The Chief in his Chiefdom: On the Legacy of Chief Justice B.R. Gavai

When I was a child, I loved reading “Choose Your Own Adventure” storybooks. In these stories, you – the reader – were the protagonist. Every few pages, you would be asked to make a choice, which would lead you to different scenarios, and – eventually – to different endings. But the great thing about Choose Your Own Adventure stories was that if you didn’t like it, you could just go back, make a different decision, and change the ending. It was that simple.

If there is one common thread that runs through the six-month tenure of Chief Justice B.R. Gavai – which ends today – it is that he treated his Chief Justiceship as a Choose Your Own Adventure story. If there was some long-standing precedent that he did not like, he used his powers as the master of the roster to overturn it. If another bench of the Court issued a ruling that he didn’t like, he set up his own bench and overruled it. It did not matter that in doing so established norms and conventions were shredded, entirely new jurisdictions were invented, and the rest of the Court was diminished. What mattered was the outcome, which – just incidentally – often happened to be aligned with the outcome that the central executive wanted. 

Let us be clear. Every Chief Justice that this blog has analysed has manipulated the roster to nudge the Court towards their preferred outcomes. CJI Misra did it. CJI Gogoi did it. CJI Bobde did it. CJI Ramana did it. CJI Lalit did it. CJI Chandrachud did it. CJI Khanna did it. Some of these manipulations were egregious, but whether out of a sense of self-restraint, or comity with their fellow judges, or some lingering concern for institutional integrity, these judges did not cross a certain line. In CJI Gavai’s tenure, the line ceased to exist. If his predecessors were – as the old saying goes – still only “first among equals,” despite everything, then he cast off the last remaining shackles of due process, and became a Chief in his Chiefdom.   

As with previous posts, this blog will consider the tenure of CJI Gavai from three lenses: as the head of the collegium (A), as the master of the roster, in control of the direction of the Court (B), and as a judge of a constitutional court (C).*

A. Head of the Collegium

    There are three major criticisms levelled at the collegium system of judicial appointments: first, that it is nepotistic; secondly, that it is non-transparent; and thirdly, that its opacity allows the central executive to exercise an informal, back-channel influence over appointments. Previous Chief Justices’ tenures would rank poorly on one, or perhaps two, of these indices. Chief Justice Gavai’s tenure hit rock bottom on all three. 

    On nepotism: during his tenure, the Collegium (that he was the formal head of) appointed his own nephew a judge of the Bombay High Court. Such a direct conflict of interest has not been seen in recent memory. Of course, it was informally leaked that he did not take part in the specific meeting in which his nephew’s name was put forward. We do not know this, because the records of the meetings remain opaque. But either way, it does not matter: whether or not he was personally present at the meeting does not change the fact that he was the head of the Collegium at the time, and does not cleanse the conflict of interest. 

    On non-transparency: as we have previously noted, CJI Chandrachud’s tenure saw some basic information being provided about the deliberations of the collegium in selecting judicial candidates. This was not very much – and in some ways, it was even counterproductive in establishing hard limits to the degree of transparency that would come from the collegium – but it was something. Under CJI Khanna’s tenure, this something was also rolled back, and under CJI Gavai it vanished completely, taking the collegium back to the cabal-of-high-priests-meeting-in-a-secret-conclave that it has long been accused of being (a late disclosure of the personal details of appointees – such as their caste and gender – does nothing to remedy this basic defect and – once again – provides an appearance of transparency where there is none) . 

    The nadir of this opacity was reached in the controversy around the elevation of Justice Vipul Pancholi to the Supreme Court. This elevation was controversial for a number of reasons involving supersession of multiple other High Court judges and the bypassing of senior, qualified women judges (in a moment of supreme irony, in his farewell speech, CJI Gavai lamented having been “unable” to appoint any women judges to the Supreme Court – while himself being responsible for superceding them!) 

    Now, in the absence of any information, it is impossible to know what – if any – reasons the Collegium had for this appointment, just like it is impossible to know what reasons the Collegium has for any appointment. But for our purposes, it is crucial to note that one of the judges (in fact, the only woman judge on the Collegium) – Nagarathna J. – dissented from the Collegium’s decision, and even her dissent note has been kept a secret. In an interview published today, the CJI stated that if Nagarathna J.’s dissent had “merit,” it would have been accepted by the other judges in the collegium (we all know that is how things work – after all, if Galileo’s views had merit, surely the Catholic Church would have accepted them!). But by this logic, we might as well stop publishing dissenting judgments, because if they had “merit,” they would not be dissents at all. Such flimsy, post-facto justifications need not detain us further, were it not for the frightening future that they portend: a Supreme Court that is so centralised under the Office of the Chief Justice, that even expressions of dissent from within the institution are kept private: what Justice Robert Jackson once called “the unanimity of the graveyard.”

    In sum, thus, not only do we have no access to the Collegium’s reasons for elevating someone as a judge of the Supreme Court, because these reasons are no longer recorded, we do not even have access to a judge’s dissent note that is formally on the record. Indeed, CJI Gavai’s decision to keep Nagarathna J.’s dissent note private was a harbinger of a theme that would be repeated throughout his tenure: a repeated diminishment of the collegial nature of the Supreme Court, and of other judges of the Court, at the expense of the all-powerful Office of the Chief Justice. 

