situation” and “anomalous.”[80] But, his explanation for this voting rarity mixes the issues in Borden with the issues in the relevant precedent and fails to clearly explain exactly what the voting pattern was in this case and how Justice Thomas’s switch played a role.[81] That these Justices’ writings failed to accurately explain or assess the situation indicates that, even if they really do understand the voting complexities here, they are unable to clearly convey what is happening and thus are likely to be unable to evaluate these situations in the future. If that is so, the possibility that they act inconsistent with long-standing Court norms around voting paradoxes and vote switches and then unnecessarily switch votes is increased.

Finally, Justice Thomas’s unnecessary vote switch in Borden could create a precedent for others to do the same. Justice Thomas’s is not the first unnecessary vote switch in Court history,[82] but that he has openly done so here with an attempted (yet analytically unsound) justification provides what appears to be a (wrongly) reasoned precedent to support future such switches. In a legal system outwardly tied to following precedent, the more unnecessary switches there are, particularly ones that Justices try to justify, the more likely there could be unnecessary switches in the future.

Obviously, this is all just (informed) speculation. We have no way of knowing whether Arthrex and Borden are going to inspire a wave of public vote switching. Only time will answer that. What we do know now though is that these two rapid-succession vote switches have in common that they are part of a rare breed of Supreme Court cases. What they do not share is that Arthrex was justifiable, whereas Borden was not.

  1. .See generally, e.g., Lee Epstein & Jack Knight, The Choices Justices Make (1998).
  2. .141 S. Ct. 1817 (2021) (plurality opinion).
  3. .Id. at 1821–22.
  4. .18 U.S.C. § 924(e)(2)(B).
  5. .135 S. Ct. 2551 (2015).
  6. .Id. at 2557, 2563.
  7. .Petition for Writ of Certiorari at ii, Borden, 141 S. Ct. 1817 (No. 19-5410).
  8. .Borden, 141 S. Ct. at 1820, 1822 (plurality opinion).
  9. .See id. at 1825.
  10. .Id. at 1824–25 (2021) (citing Voisine v. United States, 579 U.S. 686 (2016)).
  11. .Id. at 1825.
  12. .Id. Later in the opinion she used the example of a reckless driver who speeds through an intersection’s red light, hitting a pedestrian the driver did not see. “[T]he reckless driver has not directed force at another: He has not trained his car at the pedestrian understanding he will run him over. To the contrary, his fault is to pay insufficient attention to the potential application of force. Because that is so—because his conduct is not opposed to or directed at another—he does not come within the elements clause. He has not used force ‘against’ another person in the targeted way that clause requires.” Id. at 1827 (internal citation omitted).
  13. .Id. at 1839–44 (Kavanaugh, J., dissenting).
  14. .Id. at 1844–49.
  15. .Id. at 1835 (Thomas, J., concurring in judgment) (rejecting the holding of Voisine, 579 U.S. 686).
  16. .Id.
  17. .Johnson v. United States, 135 S. Ct. 2551, 2563–64 (2015) (Thomas, J., concurring in judgment).
  18. .Borden, 141 S. Ct. at 1835–36 (Thomas, J., concurring in judgment).
  19. .Id. at 1835.
  20. .Id. at 1836.
  21. .Id. at 1836–37 (citation omitted).
  22. .United States v. Arthrex, Inc., 141 S. Ct. 1970 (2021) (plurality opinion).
  23. .Id. at 1977.
  24. .Id.
  25. .Id. at 1978.
  26. .Id.
  27. .Id.
  28. .Id. at 1979–83.
  29. .Id. at 1978–79.
  30. .Id. at 1985.
  31. .Id. at 1998 (Thomas, J., dissenting).
  32. .Id. at 1987 (plurality opinion).
  33. .Id. at 1986.
  34. .See id. at 1991 (Gorsuch, J., concurring in part and dissenting in part).
  35. .Id. at 1990 (quoting Novartis AG v. Torrent Pharmaceuticals Ltd., 853 F.3d 1316, 1323–24 (Fed. Cir. 2017)).
  36. .See id. at 2006 (Thomas, J., dissenting).
  37. .Id. at 1997 (Breyer, J., concurring in judgment in part and dissenting in part).
  38. .Id. at 1994, 1997; id. at 1997–98 (Thomas, J., dissenting).
