Who will judge the judges?

Recusal jurisprudence needs a fairness push. Judges cannot decide on their own bias as perceived by the litigants

Recently, Justice Swarana Kanta Sharma, in a detailed order, denied Arvind Kejriwal’s request for recusal. While this judgment might be a passionate cry against politicising the courts, it provides a window into the deeper problems and structures that plague recusal jurisprudence in India. Beneath the rhetoric of judicial independence lies a framework that allows judges to decide their own impartiality, raising uncomfortable questions about fairness and accountability.

Recusal jurisprudence is primarily influenced by legal precedents due to the lack of any law or guideline governing judicial disqualifications. This absence has led to a scenario where tests to regulate judges’ conduct are formulated by judges themselves. This vacuum can be attributed to a vigilante judiciary that strikes down any law which attempts to govern or streamline its functioning. Courts use the grounds of judicial independence to render legislation and amendments unconstitutional.

The question is about fairness – the judge assesses the allegation, applies the legal test, and decides whether to step aside. In effect, the system asks a judge to be a judge in their own cause. This violates a basic principle of natural justice. Disqualification is needed to prevent and intercept judges who have a particular ‘bent of mind’. This is required to promote the appearance of justice and restore faith in judicial administration. Such a framework does not reasonably account for the litigant’s faith in judicial administration.

If a court cannot be the first place where justice appears to be achieved transparently and fairly, it will largely affect public expectations. Judges need to acknowledge that recusal petitions are not always admitted to malign them. The innate structures of the court instead purport a sense of injustice being perpetrated. Compromising the constitutional right to a fair trial and being outright dismissive of petitions hampers the perception of governance.

Justice Sharma’s judgment illustrates another concern: the choice of legal test. She uses the ‘real danger’ test to refuse recusal. This test, laid down in Supreme Court Advocates-on-Record Association v. Union of India, states that disqualification can occur solely on substantive and tangible evidence that conclusively highlights the presence of judicial bias and prejudice. Recusal must happen on the possibility of bias and not its mere probability.

This sets an extremely high threshold, almost impossible for a litigant to meet. Bias, by its very nature, is rarely visible. It operates internally, shaped by experience, perception, and subconscious influence. At best, only circumstantial evidence may be gathered, which might hint towards the presence of bias. Litigants cannot access a judge’s state of mind. Expecting “real danger” to be proven through hard evidence misunderstands how bias works. Such a high burden forces litigants to compromise their perception of justice.

A sharper scrutiny

A more appropriate standard is the “reasonable apprehension” test – whether a fair-minded observer would reasonably suspect bias. This shifts the focus from proof to perception. It acknowledges that justice must not only be done, but must be seen to be done. It places the litigant’s perspective at the centre, rather than treating recusal as a personal affront to judicial integrity.

The inconsistency of a self-evolving jurisprudence has allowed the court to mould the parameters according to its convenience. Litigants face a system where raising a recusal plea is itself an uphill battle: one that risks being seen as forum-shopping or an attempt to delay proceedings. While such misuse is a legitimate concern, it cannot justify a framework that effectively shuts out genuine claims. If anything, the current approach risks the opposite problem. It creates the impression that judges are insulated from scrutiny, and that the system prioritises institutional comfort over litigant confidence: a dangerous perception for any constitutional court.

The solution does not lie in weakening judicial independence, but in strengthening procedural fairness. Recusal decisions could be placed before a different bench, or, at least, involve other judges. Most importantly, courts must move towards standards that recognise how
bias is perceived, not just how it is proven.

At stake is not the reputation of individual judges, but the credibility of the institution. When judges decide their own recusal using standards that are nearly impossible to meet, the process begins to look closed, even if it is not. A justice system that asks for trust must first make space for doubt and address it transparently. Until then, recusal in India will remain less about fairness and more about faith.

(The writer teaches Constitutional Law at the Jindal Global Law School)

Disclaimer: The views expressed above are the author's own. They do not necessarily reflect the views of DH.