Gremlins, Borgs, and judicial impartiality in multi-member courts: QYFM v Minister for Immigration, Citizenship, Migrant Services and Multicultural Affairs [2023] HCA 15
Sarah Fulton and Geneviève Murray
20.06.2023
Judicial impartiality — and within that, an absence of bias — is at the heart of the Australian judicial system and central to how judges see themselves. But while serving and retired judges of the High Court have had a lot to say about when judicial bias arises, they have (with some notable exceptions, as noted in the ALRC Judicial Impartiality Report, p 234) said little publicly about how such matters should be raised with and considered by the courts. Until now.
In QYFM v Minister for Immigration, Citizenship, Migrant Services and Multicultural Affairs [2023] HCA 15 (‘QYFM’), judges of the High Court not only clarified the law on apprehended bias as it applies to a judge who was previously involved in the prosecution of a party, but also took the opportunity to set out their views on the processes for determining issues of bias raised in multi-member courts (such as Courts of Appeal, Full Courts, or the High Court itself).
QYFM signals a significant shift in the orthodox practice in Australian appellate courts, with a majority of judges holding that, ultimately, the question of the apprehended bias of one member of a multi-member court is a matter of jurisdiction on which the whole court must be satisfied, not simply an issue to be determined by the challenged judge alone.
The Court’s decision comes against the backdrop of a report by the Australian Law Reform Commission on judicial impartiality published last year (in which we were involved). The ALRC’s inquiry, referred in the wake of several high-profile cases concerning questions of judicial bias, considered the processes federal courts use to determine bias challenges. It made recommendations (see p 13) for greater transparency about recusal and disqualification procedures and specific reforms to those procedures, particularly involving multi-member benches.
As the delivery of six separate judgments in QYFM shows, these are matters on which reasonable minds can differ. The judgments lay bare different views about central and complex questions of jurisdiction, judicial power, and the hierarchy of courts.
Nevertheless, the judgments, which traversed the globe for instruction and reached deep into academia and law reform, signal an appetite for greater clarity and provide a path forward for Australian courts to develop more transparent procedures.
The post will set out the case’s background and briefly touch on the court’s consideration of the substantive question of disqualification on the facts, before turning to the procedural aspects and their implications.
Background
QYFM was an appeal from a decision of the Full Court of the Federal Court. The appellant’s visa had been cancelled on character grounds because of a serious criminal conviction for drug offences in 2013. The Minister’s delegate decided not to revoke the cancellation. The Administrative Appeals Tribunal affirmed the decision not to revoke the cancellation. The Federal Court dismissed the appellant’s application for judicial review. QYFM then appealed to the Full Court of the Federal Court, comprising Bromwich J and two other judges.
On the day of the hearing before the Full Court, Bromwich J discovered and disclosed to the parties that he had personally appeared for the Crown in the appeal of QYFM’s 2013 criminal conviction in his previous role as Director of Public Prosecutions. His Honour did not consider the circumstances to give rise to an apprehension of bias, as the earlier appeal ‘related to a pure legal question’, but he recognised counsel might wish to apply for his recusal.
Counsel for QYFM sought that Bromwich J recuse himself, and was invited by the presiding judge, McKerracher J, to make oral submissions. Following the usual practice, after a short adjournment, Bromwich J provided brief oral reasons for his decision not to recuse. The Court proceeded to hear the matter and unanimously rejected the appeal.
QYFM then appealed to the High Court. QYFM argued that there was a ‘reasonable apprehension of bias’ in relation to Bromwich J because of his prior involvement in QYFM’s criminal prosecution. QYFM also argued that the Full Court erred in the process by which it determined the issue of bias — arguing that the question of recusal should have been determined by all members of the Full Federal Court, rather than by Bromwich J alone.
A 5:2 majority of the High Court found that Bromwich J was affected by reasonable apprehension of bias. The Court allowed the appeal, set aside the Full Court’s judgment and ordered that the matter be remitted to be heard and determined by a differently constituted Full Court.
Apprehended bias: No automatic categories but a ‘relatively undemanding’ test
The separate judgments in QYFM contain much of interest on the substance of disqualification for bias — including the interaction between judicial impartiality and independence, the links between actual and apprehended bias, the attributes of, and knowledge attributed to, the fair-minded lay observer, the relevance of how an actual litigant would perceive the situation, and the ‘gremlins’ of unconscious bias.