  39. .See generally John M. Rogers, “I Vote This Way Because I’m Wrong”: The Supreme Court Justice as Epimenides, 79 Ky. L.J. 439 (1991); Maxwell L. Stearns, Should Justices Ever Switch Votes?: Miller v. Albright in Social Choice Perspective, 7 Sup. Ct. Econ. Rev. 87 (1999).
  40. .Stearns, supra note 39, at 125 (“[The Supreme Court] employs outcome voting, which requires a binding and unconditional vote on judgment coupled with a preferred statement of the rationale.”).
  41. .See David Post & Steven C. Salop, Rowing Against the Tidewater: A Theory of Voting by Multijudge Panels, 80 Geo. L.J. 743, 744 (1992); David G. Post & Steven C. Salop, Issues and Outcomes, Guidance, and Indeterminacy: A Reply to Professor John Rogers and Others, 49 Vand. L. Rev. 1069, 1069 (1996).
  42. .See John M. Rogers, “Issue Voting” by Multimember Appellate Courts: A Response to Some Radical Proposals, 49 Vand. L. Rev. 997, 997–98 (1996); Maxwell L. Stearns, How Outcome Voting Promotes Principled Issue Identification: A Reply to Professor John Rogers and Others, 49 Vand. L. Rev. 1045, 1046 n.2 (1996); David S. Cohen, The Precedent-Based Voting Paradox, 90 B.U. L. Rev. 183, 222–24 (2010).
  43. .Rogers, supra note 39, at 440 (explaining that “justices almost always vote to reverse or affirm based purely on their independent analysis, regardless of the majority’s rejection of a portion of that analysis”).
  44. . United States v. Arthrex, Inc., 141 S. Ct. 1970, 1997 (2021) (Breyer, J., concurring in judgment in part and dissenting in part).
  45. .Rapanos v. United States, 547 U.S. 715, 810 (2006) (Stevens, J., dissenting).
  46. .Screws v. United States, 325 U.S. 91, 134 (1945) (Rutledge, J., concurring in result) (emphasis added); see also Hamdi v. Rumsfeld, 542 U.S. 507, 553 (2004) (Souter, J., concurring in part) (“Since this disposition does not command a majority of the Court, however, the need to give practical effect to the conclusions of eight Members of the Court rejecting the Government’s position calls for me to join with the plurality in ordering remand on terms closest to those I would impose.”).
  47. .See Ryan C. Williams, Questioning Marks: Plurality Decisions and Precedential Constraint, 69 Stan. L. Rev. 795, 850 n.264 (2017) (“[A] longstanding, informal convention of the Court demands that the Justices’ collective votes always lead to a conclusive outcome in the case—for example, an affirmance, a reversal, or remand: where an initial vote reveals a three-way split between such options, some Justices have switched their votes to produce a majority-supported result.”); Rogers, supra note 39, at 459–61 (collecting five such cases and concluding that they “represent the exception that demonstrates the rule. . . . [T]he only principle that will trump the uniform practice of Supreme Court justices’ voting consistently with their own reasoning is the fundamental rule that a court must decide the case before it.”).
  48. .Borden v. United States, 141 S. Ct. 1817, 1836 (2021) (Thomas, J., concurring in judgment).
  49. .570 U.S. 421 (2013).
  50. .Id. at 424.
  51. .Id. at 450–51 (Thomas, J., concurring). Justice Thomas dissented in both Burlington Industries, Inc. v. Ellerth, 524 U.S. 742 (1998), and Faragher v. Boca Raton, 524 U.S. 775 (1998), because he believed that “absent an adverse employment consequence, an employer cannot be held vicariously liable if a supervisor creates a hostile work environment.” Faragher, 524 U.S. at 810 (Thomas, J., dissenting).
  52. .Borden, 141 S. Ct. at 1835 (Thomas, J., concurring in judgment).
  53. .See id.
  54. .Id. at 1836.
  55. .Id. at 1836 n.3.
  56. .Again, this chart uses Justice Thomas’s views prior to his concluding vote switch.
  57. .Interestingly, embedded within the elements clause determination is a separate voting paradox. An eight-Justice majority (Kagan’s plurality and Kavanaugh’s dissent, but not Justice Thomas) believes the “use of physical force” language alone covers reckless crimes. Only four Justices (Kagan’s plurality) believe the “against the person of another” language forecloses the coverage of reckless crimes. The dissenting four Justices believe that language covers reckless crimes, and Justice Thomas does not address the issue. Thus, even though only one Justice believes “use of physical force” excludes reckless crimes and only four Justices believe “against the person of another” forecloses reckless crimes, a majority holds that the elements clause forecloses reckless crimes.
