“Constitutional guarantees cannot be subject to the vicissitudes of technology.” (Chandrachud J., dissenting, paragraph 269)

As we discussed in the last post, the Majority judgment in the Aadhaar Case is premised upon a series of factual assumptions, which are either unsubstantiated, or lifted from a Power-Point presentation given by the UIDAI Chairperson in the Court. The Majority agrees, for instance, that profiling is bad, and surveillance unconstitutional – but finds, on fact, that the Aadhaar framework does not permit either. The Majority agrees that data minimisation is a constitutional principle, but finds, on fact, that Aadhaar does collect minimal data. And the Majority agrees that the legal standard is that of proportionality, but finds that because of its “uniqueness”, biometric authentication successfully targets deserving beneficiaries, and that therefore, is proportionate.

For this reason, while reading Justice Chandrachud’s dissenting opinion, there is a sense that the two judgments disagree not only on legal standards (which they do), but also, on something far more basic: they disagree about the very state of the world within which Aadhaar operates.

Why does this matter? It matters because Judges are entitled to declare the law – and indeed, bring it into being by declaring it – but facts have an independent existence. For example, one may take sides on whether the Majority or Justice Chandrachud is correct in holding that Section 59 of the Aadhaar Act validates past action, while acknowledging that the Majority – by virtue of being the Majority – has laid down the law. However, if the Majority and Justice Chandrachud disagree on the uniqueness of biometrics, or on the existence of exclusion, then – quite simply – one of them is right, and the other one is wrong. And if the Majority is wrong on facts, then serious questions must be asked about the sustainability of that judgment.

Overarching Assumption: The Uniqueness of Biometrics

As we saw, the factual foundation of the Majority Judgment is that biometrics are unique. This foundation is at the heart of the Majority’s decision to uphold Section 7, as well as the mandatory Aadhaar-PAN linkage. And it is with this foundational assumption that Chandrachud J. takes issue. In paragraph 132, he notes that “errors will inevitably occur” (with biometric use). In paragraph 150, he puts the point in a stronger way: “The uniqueness of a fingerprint in forensic science remains an assumption without watertight proof.” Unlike the Majority, he then goes on to substantiate this claim, citing scholarly books (Fn. 154 & para 260) and the text of the Aadhaar Act itself, which envisages updation of biometrics. This becomes crucial in the latter part of the judgment dealing with exclusion, where (as we shall see) both authentication failures and the existence of false positives and false negatives are a crucial reason for his finding of unconstitutionality.

This foundational factual disagreement between the Majority and Justice Chandrachud is important – and virtually determinative – to the outcome of the case. Recall that the Majority elects to pitch its case very high: not merely that a biometric database is good or efficient, but that it is flawless: i.e., there cannot be duplicates. The Majority does this because it makes the rest of the case very easy. If Aadhaar is truly unique, then ipso facto, it efficiently targets beneficiaries (and so, Section 7 is upheld) as well as fake PAN Cards (and so, S. 139AA is upheld); and furthermore, “unproven” stories of exclusion cannot be taken seriously (and, in any event, are being dealt with by a Circular). There is no need to engage in a messy proportionality analysis about whether biometric authentication actually accomplishes what the State claims it does, and whether it is indeed the least restrictive way of accomplishing it. However, by stark contrast, Chandrachud J.’s finding that biometrics can be erroneous – as we shall see – opens his judgment up to a large number of issues: do errors disproportionately affect the most vulnerable? Shouldn’t failure be forestalled, rather than compensated? And so on.