    On executive influence: That the executive has a role to play in judicial appointments is a proposition accepted in most constitutional democracies. What this role ought to be is a matter of debate. But what is not a matter of debate is that this role should be transparent and clearly defined by law, and not something informal and shadowy. This is true whether a country has its executive literally appointing judges (as in the United States), or where the executive is one voice in a multi-member appointment commission (as in the United Kingdom), or where the executive has a downstream role in selecting from a shortlist (as in South Africa). The reason for this is obvious: as there is so much constitutional litigation against the executive, it is imperative to know precisely what role it has had in selecting the judges who will be adjudicating these cases. 

    The non-transparency and opacity of the Collegium is corrosive to this, and the more non-transparent the Collegium gets, the more corrosive things become. Nobody knows the exact role the executive plays, but everyone knows that through the pocket veto and other forms of pressure, it does play a role. And in the absence of specific information, as is inevitable, knowledge is replaced by the kind of speculation that is particularly toxic to maintaining public trust in institutional integrity. 

    None of this is new as far as the Collegium goes, but here again CJI Gavai broke new ground, and not in a good way. This involved the transfer of the High Court judge, Atul Sreedharan J. It was alleged that Sreedharan J. was repeatedly shunted between High Courts because he was a firm and pro-liberty judge whose orders were embarrassing the executive. We cannot comment on this because – once again – there is no transparency with respect to when, where, and for what reason the Collegium orders the transfer of a High Court judge. What we can comment on, however, is that when Sreedharan J.’s transfer order to the Chhattisgarh High Court (where he would have been a member of the High Court Collegium) was countermanded, and instead, he was transferred to the Allahabad High Court (where he would not be), the Collegium Resolution – for the first time – explicitly stated that this was being done at the recommendation of the central executive. 

    But if the whole purpose of the Supreme Court taking over the power of appointment and transfer of judges was to insulate these processes from the executive, then what role does an executive recommendation have to play in a transfer decision? More importantly, what was the recommendation, and why was it made? Yet again, we do not know, because nothing is transparent; what we do know now, however, is that it is on the record that a judge’s transfer was influenced by the executive (to the detriment of that judge’s seniority). 

    One does not need to think too far to see the impact this will have on other High Court judges: earlier, they were subject to transfer by an unreasoned collegium order. Now, they are subject to transfer by an unreasoned collegium order on the recommendation of the executive. It cannot be understated how deeply this undermines judicial independence.  

    B. Master of the Roster and the Direction of the Court 

      For the last eight years, this blog has critically examined the extraordinary administrative powers enjoyed by the Chief Justice as the “Master of the Roster”: this includes the power to assign cases to any judge or combination of judges (the CJI has thirty-three to pick from), the power to decide which cases are to be heard, and the power to leave cases hanging in cold storage. These powers – and the way that they have been wielded by each Chief Justice – means that it is, in large part, the CJI who decides not just the cases he places before himself, but also, in a very significant way, the direction of the Court under his tenure. 

      I. The Direction of the Court

      What were some of the significant features of the direction of the Court under CJI Gavai? To start with, the Court continued its time-honoured tradition of judicial evasion: that is, its refusal to hear constitutional cases against the executive, and its maintenance thereby of a status quo that benefits the executive. During CJI Gavai’s tenure, long-pending constitutional challenges continued to go unheard, while – on the other hand – fresh cases pertaining to a very specific issue (the appointment of the district judiciary) were swiftly taken on board, heard, and decided. 

      However, the most egregious case of judicial evasion during CJI Gavai’s tenure was how the Supreme Court dealt with the challenge to the Election Commission’s decision to carry out a Special Intensive Revision (S.I.R.) of the electoral rolls in advance of Bihar state elections. The S.I.R. was challenged soon after it was announced, in early July. It raised fundamental constitutional issues that went to the heart of the electoral and democratic process, such as the competence and powers of the Election Commission to conduct such an exercise in the manner that it was doing, the constitutionality of placing the burden of proof on individuals to demonstrate eligibility through a list of documents, the possibility of mass disenfranchisement, and so on. It was assigned by CJI Gavai to a bench presided over by his next-senior-most colleague. 

      It is now late November. More than four months have passed, and the Supreme Court has not held a single substantive hearing on the constitutional challenge. What it has done is hold multiple, hours-long hearings on issues such as whether “Aadhaar” should be added to the list of documents or not, issues that are all obviously downstream from the constitutional challenge itself. In the meantime, the Bihar elections came and went, and the S.I.R. was subsequently announced in multiple other states. 

      To those who have been following and been involved with the Supreme Court over the last decade, this is a drearily familiar story: constitutional challenges languish until they become infructuous, until what they were challenging has become a fait accompli. In the meantime, the Court holds multiple – and chaotic – hearings on issues that are entirely disconnected from the substance of the challenge itself. The S.I.R. case is a particularly egregious example, as this is a challenge about the electoral process – quite literally, the very ground rules that enable democracy to function. For a constitutional court, this should take first and urgent priority; however, on repeated occasions, on listed dates, the Court told petitioners that it was “hearing other matters”, and this case would have to wait, or be heard for a few minutes (thus precluding any substantive engagement). No constitutional court worth the name would treat democratic disputes in such a callous manner. 

