Procedure and Remedies

The grounds for disqualification taught above say when a judge must step aside, but they leave a good deal unanswered. Who is expected to notice a ground? When must a party speak up? Who rules on a challenge, and how does the loser test that ruling? And what does a court do about a case in which a judge sat who should have stood down? The first three questions concern how disqualification is raised, decided, and reviewed. The last concerns relief after the fact.

Both sets of questions require a balance that the grounds themselves never had to strike. A litigant is entitled to an impartial judge, but a disqualification motion can also be used as a tactic, and the D.C. Circuit has warned that in the wrong hands it becomes a “procedural weapon to harass opponents and delay proceedings.”¹ On the remedy side, undoing a judge’s work burdens a prevailing party who did nothing wrong, while refusing all relief would turn a mandatory duty into advice. Procedural law tends to strike such balances by leaving details to “the informed discretion of the judge, acting in the circumstances of the particular case, rather than regulating those details by rigid provisions.”² Discretion does much of the work here too, as the two subsections below show.

Motions and Review

The duty to disqualify belongs to the judge. Rule 2.11(A) says that a judge “shall disqualify himself or herself” in any proceeding in which the judge’s impartiality might reasonably be questioned, and it makes that duty depend on nothing a party must do first.³ The codes of conduct say so directly: a judge must withdraw when the standards for disqualification are met “regardless of whether a motion to disqualify is filed.”⁴ Grounds that turn on the judge’s knowledge work the same way, because the judge’s own awareness sets the duty in motion. The economic-interest ground, for example, applies once the judge knows that a covered family member has an economic interest in the subject matter or in a party.⁵

A California disciplinary case shows what follows when a judge ignores the duty. The judge there had discussed Frank’s arrest outside court with the Cerrato brothers, with whom he had a long-term friendship. The tribunal held that he should have promptly disqualified himself under the canon requiring disqualification when a judge’s impartiality might reasonably be questioned, and it concluded that his handling of the case amounted to wilful misconduct.⁶

A motion is therefore a backstop for the judge’s own duty and does not start it. The reason lies in who holds the information. A litigant can raise only what the litigant can discover, and those with the greatest access to information relevant to disqualification are often the judges themselves.⁷

That gap in access explains the practice of disclosure. Model Code commentary says that the judge “should disclose on the record information that the judge believes the parties or their lawyers might reasonably consider relevant to a possible motion for disqualification, even if the judge believes there is no basis for disqualification.”⁸ Disclosure turns what only the judge knows into something the parties can act on, either by moving to disqualify or, when the ground is waivable, by agreeing under the waiver procedure described above that the judge may stay. Silence has a price, as In re Al-Nashiri shows. For more than two years after applying for a job with the Executive Office for Immigration Review, the military judge presiding over the commission never disclosed his efforts to secure that employment, and less than twenty-four hours after receiving his start date he indefinitely abated the proceedings. The D.C. Circuit observed that, given this lack of candor, a reasonable observer might wonder whether the judge had done something worth concealing.⁹

Once a party has the facts, the burden to act quickly shifts to the party. A motion to disqualify must be filed promptly, and courts look askance at tactical delays in which the request becomes “a fallback position in the event of adverse rulings on pending matters.” They require that motions be filed expeditiously after the movant obtains knowledge of the facts that show the ground.¹⁰ The concern is that a litigant who could wait would get a free preview of the judge’s rulings before deciding whether the judge’s connections matter, and the right to an impartial judge would turn into an option exercised only when the outcome looked bad. The D.C. Circuit made a related point in United States v. Microsoft Corp.: disqualification is never taken lightly, and in the wrong hands a motion is a procedural weapon to harass opponents and delay proceedings.¹¹

The clock runs from the movant’s knowledge, so promptness does not penalize a party for failing to raise what it could not see. Microsoft supplies the contrast. The full extent of the circumstances disqualifying the district judge came to light only while the appeal was pending.¹² The D.C. Circuit was concerned about granting a disqualification request when the supporting material had not been admitted in evidence, yet it decided Microsoft’s request notwithstanding the state of the record and departed from its usual practice of declining to address issues raised for the first time on appeal.¹³ A party that sat on knowledge it had and a party that was kept from knowledge it needed stand in different positions, and only the first has committed the delay the rule disfavors.