Here again, Chandrachud J.’s disagreement with the Majority starts in factual analysis. Chandrachud J. notes that profiling and surveillance is possible under the existing Aadhaar framework. This is because, in his view, according to Regulation 17, requesting entitles can store biometric information for a temporary period (paragraph 126), that through the IP address, meta data can be used to track location and profile (para 227), that there is the access third-party vendors to the database, and – most crucially – that linking of databases can take place:

The risks which the use of Aadhaar “for any purpose” carries is that when it is linked with different databases (managed by the State or by private entities), the Aadhaar number becomes the central unifying feature that connects the cell phone with geo-location data, one’s presence and movement with a bank account and income tax returns, food and lifestyle consumption with medical records. This starts a “causal link” between information which was usually unconnected and was considered trivial. Thus, linking Aadhaar with different databases carries the potential of being profiled into a system, which could be used for commercial purposes. It also carries the capability of influencing the behavioural patterns of individuals, by affecting their privacy and liberty. Profiling individuals could be used to create co-relations between human lives, which are generally unconnected. (paragraph 244)

In addition:

When Aadhaar is seeded into every database, it becomes a bridge across discreet data silos, which allows anyone with access to this information to re-construct a profile of an individual’s life. It must be noted while Section 2(k) of the Aadhaar Act excludes storage of individual information related to race, religion, caste, tribe, ethnicity, language, income or medical history into CIDR, the mandatory linking of Aadhaar with various schemes allows the same result in effect. For instance, when an individual from a particular caste engaged in manual scavenging is rescued and in order to take benefit of rehabilitation schemes, she/he has to link the Aadhaar number with the scheme, the effect is that a profile as that of a person engaged in manual scavenging is created in the scheme database. The stigma of being a manual scavenger gets permanently fixed to her/his identity. What the Aadhaar Act seeks to exclude specifically is done in effect by the mandatory linking of Aadhaar numbers with different databases, under cover of the delivery of benefits and services. (paragraph 247)

On every point, therefore, there is a direct factual clash between the Majority and Chandrachud J. Recall that the Majority categorically says that merging of data silos cannot happen, that that is a fundamental reason why profiling and surveillance is impossible. Chandrachud J., on the other hand, makes it clear that the seeding of Aadhaar across databases serves to break the silos. As I had pointed out above, this is not a matter of different legal interpretation, with the Majority’s view being “correct” simply because it is the Majority. Here, one of them is right and one of them is wrong.


Chandrachud J.’s disagreement with the Majority on this is legal in character. Recall that the Majority undertakes the proportionality test by diminishing our privacy interest in our bodily characteristics, and devaluing the importance of biometric details (fingerprints or iris scans). By contrast, Chandrachud J. holds that it our privacy interests in our biometric details is high: both from an informational self-determination point of view, as well as from a bodily integrity and physical safety point of view (paragraph 125). In particular, Chandrachud J. avoids two pitfalls that the Majority falls into: that is, mixing up “minimal information (collected)” with “minimal interference with privacy”, and applying the American “reasonable expectation” standard. Ultimately, however, the difference is one of framing: the Majority uses the fact that biometric details are given frequently and for a multiplicity of purposes, to argue that we don’t have a heightened privacy interest in them. For Chandrachud J., however, that is irrelevant: what is relevant is that a “carefully designed” biometric system may nonetheless preserve privacy (such as, for instance, ensuring anonymity), and that therefore, that is the standard we must measure Aadhaar against. (paragraph 127) And within this framework, he finds that the absence of consent within the Act, the extent of information disclosed, the expansive scope of the term “biometrics”, the burden placed upon the individual to update her own biometrics, and lack of access to the record, cumulatively constitute a serious infringement of privacy.

Section 7 and Proportionality 

Chandrachud J. accepts that the Aadhaar Act – and Section 7 – are designed to fulfil a “legitimate State purpose” – that is, making the welfare delivery system better and more effective. For him, however, the Programme fails on the proportionality prong. As he observes:

The test of proportionality stipulates that the nature and extent of the State’s interference with the exercise of a right (in this case, the rights to privacy, dignity, choice, and access to basic entitlements) must be proportionate to the goal it seeks to achieve (in this case, purported plugging of welfare leakage and better targeting). (para 198)

