In 2013, the Supreme Court described the Central Bureau of Investigation (CBI) as a “caged parrot speaking in its master’s voice” in Centre for Public Interest Litigation v. Union of India. The phrase stayed because it captured something larger than that moment: institutions don’t lose credibility only when they are compromised, but when they are unable to show, convincingly, that they are not.

The issue is no longer confined to the facts of one case; it has begun to raise larger questions about how recusal actually operates in practice. That concern now surfaces, uncomfortably, within the judicial process itself.

The controversy triggered by the affidavit of Arvind Kejriwal brings into focus a question courts have largely preferred to treat as internal: what happens when visible proximity between the Bench and the State is met not with disclosure, but with silence? The allegation - that relatives of a sitting judge are empaneled as government advocates and have received briefs in matters involving the office of the Solicitor General - cannot simply be brushed aside. It may not establish bias in the strict legal sense, but it does not sit comfortably either. It raises questions and those questions don’t disappear just because no one answers them.

The State has strongly opposed the plea, calling it motivated, only sharpening the divide between legal principle and courtroom strategy.

The law on recusal in India is, in a sense, incomplete. There is no clear statutory framework that tells us when a judge must step aside. The doctrine flows from constitutional guarantees of fairness under Articles 14 and 21 and from precedent.

In Ranjit Thakur v. Union of India (1987), the Supreme Court made it clear that the test is not actual bias but the reasonable apprehension of bias in the mind of a fair observer. That idea was reinforced in State of West Bengal v. Shivananda Pathak (1998), where the Court emphasised that justice must not only be done but must also be seen to be done.

In State of Punjab v. VK Khanna (2001), the Court went a step further, recognising that even a reasonable likelihood of bias is enough to undermine decision-making. And in Supreme Court Advocates-on-Record Association v. Union of India (2015), questions of recusal did arise, but without any structured framework emerging from the Court.

So, the principles are not unclear. What is missing is a system.

Apply those principles to the present situation. A judge hears a matter involving the State. Close relatives of that judge are professionally engaged by the State. The State is represented before that very Bench by the Solicitor General. One does not need to prove bias. The only question that matters is whether a reasonable observer could apprehend it. That threshold is intentionally low.

But the real problem isn’t the conflict; it’s the silence that follows. And that silence is not neutral.

If, as set out in an additional affidavit now placed before the Court by Arvind Kejriwal, the empanelment of the judge’s relatives with the government was an existing and verifiable fact, the obvious question is simple: why was this not disclosed at the outset? Why does the system proceed as if nothing arises, only to confront the issue after it is forced into the open?

Disclosure is not an admission of bias. It is a basic safeguard. Without it, the burden shifts onto litigants to discover and raise conflicts. Once that happens, the tone changes. It stops being procedural and becomes adversarial.

At some point, the system has to decide whether it prefers silence over scrutiny.

Indian procedural law does not treat apprehension of bias lightly. Under the Code of Civil Procedure, 1908, courts can transfer cases where the ends of justice require it and the Supreme Court of India can do the same across states. The approach continues under the Bharatiya Nagarik Suraksha Sanhita, 2023, which recognises transfer where a fair trial may be in doubt.

So the system accepts that even the perception of bias can justify shifting entire proceedings. And yet, when it comes to recusal - the more immediate safeguard - it leaves the matter largely to individual discretion. That contrast is difficult to ignore.

The concern becomes sharper in the broader public environment. Once questions about institutional neutrality enter public discourse, the burden on the system becomes heavier. In that setting, even routine silence begins to carry weight. The issue is not whether the court is influenced, but whether the system does enough to ensure that it never appears to be.

That responsibility does not lie with the court alone. A law officer such as the Solicitor General is not just another counsel. The office carries a certain expectation. In situations where proximity could raise questions, the issue is not just whether appearance is legally permissible. It is whether proceeding without disclosure is wise.

Silence creates a vacuum. And in that vacuum, perception does not stay neutral. It shifts; it hardens.

It will be said that conflict of interest is not the same as compromise. That is correct. But that distinction only holds when the system openly addresses the conflict. Where there is neither disclosure nor reasoning, the line begins to blur - not in doctrine, but in perception.

That is exactly what the “caged parrot” remark warned against. Institutions do not become suspect only when they are controlled. They become suspect when they appear unable to demonstrate independence in a way that is visible and verifiable.

The larger issue is structural. Recusal in India still operates largely as a matter of personal discretion. Judges decide. Reasons are not always recorded. There is no uniform disclosure requirement, no consistent practice. Similar situations can lead to different outcomes and the system offers little explanation.

This is not about individuals. It is a failure of design.

The same tension was visible in Supreme Court Advocates-on-Record Association v. Union of India (2015), where the Court defended its independence but stopped short of embracing deeper transparency. That tension has not gone away. It has simply returned in a more immediate form.

The burden this creates is not sustainable. If a judge recuses, it invites speculation. If the judge does not, it invites suspicion. Without disclosure and without reasons, both outcomes weaken confidence.

If this is to change, recusal cannot remain informal. It has to become structured. Disclosure must come first, not last. Reasons must be given, not avoided. And responsibility cannot rest only with the Bench. Government law officers must also exercise restraint in situations that clearly raise perception concerns.

The affidavit of Arvind Kejriwal does not prove that any judge is compromised. But it does something just as important. It shows that the system does not yet have a convincing way to demonstrate that it is not.

That gap matters.

Recusal was meant to protect legitimacy. In its present form, it risks weakening it. As long as it remains opaque and discretionary, every conflict will carry a shadow. Courts may continue to ask for trust. But trust without transparency does not last.

And when that happens, the question is no longer about independence. It is about credibility.

Jitendra Mohananey is a practicing lawyer specialising in insolvency, banking and commercial litigation.