A filed motion must be decided by someone, and jurisdictions have built different designs. A blended approach, adopted by a significant number of states, has the subject judge review the motion first for legal sufficiency, asking whether the allegations, if true, would require disqualification, and then assigns another judge to decide the merits, meaning whether the allegations are true.¹⁴ It sits between two simpler designs: leaving the whole motion with the challenged judge, or sending the whole motion to someone else. Each answers a real difficulty from a different angle. Asking the challenged judge to rule is efficient, but it has that judge decide a question about his own impartiality. Sending every motion to another judge removes the self-evaluation at the price of reassignment, even for groundless motions. The blended approach lets the subject judge dispose of a motion that fails on the law alone, because the allegations are assumed to be true, while keeping that judge away from contested facts that would put his own credibility in issue.

Where the challenged judge does rule, the cases describe a two-step inquiry. A Massachusetts appellate court explained that when the judge’s capacity to rule fairly is in question, he must consult first his own emotions and conscience, and that even a judge who concludes that he can act impartially must then make an objective appraisal of whether this is a proceeding in which his impartiality might reasonably be questioned.¹⁵ The second step means that the judge’s sincere confidence does not end the matter, which matches the objective standard that opened this chapter. In the Ninth Circuit, the decision on such a motion is left in the first instance to the trial court’s discretion.¹⁶

Around twenty states seek to reduce the need for judges to evaluate their own impartiality, or that of their colleagues, through a peremptory challenge, also called substitution of judges. Each party gets a one-time opportunity to request that its assigned judge be replaced without showing, and in some jurisdictions without even alleging, that the judge’s impartiality is in doubt.¹⁷ The device works because it asks no one to decide anything: the judge is spared ruling on himself or herself, and the party is spared proving a ground it may only suspect. The same feature limits it. The opportunity comes once, and substitution says nothing about whether a ground exists, so a judge who has a real ground under Rule 2.11 must still withdraw whether or not a party has used its substitution.

On appeal, the standard of review for a denial is most commonly abuse of discretion.¹⁸ The Ninth Circuit frames that standard as a question: not whether the appellate court would have decided as the trial court did, but whether the decision represents “a clear error of judgment.”¹⁹ The Massachusetts court likewise required a showing of abuse of discretion before a refusal to recuse could be reversed, and it affirmed even though it disapproved of some statements the judge had made during another trial before him.²⁰ One state supreme court uses a manifest error standard for a judge’s refusal to recuse himself.²¹ Deference follows the premise noted at the start of this unit: the trial judge sits closest to the facts, and procedural law generally trusts informed discretion over rigid rules.²²

Appeal is not the only route to review. A party may also seek a writ of mandamus, which requires three showings: that the right to the writ is clear and indisputable, that the party has no other adequate means of obtaining the relief, and that the issuing court, in its discretion, is satisfied that the writ is appropriate under the circumstances.²³ In Al-Nashiri the D.C. Circuit granted the writ while the case was still at the pretrial stage. It reasoned that the public’s interest in efficient justice is no greater than its interest in impartial justice, and that the costs of granting the writ were not intolerably high, especially when weighed against the burdens that both the defendant and the public would bear if the military commission proceeded under a cloud of illegitimacy.²⁴

At the top of the system, the judge whose recusal is sought decides the motion himself, because no higher court sits above a Supreme Court Justice. In Cheney v. United States District Court, a party moved for Justice Scalia to recuse himself because of his friendship with the named defendant, the Vice President of the United States. Justice Scalia denied the motion and issued a memorandum explaining why. He framed the question as the same objective test applied to himself: whether someone who thought he could decide the case impartially despite the friendship would reasonably believe that he could not decide it impartially because he had gone hunting with that friend and accepted an invitation to fly there with him on a Government plane.²⁵ His memorandum concluded that if it is reasonable to think a Justice can be bought so cheap, the Nation is in deeper trouble than he had imagined.²⁶ Where no reviewing court exists, the reasoned memorandum is the way the judge answers to the public for his application of the standard.