      The Court’s refusal to hear the case – and instead, to rule through a series of one-line interim orders – raises another issue. Constitutional adjudication is not just about settling the rights and obligations of parties, and protecting individuals from State power: it is also about judicial accountability. When Courts write judgments, they must necessarily give legal reasons, and in that process, explain themselves to the public: how, why, and in what manner have they reached their decision? This opens up Courts to public scrutiny and critique, and remains the only real node of accountability. When, therefore, in a case like the S.I.R., the Court takes highly consequential actions over a period of months, but there is no reasoned judgment (as everything is happening through interim orders), it also insulates itself from the accountability that comes with publicly and openly deciding the cases brought to it. This is of a piece with the opacity and non-transparency we discussed in the previous section, and both pertain to how the Court has, in effect, liberated itself from accountability. Judicial evasion is one of the most potent weapons through which it does so. 

      The other significant event which reflected on the direction of the Court under CJI Gavai’s tenure pertained to the “Vantara case.” In a hasty and irregular process, a bench of the Supreme Court – acting on two half-baked public interest petitions – set up a Special Investigation Team to investigate whether Vantara was involved in any illegality when it came to trafficking of wildlife and other related laws; equally hastily (in under three weeks, to be precise), the SIT filed a report clearing Vantara of all charges. At this point, it must be noted that there seems to be absolutely no principle or standard that determines when, and on what basis, the Supreme Court decides to set up these “SITs,” with all the far-reaching consequences that that entails; it apepars to depend entirely on the uncanalised discretion of the judges involved. 

      In any event, the Court then passed an extraordinary judgment, in which the most extraordinary part was two directions. 

      The first was an order that “no further complaint or proceedings based upon such same set of allegations shall be entertained before any judicial statutory or administrative forum.” In one stroke, the Court erased the entire criminal and civil law machinery that exists in the country, with respect to Vantara, on the strength of an SIT that it itself had constituted, and whose report has never been made public (of course, this is not the first time the Supreme Court has done this, and it remains now – as it remained then – unjustifiable).

      The second was a direction that “we may leave it open to the respondent – Vantara to pursue its remedies in accordance with law for the deletion of any offending publication or for any action against those responsible for the misinformation or for actions for defamation…”

      I have previously referred to this as the Supreme Court’s “Idi Amin jurisprudence.” One does not need to be a clairvoyant to read between the lines and understand this for what it is: a signal that it is open season on critics and dissidents. One does not need to be a clairvoyant because, again, one has seen this before: in cases involving Teesta Setalvad and Himanshu Kumar, where – utterly unconnected to the actual case before it – the Court makes these innocuous suggestions to prosecute or otherwise use the law against specific people. The continuation of this judicial vindictiveness under the tenure of CJI Gavai is disturbing. 

      As an aside, it is interesting to note that from start to finish, this case took forty-five days before it was closed, and all of that without the Court ever formally issuing notice on the petition. 

      What of other areas of constitutional law? Things largely stayed the same under CJI Gavai’s tenure as they had under his predecessors, which is not to say that things were good. Personal liberty continued to suffer. Political prisoners still languished in jail for years without trial, even as interminable hearings continued (and continue) before the Supreme Court. Challenges to administrative detention – such as that of the political leader Sonam Wangchuk – were kept pending with long dates until they became effectively infructuous. The Court continued to exhibit a peevish hostility towards the freedom of speech, berating one person, setting up an SIT (that SIT again!) to investigate the Facebook posts of another, and sanctioning continued police harassment by refusing to grant anticipatory bail to a third. The language of the Court continued to coarsen and cheapen towards some of the most vulnerable members of society, such as refugees and migrants. Indeed, while a dislike of refugees and migrants – and of the international legal framework that seeks to protect them from persecution – has long been part of the Supreme Court’s judgments, jibes and taunts from the bench towards the UN High Commission one day, and towards petitioners protesting the physical dropping of refugees into the ocean for deportation another day, marked a new low in the quality of the conversation that takes place within the halls of the nation’s highest constitutional court. While, naturally, one cannot hold the CJI directly responsible for specific oral remarks made by other judges, all in all, the Court’s drift from an institution that engages seriously and dispassionately with law and legal reasoning, to a forum where our society’s nastiest and most parochial sentiments now get a public airing from the bench, continued. 

      II. The Master of the Roster 

      It was in his actions as the Master of the Roster, however, that we saw CJI Gavai as the protagonist of his own “Choose Your Own Adventure Story.” The most prominent example of this, of course, was the manner in which he took up a Presidential Reference filed in the aftermath of the Court’s ruling (State of Tamil Nadu) that Governors and Presidents were bound by timelines to act on state legislative assembly bills. 

      In setting up a bench post haste, which then went on to render an opinion effectively overruling State of Tamil Nadu, CJI Gavai and his bench did something truly extraordinary: as I have noted at some length in this post, they created a new, intra-Court appellate jurisdiction which only one party can utilise: the central executive. After this Presidential Reference, whenever the central executive is unhappy with a case that they have lost before the Supreme Court, they can simply file a Presidential Reference (a privilege which no other litigant in the country has), and get a second bite of the cherry before a five-judge bench, where the previous decision is entirely ignored and the case considered afresh. 

      Yes, we have had Chief Justices previously who have administered what I have referred to as an “executive court”: but this is the first time we have had a Chief Justice who has formally created a privilege reserved only for the executive, and entrenched it within the Constitution. Perhaps we now need a term that goes beyond the “executive court” (see the Conclusion for a suggestion). 