  58. .See, e.g., Cohen, supra note 42, at 231; Rogers, supra note 39, at 474.
  59. .Although not eliminated, see Cohen, supra note 42, at 222–24.
  60. .See generally id.
  61. . 561 U.S. 742 (2010).
  62. .See generally David S. Cohen, The Paradox of McDonald v. City of Chicago, 79 Geo. Wash. L. Rev. 823 (2011).
  63. McDonald, 561 U.S. at 748–49, 791.
  64. .This chart is a slightly altered version of the chart in Cohen, supra note 62, at 830 (here, combining the Stevens and Breyer rows into one).
  65. .See supra chart accompanying note 56.
  66. .Borden v. United States, 141 S. Ct. 1817, 1836 n.3 (Thomas, J., concurring in judgment) (quote from Borden altered for issues raised in McDonald).
  67. .Williams, supra note 47, at 831, 833 (“But despite these seeming theoretical difficulties, lower courts have not struggled to extract guidance from the decision.”).
  68. .See David S. Cohen, McDonald’s Paradoxical Legacy: State Restrictions of Non-Citizens’ Gun Rights, 71 Md. L. Rev. 1219, 1220 (2012).
  69. .See Justine Farris, Note, The Right of Non-Citizens to Bear Arms: Understanding “The People” of the Second Amendment, 50 Ind. L. Rev. 943, 944–45 (2017) (explaining the circuit split).
  70. .All it would have taken would have been a sentence at the end of his McDonald dissent in the same form as his sentence about Johnson in his Borden dissent—“I reluctantly conclude that I must accept [Slaughterhouse] in this case because to do otherwise would create further confusion and division about whether [the Second Amendment is incorporated].” 141 S. Ct. 1817, 1836 (2021) (quote from Borden altered for issues raised in McDonald).
  71. .See, e.g., Hein v. Freedom from Religion Found., 551 U.S. 587 (2007); Miller v. Albright, 523 U.S. 420 (1998); Planned Parenthood Se. Pa. v. Casey, 505 U.S. 833 (1992).
  72. .Borden v. United States, 141 S. Ct. 1817, 1836 (2021) (Thomas, J., concurring in judgment).
  73. .See id. at 1837.
  74. .It is possible an unstated reason supports Justice Thomas’s vote switch—some form of the “rule of lenity” that gives a criminal defendant the benefit of the doubt when there is a voting paradox. If this were the case, though, Justice Thomas has not followed such a principle in past voting paradox cases. See, e.g., Miller, 523 U.S. at 459. Moreover, immediately before announcing his vote switch, Justice Thomas indicates that he is not willing to change his views based on sympathy for a criminal defendant. Borden, 141 S. Ct. at 1836 (Thomas, J., concurring in judgment) (“I hesitate to give petitioner the benefit of Johnson, because his crime is a ‘violent felony’ as Congress defined the term.”).
  75. .As well as, possibly, his long-standing antipathy toward the PTAB, see Oil States Energy Servs., LLC v. Greene’s Energy Grp., LLC, 138 S. Ct. 1365 (2018) (Gorsuch, J., dissenting).
  76. .See, e.g., U.S. v. Lopez, 514 U.S. 549, 585 (Thomas, J., concurring).
  77. .See, e.g., Barr v. Am. Ass’n Pol. Consultants, 140 S. Ct. 2335, 2363–67 (2020) (Gorsuch, J., concurring in judgment in part and dissenting in part).
  78. .See, e.g., Fulton v. City of Philadelphia, 141 S. Ct. 1868, 1883–1926 (2021) (Alito, J., concurring in judgment) (writing a 43-page concurrence on freedom of religion); Bostock v. Clayton County, 140 S. Ct. 1731, 1754–1822 (2020) (Alito, J., dissenting) (writing a 30-page dissent plus a 38- page appendix in an anti-discrimination case).
  79. .Borden, 141 S. Ct. at 1829 n.6 (plurality opinion).
  80. .Id. at 1838 n.3 (Kavanaugh, J., dissenting).
  81. .See id.
  82. .Rogers, supra note 39, at 461–74 (arguing the vote switches in United States v. Vuitch, 402 U.S. 62 (1971), and Pennsylvania v. Union Gas Co., 491 U.S. 1 (1989), were unnecessary and unwise); Stearns, supra note 39, at 129–41 (analyzing the same two cases plus Arizona v. Fulminante, 499 U.S. 279 (1991)).