The seriousness of recusal applications and the need for a third party mechanism

The recent order of the Delhi High Court in CBI v. Arvind Kejriwal has exposed a fault line in India’s recusal system. The order calls this a Catch-22 situation, where a political litigant gains either way: if the application is allowed, the bench stands reconstituted and the bias allegation is impliedly vindicated; if it fails, the rejection itself is recast as further proof of the very bias alleged.

This is exactly what happened since Arvind Kejriwal’s application was rejected, wherein he released a public letter boycotting the proceedings stating that he has lost all hope of a fair trial and believes that the bench is biased against him.

Some may see this as bench-hunting of the kind deprecated in Subrata Roy Sahara v. UoI. Even so, the present structure leaves room for such maneuvers. When a recusal application is decided by the same judge against whom the challenge is made, it is easier to create narratives of bias irrespective of how the chips fall. This is a serious issue given that recusal applications are inherently more damaging to the system than voluntary recusals.

Voluntary recusals stem from a judge’s own conscience. The judge himself doubts his ability to render, or be seen to render, an impartial judgment. Contrastingly, a recusal application is made when even though the judge finds himself fit to preside, the litigant disputes that fitness. Therefore, in some sense, recusal applications raise a motion of ‘no-confidence’ on the impartiality of a presiding judge. This is a serious challenge to the court’s image as a neutral arbiter and is directly tied to the judiciary’s public perception.

Optically too, voluntary recusals, as a cautionary step, may add to build up judicial trust. However, every recusal application, being a ‘demanded’ measure from the litigant, prejudices this trust even before the application gets assessed on its merits. Inevitably, application based recusals are more aggravating than voluntary ones.

Naturally, in order to maintain this trust, it is important that more avenues are provided for voluntary recusals and recusal applications are limited to rare cases. Unlike India, multiple jurisdictions already list down situations where a judge is automatically disqualified to preside over a hearing, thereby triggering a voluntary recusal. For example, 28 USC § 455(b) provides the situations where a judge must disqualify himself, a principle elaborated by the US Supreme Court in Liteky v. United States. Similarly such lists are provided under Article L111-6 of the French Code of Judicial Organization and Section 41 of the German Code of Civil Procedure.

In sum, in all these rules, common instances include; (a) personal or financial interests in the dispute; (b) family or personal relationship with the parties; (c) prior participation as lawyer for any of the parties; or (d) appearance as a witness or expert in the matter. Indian law already accepts the first category as an absolute disqualification [Manak Lal v. Dr Prem Chand (1957), reaffirmed in the NJAC case] but stops short of codifying the rest. Similar standards must be introduced in India, either through statute or by amendment to the court rules.

In such rare instances, the application is assessed based on tests such as whether: (i) there is a ‘real danger of bias’ - previously followed in the UK in R v. Gough, (1993) and imported into Indian law in State of Punjab v. Davinder Pal Singh Bhullar (2009) and refining the earlier ‘reasonable apprehension’ test laid down in Ranjit Thakur v. UoI (1987); or (ii) there is ‘real possibility of bias’ [followed in the UK after Porter v. Magill (2001)]; or (iii) ‘if the impartiality can be reasonably questioned’ [followed in the USA; see 28 USC § 455(a) and Liteky].

Despite these moderately varying tests, their application is always from the perspective of a ‘reasonable observer’ or ‘reasonable person’ which is a common standard seen in most jurisdictions, including India. Arguably, a truly ‘reasonable person’ would be nothing but a judicial person. Thus, if such persons are required to assess recusal requests, then it indirectly indicates that the need is to introduce judicial persons for assessing applications.

The test of ‘reasonable person’ is unfounded. The Supreme Court in Ranjit Thakur itself accepted that judicial introspection is an unreliable guide, directing instead that the judge “look at the mind of the party before him”. The critique that follows is a logical extension of that admission. McKoski argues that such a reasonable person does not exist, since to achieve the desired reasonability a lay person would need to shed every societal influence and so comes to resemble the very qualities desired of a judge. The reasonable lay observer thus shares little with the common public; he is, in effect, another judge. The test purports to create a distance between the adjudicating mind and the societal perspective it claims to draw on, but that distance is imaginary.