      Other than the Presidential Reference case, CJI Gavai repeatedly manipulated the roster in other cases to overturn precedent that he didn’t like. A two-judge bench of the Supreme Court had passed a judgment holding that post-facto environmental clearances to big corporations involved in extractive activities (such as mining etc) were illegal. Against this judgment, a review petition was filed. A review is an extraordinary remedy requires three things: one – that it be heard by the same bench that passed the original judgment; two – that it be decided in circulation in chambers unless there is a specific, stated reason to hear it in open court; and three – that a review can succeed only if the original judgment has an error on the face of the record (and not simply because another judge believes the law should be decided differently). 

      After passing the environmental clearances judgment, one of the judges – Oka J. – retired. CJI Gavai then set up a review bench that had three judges instead of two: the other judge on the original bench (Bhuyan J), the CJI himself, and a third judge (Chandran J). Without any circulation or reason why the case was being reviewed in open Court, this three judge bench now passed a judgment “recalling” the original judgment (needless to say, Bhuyan J dissented). 

      And needless to say, the “recall” judgment made no effort to demonstrate how Oka J.’s original judgment was so erroneous that it had to be “recalled.” The consequences, as many have noted, are going to be devastating for the environment; but the consequences are also devastating for the Supreme Court as an institution, because clearly, even final judgments of the Court can be very quickly overturned if the Chief Justice does not like them. 

      These examples would have been bad enough on their own, but there are a litany of such cases: from the Court’s Constitution Bench decision on the district judiciary, to the notorious matter involving stray dogs, to an unreasoned stay upon another judge’s order that somehow came to him in “review” in open court after a split bench (see above), CJI Gavai’s tenure was marked by the manipulation of the master of the roster’s powers to not just assign and list cases, but to actively interfere with decided (or pending) cases (for a more detailed analysis, see Indira Jaising’s piece here). The upshot of this – other than the gutting of the doctrine of precedent and the rule of law – is, of course, a diminished institution. It was bad enough that as “first among equals,” the Chief Justice had vast administrative powers of assignment and listing, which could influence outcomes. This influence, however, was indirect; under CJI Gavai’s tenure, this became direct interference, thus effectively rendering the Office of the Chief Justice far more than “first among equals”, and in every way, a superior among subordinates.

      Or, a chief in his chiefdom. 

      C. Constitutional Judge

        What of CJI Gavai’s time as a constitutional judge during this six-month tenure? We do have some judgments involving the district judiciary (as noted above) and tribunals, but at least as far as substantive constitutional law goes, there is only the Presidential Reference (which was authored per curiam). On substantive issues, the Presidential Reference “opinion” has been critiqued elsewhere on this blog; here, in addition to what I have noted above, I will add one further point, which is in line with the theme of this post, and of CJI Gavai’s tenure. The effect of the Presidential Reference Opinion – other than gutting Indian federalism – is to place elected state governments not only at the mercy of Governors and Presidents, but – even worse – at the mercy of the Supreme Court

        At its heart, the Presidential Reference Opinion is an expansion of unprincipled and discretionary judicial power, because what it does is to give to itself the final call on whether there has been “enough” delay on part of Governors and Presidents to warrant judicial intervention – but, unlike the State of Tamil Nadu – with absolutely no standards to check the arbitrariness of the Court in making that determination. Henceforth, therefore, for every bill that a Governor pocket-vetoes, states will have to come to the Supreme Court and plead for intervention so that they can go about their business of legislation. And we have seen where this road goes before, with the 10th Schedule and the anti-defection cases: the Court turns into a political actor and a power-broker between the centre and the states, tipping the scales this way and that. And, just like in the 10th Schedule cases, because there is no law and there are no standards, the Court will not explain itself even as it makes hugely consequential decisions. Just like in 2022, the Maharashtra government flipped with two one-line interim orders of the Court, states will be informed through one-line diktats whether they can legislate or not. 

        It would be remiss not to conclude with a brief coda of the hypocrisy involved in the Presidential Reference Opinion. Forget, for a moment, the hypocrisy of the Indian Supreme Court invoking the “separation of powers” in 2025, given its sweeping role in Indian public life. Forget the hypocrisy of the Court suddenly becoming chary of invoking Article 142 of the Constitution in this case. Forget the sudden, odd fidelity to text when each of the judges on the bench owe their own position to the Collegium system, which was invented out of whole cloth. The fact is that the same CJI Gavai who headed this bench saw no problem with imposing a three-month deadline on an Assembly Speaker to decide on disqualification petitions. Walt Whitman may contain contradictions. A Chief Justice does not have that luxury. Or so one would have thought. 

        D. Conclusion 

          Through his tenure, the Chief Justice liked to speak about his achievements. He praised himself for having stopped “bulldozer raj” and restored the rule of law, even as the bulldozers continued to demolish homes. And on his farewell day, he praised himself for not having used a single foreign precedent in the Presidential Reference Opinion. He called it “swadeshi jurisprudence.” 

          But what exactly, one might ask, is “swadeshi” about a judgment written in English, relying primarily on the English common law of precedent, considering the powers of the English-invented institution of the “Governor,” and propounding a doctrine of executive deference that was originally propounded by the German jurist, Carl Schmitt? 