There was widespread agreement about how the case should be approached in theory. All members of the Court applied the well-settled ‘double-might’ test for apprehended bias, which asks whether ‘a fair-minded lay observer might reasonably apprehend that the judge might not bring an impartial mind to the resolution of the question the judge is required to decide’ (Ebner v Official Trustee in Bankruptcy (‘Ebner’) at [6]). They all then applied the analytical steps set out in Ebner. In doing so, they reaffirmed that there are no ‘automatic’ categories of apprehended bias, even if ‘the outcome of the Ebner analysis in some cases may be so obvious as to warrant little or no elaboration’ (at [40] (Kiefel CJ and Gageler J)).
The judges were unanimous in their conclusion that the first step was satisfied — the fact that Bromwich J had previously appeared against the appellant might lead a fair-minded lay observer to conclude that he might decide the case other than on its legal and factual merits. On the second step, however, Steward and Gleeson JJ were not satisfied that there was a logical connection between Bromwich J’s previous role and the feared deviation in this case.
For those in the majority on this issue, it was not to the point that Bromwich J’s involvement was in separate proceedings and only on a matter of law. The fair-minded lay observer would know the risk of the judge having an ‘unfavourable mindset’ towards the appellant, even unconsciously, and would be mindful of how the appellant would feel seeing the past prosecutor now involved in deciding his case.
Marking a clear departure from some earlier case law (which Bromwich J had ‘understandably’ been guided by (at [54])), the judges in the majority declared three intermediate appellate criminal court judgments concerning past prosecutions by judges to be either incompatible with contemporary practice or as having had been wrongly decided (R v Garrett (1988) 50 SASR 392, McCreed v The Queen, and Muldoon v The Queen). As Kiefel CJ and Gageler J emphasised, the fair-minded lay observer is not a lawyer and ‘may be less inclined to dissociate the advocate from the cause advocated than would someone steeped in the adversary process with a cultivated sense of the ethics of the legal profession and the profundity of the judicial oath’ (at [48]).
Procedure: disqualification in multi-member courts
On the substance, the High Court’s judgment continues a line of recent cases (including Charisteas and CNY17) adopting a determinedly realistic approach to the contemporary fair-minded lay observer. This observer lives in a society where ‘habits of deference to authority are no longer ingrained’, the ‘vagaries of “human frailty”, even in those wielding power, are now accepted’, and the ‘unknowable effects of the subconscious are also acknowledged to exist’ (at [297] per Jagot J). But while the substantive law has adapted significantly to reflect this reality, the ALRC report noted that procedures by which bias claims are raised and determined, and their ability to be known by those outside the ‘club’, have not necessarily kept pace (from p 410). For this reason, the ALRC recommended that the Commonwealth courts consider changes to their disqualification procedures and publish guidelines on how they operate, modelled on examples from the New Zealand courts.
By addressing aspects of procedure in considerable detail, the High Court’s decision is a significant step forward. The Court recognised that the procedural issues did not necessarily need to be decided to dispose of the case. But, having been raised by the appellant and fully argued, all members of the Court except Gleeson J considered it appropriate to consider them. In doing so, Gordon J noted that ‘the procedure for deciding objections for apprehended bias raises issues that go to the heart of judicial duty, process and, indeed, the protection of impartiality, independence and procedural fairness, all essential characteristics of Ch III courts under the Constitution’ (at [87]).
Boldly going where (almost) no (Australian) court has gone before
The conventional practice in Australia has been that an application for disqualification of a judge is heard and determined by the challenged judge alone. This practice applies in single-judge cases and generally (although not universally) carries through to multi-member courts. In QYFM, a majority of the High Court (Jagot J dissenting and Gleeson J not deciding), determined that the question of apprehended bias of a member of a multi-member court is a matter of law going to jurisdiction that should ultimately be considered by the court as a whole, and offered guidance as to how such matters should be dealt with. In doing so, the judges looked at the practice of several other common law jurisdictions (including changes to practice in England and Wales, Northern Ireland, New Zealand, Singapore and South Africa). Members of the majority also cleared up some remaining areas of uncertainty, including the fact that interlocutory decisions on bias should be recorded in an order that is subject to appeal or review (see, for example, Kiefel CJ and Gageler J at [28]).