Exam Analysis Steps

Because a disqualification problem can fail at the procedural stage even when the ground is strong, it is best run in the order the events occur. The steps are these:

Ask whether a ground exists that the judge knows of, and if so treat the duty to withdraw as already operating without any motion, remembering that a waivable ground can be released only through the disclosure procedure taught above.
Ask whether the judge disclosed on the record what the parties might reasonably consider relevant, since silence about a connection can deepen the appearance of partiality.
Ask when the movant learned the facts and when the motion was filed, and treat delay that held the motion back as a fallback after an adverse ruling as a reason to deny it, unless the movant could not have known the facts earlier.
Identify which approach the jurisdiction uses to decide the motion (the subject judge, a different judge, or a sufficiency screen followed by another judge) and whether it also offers a one-time peremptory substitution.
If the motion is denied, identify the available review, appeal or a writ such as mandamus, and apply the deferential standard of review, which is most commonly abuse of discretion.

The third step is where fact patterns usually test promptness, since the clock runs from the movant’s knowledge and not from the date of assignment. The fifth step supplies the lens for any appeal, because the reviewing court asks whether the ruling reflects a clear error of judgment and does not ask whether it would have decided the motion the same way.

Remedies for Violations

The grounds above say when a judge must step aside, and they do not say what a court should do after a judge sat anyway. The federal appearance statute, 28 U.S.C. § 455(a), makes disqualification mandatory for conduct that calls a judge’s impartiality into question but does not prescribe how far disqualification reaches.²⁸ As the Supreme Court put it in Liljeberg v. Health Services Acquisition Corp., the statute “neither prescribes nor prohibits any particular remedy for a violation of that duty,” and Congress “wisely delegated to the judiciary the task of fashioning the remedies that will best serve the purpose of the legislation.”²⁹ Rule 2.11(A) likewise states a command to the judge and prescribes no remedy.³⁰ Remedy is therefore a matter of judicial craft, and the Court added that the court of appeals is often better placed than the Supreme Court to judge how much a particular violation matters.³¹

The Court’s guidance points in two directions. Busy judges can inadvertently overlook a disqualifying circumstance, so the law leaves room for harmless error and does not demand a draconian remedy for every violation of § 455(a). Yet it would be equally wrong to forbid any relief when the offending judge was merely forgetful.³² Between those poles lies a spectrum of relief, and its shape follows from who bears the cost. Undoing a judgment burdens a winning party who took no part in the violation, while refusing every remedy would reduce the duty to disqualify to advice. Each step up the scale protects the movant and the public more, at the price of more disruption. Relief within the case is also not the only consequence, since a judge who ignores a clear duty may face discipline, as the California finding of wilful misconduct above shows.³³

Courts decide how far up the scale to climb by weighing three risks: the risk of injustice to the parties in the particular case, the risk that denying relief will produce injustice in other cases, and the risk of undermining the public’s confidence in the judicial process.³⁴ The test comes from Liljeberg, where the judgment had already become final and the movant sought to vacate it under Rule 60(b). The D.C. Circuit applied the same test in United States v. Microsoft Corp. because there, too, the full extent of the disqualifying circumstances came to light only while the appeal was pending.³⁵ Each risk looks to a different group. The first protects the litigants before the court, including the winner who took no part in the violation. The second protects future litigants, since a rule of no consequences would invite judges to stay quiet about conflicts. The third protects the public, whose confidence in judicial work is the reason the appearance standard exists.

At the bottom of the spectrum is prospective disqualification. At a minimum, § 455(a) requires that the offending judge hear no further proceedings in the case.³⁶ That step ends the continuing problem without undoing completed work, so it costs the parties nothing they have already earned. If the ground surfaces before the judge has made any substantive ruling, nothing has been tainted and the floor is also the ceiling. The harder question, what to do about rulings already made, arises only when the judge has acted.

Further up the scale, relief can be aimed at the part of the case that the problem touched. In Microsoft the district judge’s conduct called his impartiality into question, but Microsoft neither alleged nor demonstrated actual bias or prejudice, and the D.C. Circuit saw no reason to presume that everything the judge did was suspect. After reviewing the entire record, it discerned no basis to suppose that actual bias infected his factual findings.³⁷ The court also observed that the most serious misconduct occurred near or during the remedial stage, and it concluded that its remedy should therefore focus on that stage of the case.³⁸ The scope of relief tracked the location of the taint.