          Indeed, the greatest exponent of that same English language, one Shakespeare, famously asked us: what’s in a name? What we call the executive court, by any other name, would judge the same (or not judge, if it is performing judicial evasion). If CJI Gavai’s parting achievement is to bring together all these diverse elements of the executive court – judicial evasion, non-transparent appointments, manipulation of the roster, callousness towards personal liberty, “Idi Amin rulings,” coarse judicial language, and executive deference before all else (including federalism) – within one label, that of swadeshi jurisprudence, then perhaps we should be grateful to him for that. After all, brevity is the soul of wit, Shakespeare also said. 

          We look forward to the further evolution of this swadeshi jurisprudence. 

          Illegal…For Now: the Supreme Court on the Maintainability of Special Presidential Reference No. 1 of 2025 [Guest Post]

          [This is a guest post by Karthik Ravichandran.]


          A five-judge bench of the Supreme Court recently gave judgement in an Article 143 Reference in which the President sought answers to fourteen constitutional questions concerning assent of bills by State Governors and the President (“Re Assent to Bills”). As widely noted, many of these questions had been explicitly answered in State of Tamil Nadu v. Governor of Tamil Nadu (“Tamil Nadu”), barely a week before the Reference was made. The Reference itself was obviously a response to the Tamil Nadu judgement.

          In Tamil Nadu, a Division Bench of the Court had laid down rules for Governors acting under Article 200. These had to do with, among other things, the width of the Governor’s powers in relation to withholding assent to a State bill, the Governor’s discretion in doing so against the wishes of the Cabinet of Ministers (it had none), the justiciability of the Governor’s actions under Article 200 (widely justiciable under various grounds), and timelines under which the Governor was generally expected to exercise Article 200 powers (specific timelines were stipulated for the various options, with exceptions). Tamil Nadu arose out of a situation in which the Governor sat on bills for months, initially withholding assent without sending them back to the legislature, then sending them to the legislature after much coaxing by the Supreme Court, and then sitting on the bills again after they were returned. It is hard to argue that the Governor wasn’t essentially exercising a pocket veto on the State’s legislative agenda – something the Tamil Nadu Court seems to have acknowledged. Aside from interpreting Article 200, the Court decided that in this particular instance, the Governor was in gross violation of the timelines that it had just laid down. It used Article 142 to deem assent to the bills that the Governor had been illegally sitting on. A judicially deemed assent of a legislative act was a first in India’s history, and was subject to much controversy.

          We will refrain from delving into the merits of the Tamil Nadu judgement, and restrict ourselves to the manner in which the Court answers questions on the maintainability of this Reference. An earlier post on this blog rightly argued that the Reference was clearly not maintainable in as much as it sought to address questions that had been explicitly addressed in Tamil Nadu, considering that it would basically amount to entertaining a second round of litigation where the executive had lost in the first round. In other words, it would amount to creating a special de novo review/appeals process for the Union Government, for judgements that went against it, thus by-passing the high thresholds of invoking the review and curative jurisdictions – what Gautam Bhatia calls “constitutional mulligan” (This author likes the term constitutional trial ball). Presidential References are thus, only meant to answer questions that the President could possibly not have known the answer to from the Court’s jurisprudence, and not as a means to overturn settled law. This argument finds explicit support in the Supreme Court’s judgement in the Cauvery Reference (“Cauvery”). The In Re Assent to Bills Court disagrees.

          For some reason, the Court begins its refutation of this argument by distinguishing this Reference from the fifteen that have come before it, terming this one as a “functional reference” (paragraph 10). This is purportedly because it deals with the “day to day functioning of constitutional functionaries” and “strikes at the root of the continuation of our republican and democratic way”. It is unclear what this means, not in the least because one of the other References – all of which the Court lists out – is the Third Judges Case, which, one could say, did deal with the day to day functioning of powerful constitutional authorities, and was equally consequential to democracy. But more importantly, one wonders why the nature of the subject-matter of this Reference, notwithstanding the uniqueness that the Court ascribes to it, would have any bearing on its maintainability, if it doesn’t fulfil some basic requirements. That the Court appears to decide it does in this case, is emblematic of the way it deals with the entire maintainability question. This is basically laying the groundwork for ascribing a certain inevitability to answering these Reference questions – to the extent that one must do away with any principled engagement with Article 143. Having done that, the Court proceeds to actually deal with the maintainability arguments. 

          The Court begins by arguing that it is generally permissible for it to overrule its judgements even in its Article 143 jurisdiction. It begins substantiating this by extracting a passage from In Re: Special Courts Bill, 1978 (“Special Courts”), another Presidential Reference.

          …We are inclined to the view that though it is always open to this Court to re-examine the question already decided by it and to over-rule, if necessary the view earlier taken by it insofar as all other courts in the territory of India are concerned they ought to be bound by the view expressed by this court even in the exercise of its advisory jurisdiction under art. 143(1) of the Constitution.

          Now this is barely helping the point the Court is trying to make. In Special Courts, the Supreme Court was contemplating, after substantively considering the questions of that Reference, whether an Article 143 opinion was binding on other Courts. The Court ended up deferring that question for a later day but expressed its inclination to answer in the affirmative. But, more importantly, in doing so, the Court observed, in the extracted paragraph, that only the Supreme Court has the power to overrule its judgements. This was merely a reiteration of the Court’s general power under Article 141, and not a comment on whether there is an exception in the Court’s overruling powers for when it is hearing a case in its Article 143 jurisdiction. Now, the Court correctly points out that this same passage has been used by it earlier, in the Natural Resources Allocation Reference (“2G Reference”), to explicitly say that the Court does have power to overrule judgements in its Article 143 jurisdiction. Be that as it may, this reliance, as I have pointed out above, is incorrect.