Amongst these judges, however, a significant division remained. On one hand, Kiefel CJ and Gageler J considered that an application for disqualification on the grounds of actual or apprehended bias should be determined from the outset by all members of the court. In their view, even if the challenged judge considers that they should recuse themselves for apprehended bias, the judge should continue to sit if a majority of the court disagrees (at [32]–[33]).
On the other hand, Gordon, Edelman, and Steward JJ — for both theoretical and practical reasons — considered that the challenged judge should first decide any disqualification application. Only if the judge decides not to recuse themselves, and the application is maintained or the Court raises the issue of its own motion, should it be considered by the other judges. An important factor for both Gordon and Edelman JJ was that bias has an ethical, as well as jurisdictional, dimension, which only the subject judge can answer (at [96] and [107]). Confirming that judges are not like cyborgs in Star Trek sharing a ‘hive mind’, Edelman J notes that: ‘The judge’s conscience is not melded like a Borg with the conscience of the other members of the court. Rather, the subject judge remains throughout the proceedings under the same ethical duty to recuse themself if they consider that they are biased or are the subject of an apprehension of bias’ (at [131]).
Jagot J took a very different view — that even in a multi-member court, bias is a unique ethical and jurisdictional matter for the challenged judge alone. In a detailed judgment, Jagot J offered a vigorous defence of the existing practice. She questioned both the need for change and the source of judicial power for judges at the same level of the judicial hierarchy to prevent another judge sitting (at [311]-[315], [322]). The latter raises difficult issues that the ALRC also grappled with (from p 246) in framing its recommendations for procedural reform — issues that are even more difficult in the context of single-judge cases (see briefly below). However, the majority did not see this as a barrier to multi-member courts determining recusal disqualification relating to one of its members.
A way forward?
Where does this leave us? Four High Court judges (Gordon, Edelman, Steward, and Jagot JJ) consider that any issue of bias is, at least initially, a question for the challenged judge. Five judges (Kiefel CJ, Gageler, Gordon, Edelman and Steward JJ) held that ultimately the question of bias is one of jurisdiction for all members of a multi-member court to consider.
With this in mind, a possible way forward is provided by Gordon J, who fell into both of these camps. She sets out a broader set of steps multi-member courts can take to identify and address apprehended bias issues, ‘[a]dopting and adapting what was outlined and discussed in the ALRC Bias Report’ (at [100]). These procedures attracted the specific endorsement of Edelman and Steward JJ and, as such, could conceivably inform future practice and guidance in both the High Court and other federal and state courts. They include:
An emphasis on preventive administrative processes from the outset, including through circulating case lists and the opportunity for judges to recuse themselves.
Notification of potential issues to parties and other members of the court as constituted if a potential issue arises but the judge decides not to recuse.
Objections on bias grounds should first be considered by the challenged judge.
Other members of the court then to satisfy themselves of jurisdiction if the judge decides not to recuse.
Reconstitution will be required if the court is not satisfied of jurisdiction. In cases of real doubt, ‘it may be prudent for the Court to reconstitute’.
Formal order and reasons are required where a multi-member court considers the issue of jurisdiction (on its own motion or because a challenge is maintained).
Within this framework, Gordon J emphasises that each case and court is unique and that ‘there can be no universally applicable rules or procedures’ (at [93]). Appropriate procedures may be affected by the ‘practices, jurisdiction, level in the judicial hierarchy, composition and work of the [court] … and the various stages of a proceeding … in which the question of an apprehension of bias may arise’ (at [93]). This is borne out, for example, by differences in the procedures for recusal publicly adopted by different appellate courts within New Zealand. However, the potential for different approaches between courts and at different stages of proceedings is also why transparent court-specific guidance would be helpful for litigants and lawyers and is important to uphold public confidence. While guidelines, by their nature, would allow for variation in a given case, the flexibility should be carefully balanced against the need to promote public confidence and the benefits of predictability.