Practically, what is required is that there is actual distance between the mind that has been alleged to hold bias and the mind that is deciding on the allegation. The solution to this is simple: rather than the objected judge trying to create a split person (a reasonable observer) in their mind and then assessing from their perspective, the system should allow another judge to look at the litigant's recusal application. This also aligns the system with the principle of nemo judex in causa sua (no person should be a judge in their own cause).

Many would propose that the judge deciding on the recusal application should be the Chief Justice of the court. There is merit in this argument. Given that we follow the master-of-the-roster convention affirmed in Campaign for Judicial Accountability and Reforms v. UoI (2017), Asok Pande v. Supreme Court of India (2018) and Shanti Bhushan v. Supreme Court of India (2018), any dispute raised regarding the fairness of the allocated judge, should logically go back to the master. It must be clarified that a recusal panel does not displace the Chief Justice’s allocation power; it performs a distinct adjudicatory function on impartiality, akin to the routine reconstitution that follows a voluntary recusal.

A catch in this system may arise when the application is submitted against the Chief Justice himself. Nonetheless, once it has been principally agreed that applications need to be decided by a separate judge, such procedural issues are easy to resolve. One way could be by having the application against the Chief Justice, decided by the judge next in line of seniority. Understandably, an argument against this could be that such a judge may be wary of accepting the recusal request against the Chief in fear of displeasing the roster’s master and to avoid future administrative backlash.

An alternative solution is to have a system wherein for all recusal applications across the board, the decision is made by a panel of two or three judges. The additional panel duty can be decided when the roster is decided. The panel can arrive at a decision based on the written application without the need to summon the applicant or the challenged judge.

The proposed measure is not novel and many jurisdictions are known to follow a third-party structure for recusal applications. In France, for example, a challenged judge has no say in deciding a recusal request, which is decided by the higher court (Article 344, French Code of Civil Procedure). Germany goes a step further in formalising this principle. The German Code of Civil Procedure (Section 45) explicitly states that a challenged judge cannot decide on the question of his bias and the matter must be decided by other members of the same court. The European Court of Human Rights (ECHR) explicitly rules that a judge would not decide any impartiality challenge against him and the same would be decided by either the court’s President or the rest of the chamber. (Rule 28, Rules of the ECHR).

Preventing Catch 22 scenarios - Where the application is rejected by the challenged judge, it is easier to create a narrative of continuing bias. However, such narratives are difficult to promote when the application has been rejected by a panel of judges who are inherently positioned at a distance from the allegations. This also ensures that justice is not only being done but also seeing being done.

Upholds the dignity of the judiciary and the judge who has been challenged - Once an application is submitted, a question on the impartiality gets raised irrespective of the substance of such allegations. This may create a situation where the judge ends up defending their right to preside before the litigant, turning the tables of the courtroom. In CBI v. Arvind Kejriwal, Sharma J. observed that the litigants had “put the Judge and the institution on trial,” a danger Shivananda Pathak v. State of WB had earlier warned against. However, if the decision to preside or not is decided by a panel, the challenged judge has to simply follow its decision similar to a roster decision and does not have to defend themselves before the litigant.

Promotes only evidence-based applications to be filed - When an application is made to the challenged judge, then any litigation coming in bad faith would find it sufficient to cast mere aspersions to create a perception of bias. Psychologically, an aspersion-based challenge has the risk of being taken personally. These can, therefore, either be rejected outrightly without assessing any merits or on the other extreme ,may see the challenged judge end up overcompensating in order to show that there is no bias. In such cases, the judge may end up recusing despite weak reasons in the application. Contrary to this, when it is known that the application will be assessed by a separate panel of judges, the litigant is likely to have more pressure to substantiate their statements with evidence.

Each recusal application calls into question the court’s integrity. Therefore, the process surrounding them should be such that it leaves no room for promotion of false narratives. It is important that public confidence in the judiciary is not eroded. With the Supreme Court in 2025 declining to frame binding guidelines under Article 142 in Chandraprabha v. UoI (2025), the route now lies in formal rule-making.

To ensure this, a system must be created where disqualification metrics are clear through an amendment to the Supreme Court Rules, 2013 (Article 145) and parallel High Court Rules (Article 225) and for the rare instance where an application is filed, it is heard by a panel of judges from the court. This mechanism would prevent judges being put to trial, align with the principles of natural justice and strengthen judicial trust.

Nidhi Singh is an Advocate-on-Record at the Supreme Court of India.

Aatmik Jain is an Associate Advocate at the Chambers of Nidhi Singh.