          Let us now deal with the 2G Reference. Other than invoking that one paragraph from Special Courts, the 2G Reference Court also makes some points on first principles. It engages with Cauvery, which recall, explicitly states the opposite – that the Court cannot overrule previous judgements in its Article 143 jurisdiction, because doing so would amount to sitting in a special appeal over its own decision. In 2G Reference, the Court says that when Cauvery refers to the “overruling” of judgements, it only refers to the specific inter partes dispute or the lis in the judgement being overruled. Overruling the judgement on the lis is not permissible except in the Court’s review and curative jurisdictions, as opposed to overruling a “view of the law” which the Court says it is entitled to do under Article 143, when it is read with Article 141. And surely enough, the Re Assent to Bills Court reiterates this point.

          This is problematic for a variety of reasons. Firstly, this is a misreading of Cauvery, which makes repeated references to the overruling of “questions of law”, explicitly stating that only those “questions of law” that have not been decided by the Court are open for the President to refer under Article 143 (paragraph 85). The phrase, “questions of law” is borrowed from the text of Article 143 itself, where there is a distinction drawn between questions of law and fact. Now, from what this author understands, the entire concept of stare decisis which is the fundamental basis for the common law system, is based on the idea that a principle laid down in a particular case, even if it is in the context of a particular set of facts, is applicable to similarly placed parties in the future. That principle answers what is known as a question of law, and establishes a position of law that is supposed to influence future behaviour, and governs future disputes. Now, under this formulation, in what universe are the propositions laid down in Tamil Nadu, not answering questions of law? Under Tamil Nadu, the law would, for instance, require Governors to decide on withholding assent to bills within a stipulated time period. After Re Assent to Bills, the law does not do that. This is in the most straightforward sense, an overruling in a question of law, and contrary to Cauvery.

          Secondly, let us keep Cauvery aside for a while, and entertain on first principles the idea that the only kind of overruling not permitted under Article 143 is an overruling in terms of the inter partes lis. Under this formulation, is a Presidential Reference containing the question “Can Parliament amend the basic structure of the constitution” maintainable only because it is not framed as, “Are the 24th, 25th and 29th Constitutional Amendments valid?” Is any Presidential Reference on any Constitutional question having wide implications, previously decided by the Court, maintainable as long as the Court does not re-visit the facts out of which the question arose the last time? The Court’s answer to this question seems to be in the affirmative. In paragraph 26 of the judgement, the Court says that because the Reference does not require it to “vacate, amend or modify the relief” granted in Tamil Nadu, it is maintainable as regards questions surrounding “propositions of law” decided in that case.

          This is untenable and upends the very fabric of law that the Supreme Court declares, especially when the Union Government is involved. The Union Government’s interest in a question of law does not extinguish as soon as it loses or wins a particular case implicating that question of law. This is even more so in cases involving disputes between itself and the States, namely matters of federalism, which, as we have been told repeatedly, is a basic feature of the Constitution. For instance, the Union’s interest in the question of law surrounding the Article 131 suit between the State of Kerala and the Union, goes beyond the suit. A victory for the Union enables it to impose fiscal borrowing ceilings on every other State, whereas a loss does the opposite. But this is meaningless if the Union loses and is then enabled to reverse the position of law through a Presidential Reference, as long as the relief granted to the State of Kerala is not reversed. This is equally true in matters concerning individual fundamental rights, which is why it should be extremely concerning that Courts are, for instance, prohibiting the use of bail judgements as legal precedent.

          Lastly, consider this. If a question of law being overruled in a Presidential Reference was previously answered in a judgement involving the Union, the idea that the Court is not entering into the inter-partes dispute is, at any rate, only half true, especially when the interests of the Union in both proceedings are identical. And this is precisely where the sleight of hand is happening – because what the Court fails to mention in all of this, is that one party is omnipresent, namely the Union. The Union is involved in every dispute involving federalism, and it is only the Union that is entitled to get its interests reconsidered through a Presidential Reference.

          Conclusion

          If you believe the Constitution is a terrain of contestation of power, Re Assent to Bills, by simply admitting the Reference,helps further the tilt in the power-balance in the Union executive’s favour – and this is on all the axes of power the Union executive occupies, given that an Article 143 Reference can be brought in with regard to pretty much any question of law. And this is not even counting the substantive part of the Reference where the Court reverses some progressive ideas from the Tamil Nadu judgement that helped reign in Governors’ impunity in dealing with State bills.

          One final observation. The Indian Supreme Court is known for its “illegal but what to do now?” jurisprudence whereby it prospectively lays down propositions of law in matters where it either refuses to use said propositions to do justice to the parties before it, or places itself in positions where it is unable to do so. These days, the Court seems to have devised the mirror image to this, whereby it does justice to parties before it, but then revokes the benefit of that justice to future litigants. We can call this “illegal…for now” jurisprudence. And like its mirror image, “illegal…for now” jurisprudence is the perfect ode to the discretion-based law that the Supreme Court of India has come to champion in recent times.