In the short term at least, the area of most uncertainty is the step at which the remainder of the court considers the issue of bias. Should the challenged judge be excluded from the deliberations at that stage? Gordon J suggests that they should (at [98]), Edelman J indicates that they should not (at [134]). And how should the relevant facts be established? Should the challenged judge be required to put their knowledge of relevant facts on to the record, as explored by Kiefel CJ and Gageler J at [32] and Edelman J at [139]? Or does this run the risk of exposing the judge to cross-examination and put judges in the ‘invidious’ position of having to second-guess their judicial colleagues (Jagot J at [321])? Here, court-specific guidelines, albeit with in-built flexibility, are important to ensure that litigants and the public can have confidence that the processes being followed are legitimate.
What about single-judge cases?
What could the judgment mean for single-judge cases? The conventional practice (by which the challenged judge determines the application for recusal) has attracted criticism in that context too. Critics argue that the current process demands ‘an almost inhuman level of impartiality’ from the judge (given what we know about how difficult it is to see bias in ourselves), has a chilling effect on applications, and seems unfair to many litigants and members of the public. The ALRC considered these criticisms and, in its report, recommended incremental change by which single-judge courts should maintain the conventional practice, but introduce the option of discretionary transfer of bias applications by the challenged judge to another judge.
In QYFM several members of the Court touched on single-judge cases, noting that a plurality of the Court in Ebner had referred to the conventional approach as being ‘correct’ (see Gordon J at [89], Edelman J at [123], Jagot J at [307]). None of the judges in QYFM disagreed with this assessment.
This might suggest that there should be no procedural reform in single-judge cases. However, it does not necessarily close the door on discretionary transfer. Under that procedure, the initial responsibility for considering (though not ruling on) a disqualification application lies with the challenged judge. The judge may — having examined their conscience — then decide to transfer the question to another member of the court for determination. This would seem consistent with Edelman J’s statement that practice ‘permits, and usually requires, the first consideration [of a recusal application] to be made by the subject judge’ (at [137]) and Steward J’s qualification that responsibility lies ‘initially’ with that judge (at [193]).
In line with statements of the plurality in Ebner, Jagot and Gordon JJ questioned the source of power for another judge at the same level of the judicial hierarchy to make an order preventing another judge from sitting. The ALRC concluded these concerns could be met (from p 246). First, jurisdiction in a matter, and therefore the power to hear and determine it, lies with the court, not the individual judge. Second, any order made by a transfer judge would not be framed as preventing the challenged judge from sitting, but as returning the case to the registry for reallocation to ensure a properly constituted court. This would seem consistent with Kiefel CJ and Gageler J’s statement that ‘responsibility for ensuring an absence of bias — whether actual or apprehended — lies with a court as an institution and not merely with a member of that court whose impartiality might be called into question’ (at [27]). This may be a space to watch — things are only impossible until they’re not.
Conclusion
As the ALRC noted in its report, ‘[t]he Australian judiciary is highly respected internationally for its integrity and its impartiality’. But this should not be taken for granted. The innovation and willingness to shift from the conventional practices in QYFM reflects such a recognition from Australia’s highest court. As Kiefel CJ and Gageler J stated, the issue of determining the appropriate procedures for raising bias in multimember courts is ‘one of principle’ and ‘general practical importance’ (at [3]).
The opportunity now lies with the heads of jurisdiction to engage with the High Court’s guidance in establishing transparent guidelines for determining issues of bias. We look forward to watching the courts innovate in this space.
Sarah Fulton is a Principal Legal Officer at the Australian Law Reform Commission.
Geneviève Murray is a Lecturer at Griffith Law School.
Sarah and Geneviève were part of the ALRC’s legal team for the Judicial Impartiality Inquiry. Justice Bromwich also recently served as a Part-Time Commissioner of the ALRC for its Corporate Criminal Responsibility Inquiry (2020).
Suggested citation: Sarah Fulton and Geneviève Murray, ‘Gremlins, Borgs, and judicial impartiality in multi-member courts: QYFM v Minister for Immigration, Citizenship, Migrant Services and Multicultural Affairs [2023] HCA 15 ’ on AUSPUBLAW (19 June 2023) <https://www.auspublaw.org/blog/2023/6/gremlins-borgs-and-judicial-impartiality-in-multi-member-courts-qyfm-v-minister-for-immigration-citizenship-migrant-services-and-multicultural-affairs-2023-hca-15/>