At the far end of the spectrum is vacating the judgment. In Liljeberg the court of appeals recognized that caution is required before vacating a judgment that has become final, but it found the appearance of partiality convincingly established and the motion to vacate filed as promptly as possible, and it concluded that the appropriate remedy was to vacate the declaratory-relief judgment.³⁹ The Supreme Court agreed that its determination was well supported.⁴⁰ Relief of that kind would not produce injustice in other cases. To the contrary, a court’s willingness to enforce § 455 may prevent injustice in a later case by encouraging judges and litigants to examine possible grounds for disqualification more carefully and to disclose them promptly when discovered. Vacatur is therefore appropriate unless the movant did not make a timely request or depriving the prevailing party of its judgment would otherwise be unfair.⁴¹ Liljeberg and Microsoft differ in the reach of the problem and in what the record showed. Liljeberg combined a convincingly established appearance of partiality with a movant who acted as promptly as possible, while Microsoft involved a movant who showed no actual bias in the findings and a problem concentrated at a late stage.

The work product of a judge who should have been recused can also be set aside after the judge is gone. In Al-Nashiri the D.C. Circuit explained that when the disqualified adjudicator has left but his orders remain, the orders fall with him: if a judge should have been recused, then any work produced by that judge “must also be ‘recused’—that is, suppressed.”⁴² The court weighed efficiency against fairness and found efficiency the lighter side. The public’s interest in efficient justice is no greater than its interest in impartial justice, it said, and because the case remained at the pretrial stage the costs of granting the writ were not intolerably high.⁴³ The early stage kept the price of starting over low, which is the kind of fact that bears on the first Liljeberg factor.

The remedies so far belong to the statute and the ethics rules, and the Constitution works differently. In Williams v. Pennsylvania the Court held that Chief Justice Castille’s participation in Williams’s case was an error that affected the State Supreme Court’s whole adjudicatory framework, and it vacated and remanded so that Williams could present his claims to a court unburdened by any “possible temptation . . . not to hold the balance nice, clear and true between the State and the accused.”⁴⁴ A multimember court’s guarantee of neutrality must not be undermined, the Court explained, because the appearance of bias demeans the reputation and integrity not just of one jurist but of the larger institution of which he or she is a part.⁴⁵ The Court then held that an unconstitutional failure to recuse is structural error even if the judge in question did not cast a deciding vote.⁴⁶ A structural error is one that infects the whole framework of the adjudication, so a court does not ask whether it changed the result.

The two regimes must be kept apart. When a judge violates the statute or a rule like Rule 2.11, the court asks how much relief fits, may find the error harmless, and weighs the three factors, as Liljeberg and Microsoft did. When a judge violates due process, no weighing and no harmless-error inquiry occurs. The difference follows from the thresholds. Due process “demarks only the outer boundaries of judicial disqualifications,” and most recusal questions are governed by more stringent and detailed ethical rules, which in many jurisdictions already required disqualification in Williams, including the Model Code rule reaching a judge who served in governmental employment and participated personally and substantially as a lawyer or public official concerning the proceeding.⁴⁷ An appearance violation may reflect nothing about the decision at all, as in Microsoft, where the record showed no actual bias in the findings. Two errors follow from blurring the line: treating every ethics violation as structural error that requires automatic vacatur, and applying harmless-error review to a due process violation.

Exam Analysis Steps

Because the remedy depends first on the source of the violation and then on how far the taint reaches, it is best run in a fixed order. The steps are these:

Identify whether the failure to recuse violated the Constitution, as in the extreme patterns taught in the due process section, or only the statute or an ethics rule.
If due process was violated, treat the error as structural, so the decision cannot stand even if the judge’s vote was not decisive, and do not weigh factors or ask about harmlessness.
If only the statute or rule was violated, order prospective disqualification as the minimum, which removes the judge from all further proceedings in the case.
Decide whether to go beyond the minimum by weighing the risk of injustice to the parties, the risk of injustice in other cases if relief is denied, and the risk to public confidence in the judicial process.
If relief reaches completed work, locate where the taint sits and aim relief there, setting aside the tainted stage or orders and leaving earlier work alone when nothing suggests the taint reached it, or vacating the whole judgment when the movant acted promptly and the appearance of partiality is convincingly established.

The fourth step does the weighing, and the fifth locates the taint that any retroactive relief must follow. The first step comes before both because a constitutional violation skips the weighing altogether.