          Unconstitutional by Design: Prior Restraint and Compelled Speech in the 2025 IT Rules Draft Amendments

          [This is a guest post by Rudraksh Lakra.]


          On October 22, 2025, the Ministry of Electronics and Information Technology proposed amendments (“Draft Amendments”) to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021.  The stated objective of the Draft Amendments is to address the “growing misuse of synthetically generated information including deepfakes, misinformation, and other unlawful content” and to further the goal of creating an “Open, Safe, Trusted, and Accountable Internet.”

          While the Draft Amendments have generated extensive policy and legal debate, two significant free speech concerns have received insufficient attention: (i) the risk of prior restraint, and (ii) the imposition of compelled speech/must-carry obligations on intermediaries. This article argues that, in their current form, the Draft Amendments are unconstitutional because they introduce disproportionate compelled speech requirements and create an unlawful system of prior restraint.

          Prior Restraint

          Draft Amendments, Rule 4(1A)(a) requires users to declare whether the information they upload on a Significant Social Media Intermediary (“SSMI”) platform is synthetically generated information. “Synthetically generated information” is defined as “information which is artificially or algorithmically created, generated, modified, or altered… in a manner that such information reasonably appears to be authentic or true.”

          Further, Rule 4(1A)(b)-(c) requires SSMIs to deploy “reasonable and appropriate technical measures,” including automated tools, to verify user declarations, and where it is confirmed that the “information is synthetically generated, ensure that the same is clearly and prominently displayed with an appropriate label or notice, indicating that the content is synthetically generated.”  Now Draft Amendments, Rule 4(1A) chapeau provides that a significant social media intermediary “shall, prior to such display, uploading, or publication….” One reasonable interpretation of this language is that content cannot be posted unless it has first been verified by the intermediary. The accompanying explanation reinforces this reading by requiring intermediaries to “verify the correctness of user declarations and ensure that no synthetically generated information is published without such declaration or label.” Interpreted in this manner, the provision would amount to a form of prior restraint on free speech.

          Prior restraint refers to regulatory mechanisms that prevent speech or publication before it occurs, rather than imposing accountability after the expression has entered the public domain. Prior restraint is treated as uniquely suspect because it suppresses speech at its source, denying both the speaker the opportunity to express and the public the opportunity to receive information. Unlike post-facto regulation, which allows speech to circulate and then subjects it to accountability. Prior restraint presumes speech to be harmful unless permitted, thereby inverting the constitutional default in favour of free expression. Such systems tend to create a chilling effect, as speakers may choose silence over the uncertainty, delay, or scrutiny involved in obtaining prior approval. They also typically vest discretionary power in administrative or executive actors, which makes it easier to circumvent constitutional guarantees and heightens the risk of selective or politically motivated enforcement.

          In India, for these reasons, the Supreme Court has struck down prior restraints in Romesh Thappar v. State of Madras (1950) (“Romesh Thappar”), Brij Bhushan v. State of Delhi (1950) (“Brij Bhushan”), and R. Rajagopal v. State of Tamil Nadu (1995) (“R. Rajagopal”). In Brij Bhushan, the Court cited Blackstone for the proposition that “the liberty of the press consists in laying no previous restraint upon publications” and that “every free man has an undoubted right to say what sentiments he pleases before the public; to forbid this is to destroy the freedom of the press.” In R. Rajagopal, after relying on extensive American jurisprudence, where there is a strong presumption against prior restraint, the Court held that pre-publication censorship of material allegedly defamatory of the State or its officials was unconstitutional. Although the Supreme Court in K.A. Abbas v. Union of India(1970) (“K.A. Abbas”) upheld pre-censorship in the context of cinema; that decision was rooted in a parens patriae approach to film as a distinct medium and did not engage with the earlier rulings in Romesh Thappar and Brij Bhushan. Even in Abbas, however, the Court acknowledged that “freedom of speech and expression admits of extremely narrow restraints in cases of clear and present danger,” and further noted that censorship must be “precise.” Taken together, this line of cases establishes that prior restraint is presumptively unconstitutional and can only be justified under exceptional circumstances. Further, any such restriction must meet the requirement of demonstrating a proximate connection between the expression and the likelihood of unlawful action.

          Draft Amendments, Rule 4(1A) introduces a mandatory pre-verification requirement, and this mandate is unconstitutional under the proportionality standard because it is a disproportionate restriction. First, the labelling requirement envisaged under the Draft Amendments does not have a proximate connection to the stated aim. Labelling does not identify harmful content; it merely indicates whether a piece of content is synthetically generated information or not. The label itself does not assess illegality, misinformation, harm, or user intent. Second, the efficacy of such a labelling system is highly doubtful. Given the technical and operational limitations of automated content moderation tools, especially in non-Western contexts, non-Latin scripts, and in situations where the assessment of content requires meaningful context-based analysis. Third, a pre-verification model risks substantially altering how users engage with social media platforms, where spontaneity and immediacy are defining features. If adding a label to AI-generated content results in delays before posting, users may be discouraged from disclosing that their content is AI-generated, even when it is. This would defeat the purported objective of the draft Rule.

          Thus, even assuming the legitimacy of the State’s stated objective, the pre-verification mandate fails the balancing stage of the proportionality test: it is both operationally and technically unworkable, and it imposes a significant and unjustified burden on free expression. SSMIs now function as the modern equivalent of the public sphere, providing the primary space for political debate, social interaction, and the exchange of ideas. Thus, they are no longer merely private platforms but have become essential infrastructure through which contemporary democratic discourse is mediated. Imposing a verification burden on these platforms converts them into pre-publication censors. This concern is heightened by the fact that the definition of “synthetically generated information” in the Draft Amendments lacks objective criteria, inviting arbitrary and ad hoc decision-making under threat of liability. Further, the verification process is not accompanied by any meaningful procedural safeguards or due process guarantees.

          Compelled Speech/Must Carry Speech

          Compelled speech arises when the State requires an individual or entity to express, endorse, or disseminate a message they would not voluntarily convey. Although free speech protections are often understood in terms of restrictions on expression, compelled speech is equally problematic because the constitutional right to free speech necessarily includes the right not to speak. The Supreme Court in Union of India v. Motion Picture Association (1999) (“Motion Picture Association”) held that compelled speech constitutes a violation of free speech and must therefore be justified under Article 19(2) of the Constitution. It further noted that: “[i]f a ‘must carry’ provision furthers informed decision-making which is the essence of the fight to free speech and expression, it will not amount to any violation of the fundamental freedom of speech and expression.” However, the Court decision failed to mention that apart from pursuing a legitimate aim of enabling informed decision-making, the measure itself must be reasonably designed. In the contemporary constitutional context, the requirement of reasonableness under Article 19(2) translates into satisfying the proportionality principle [Modern Dental College v. MP(2016), Puttaswamy I (2017), Puttaswamy II (2018) and Anuradha Bhasin v. UoI (2020)].

          The Draft Amendments introduce three compelled-speech obligations. First, Rule 4(1A) requires users to declare whether their content is synthetically generated. Second, it mandates that SSMIs verify such declarations and attach corresponding labels. Third, Rule 3(3) obligates any intermediary that “enables, permits, or facilitates the creation, generation, modification or alteration” of synthetically generated information to apply a label/watermarking to such content. These requirements are akin to disclosure obligations mandating ingredient labels on food products or health warnings on cigarette packets. These were noted as examples of legitimate forms of compelled speech by the Supreme Court in Motion Picture Association. In the present context, the stated aim of enabling users to distinguish between synthetic and non-synthetic content is likewise a valid and important objective, particularly given the risks posed by deepfakes and misinformation. However, as I demonstrate below, the specific labelling measures imposed by the Draft Amendments are overbroad and unduly burdensome, as they fail the necessity and balancing stages of the proportionality test.

          Under Rule 3(3), the watermark has to cover “at least ten percent of the surface area of the visual display or, in the case of audio content, during the initial ten percent of its duration.” The use of the terms “visibly displayed” or “made audible in a prominent manner” makes clear that the type of watermarking envisaged is a visible label. However, visible watermarking is one of the weakest forms of content authentication, since such marks can be removed or obscured using basic editing tools, defeating the regulatory purpose. By contrast, robust authentication systems rely on imperceptible or embedded watermarks that remain detectable even after editing or compression. The Rule is also problematic because requiring a watermark to occupy ten percent of a visual display risks materially degrading the aesthetic, immersive, or expressive quality of content for end users. This burden becomes even more disproportionate given the provision’s sweeping scope, which applies not only to fully AI-generated content but also to trivial or routine edits such as cropping, colour adjustment, or the application of filters that do not meaningfully alter the semantic content of the underlying work.

          Comparable regulatory models take a far more calibrated approach. For instance, the European Union Artificial Intelligence Act (2024) adopts a narrower and more technically informed approach: it exempts AI systems used for ordinary editing functions and requires watermarking duties to take into account “the specificities and limitations of various types of content,” cost implications, and the state of available technology [Article 50(2)]. Accordingly, the provision is unconstitutional because it imposes a burdensome and overbroad compelled-speech requirement while relying on a technically deficient mechanism that neither reliably distinguishes between synthetic and non-synthetic content nor meaningfully advances the stated regulatory objective.

          Rule 4(1A) imposes an even more disproportionate compelled-speech obligation. It requires users to declare whether their content is synthetically generated and obligates SSMIs to verify and label such content. By comparison, the watermarking requirement in Rule 3(3), if properly designed, is a more effective and proportionate method of achieving the same regulatory objective. Watermarking allows the developer or deployer of a generative tool to embed a secure identifier directly into the content at the point of creation, enabling later authentication. In contrast, Rule 4(1A) requires SSMIs to rely on automated systems to infer whether content is synthetically generated. As such tools operate on probabilistic models, they cannot offer definitive authenticity checks; they merely assign likelihood scores that the content is AI-generated. These systems are also prone to errors, leading to compliance uncertainty and operational strain. To compensate for these inaccuracies, intermediaries may need to introduce additional layers of manual review, increasing costs, slowing content flows, and expanding exposure to liability for both under- and over-compliance. The Rule’s scope is also overinclusive: it mandates labelling for all AI-generated content without regard to context, harm, or risk. Such a blanket obligation not only exceeds what is necessary to address genuine harms but also compels speech in circumstances where the State lacks a legitimate regulatory interest. Taken together, the technical infeasibility of accurate detection, the heavy compliance and resource burdens imposed on intermediaries, and the absence of reasonable tailoring render Rule 4(1A) an unconstitutional form of compelled speech under settled free-speech jurisprudence.