This paragraph reflects three crucial differences in the way that the Majority and Justice Chandrachud approach proportionality in this case. First – as we have seen above – the Majority holds that the invasion of privacy and dignity is minimal (and it doesn’t even consider the issue of choice), thus guaranteeing a very low threshold of justification to the State. Secondly, having made the assumption that biometric authentication is flawless, the Majority cannot – and does not – admit that its use can actually impede access to basic entitlements. In the Majority’s approach, therefore, the question of entitlements comes on the other side of the justificatory ledger (i.e., the State promoting access to entitlements through Aadhaar). And thirdly, the factual assumption of accurate targeting and improvement of welfare delivery drive the Majority into a pre-decided “balancing” between minimal impairment of privacy and significant plugging of welfare leaks. Chandrachud J., however, remains more circumspect: he carefully notes that the plugging of welfare leakage and better targeting is only “purported.”

The word “purported” informs Chandrachud J.’s proportionality analysis. As he notes, this is not a domain where the Court has to be excessively deferential to the State’s assertions. This is especially true because a nation-wide biometric programme is not merely one of those initiatives that applies to a one-time transaction. Rather:

… by collecting identity information, the Aadhaar program treats every citizen as a potential criminal without even requiring the State to draw a reasonable belief that a citizen might be perpetrating a crime or an identity fraud. When the State is not required to have a reasonable belief and judicial determination to this effect, a program like Aadhaar, which infringes on the justifiable expectations of privacy of citizens flowing from the Constitution, is completely disproportionate to the objective sought to be achieved by the State. (paragraph 217)

In this context, the lack of verification mechanisms by UIDAU, no proper exemption handling process (if authentication fails), the lack of accountability mechanism with respect to the UIDAI (para 235), the absence of an overarching regulatory framework, and the vague and unbridled nature of Section 7, which allows it to interpreted in an open-ended way, without checking whether each separate use violates the proportionality standard (paragraph 248), all militate against a finding of proportionality. And the position is sealed when Chandrachud J. observes that:  “the state has failed to demonstrate that a less intrusive measure other than biometric authentication will not subserve its purposes.”

This time as well, the difference between the Majority and the Dissent is purely legal: recall that the Majority had argued that the Petitioners had failed to demonstrate an alternative; Chandrachud J., however, reverses the legal burden: it is for the State to demonstrate that there is no feasible alternative, since it is the State, after all, that is infringing my rights. As should be obvious, in cases where there is no evidence on other side, this finding of burden changes everything.

The Argument from Inequality

Chandrachud J’s acknowledgment that biometric authentication is error-strewn (something that the Majority refuses to acknowledge) leads him to study the exclusion issue carefully (which the Majority dismisses in a line). Exclusion is directly linked with discrimination, because – as he notes – “exclusion as a consequence of biometric devices has a disproportionate impact on the lives of the marginalized and poor.” (para 253) He deals in some detail with the work of Virgnia Eubanks, who – using the concept of the “digital poorhouse” – has demonstrated that the discriminatory effects of technological solutions are inevitably visited upon the most vulnerable. Consequently, Chandrachud J. holds that “the fate of individuals cannot be left to the vulnerabilities of technological algorithms or devices.” (para 262).

He then substantiates this by looking at the Economic Survey of 2016-17, government reports involving pilot projects in Andhra Pradesh, and at the work of grassroots scholars such as Jean Dreze and Reetika Khera, to show that the exclusion percentages have been substantial (paragraphs 263 – 268).  Now recall that the Majority dealt with this issue in a line, noting that the Attorney-General had a made a statement that people would be allowed to use another identification, and that there was also a Circular providing for this. In a following post, I will explain quite how bizarre this approach is, which upholds an unconstitutional statute on the basis of a promise to interpret it fairly; for now, however, Chandrachud J. has an answer himself, when he notes:

Technological error would result in authentication failures. The concerns raised by UIDAI ought to have been resolved before the implementation of the Aadhaar project. Poor connectivity in rural India was a major concern. The majority of the Indian population lives in rural areas. Even a small percentage of error results in a population of crores being affected. Denial of subsidies and benefits to them due to the infirmities of biometric technology is a threat to good governance and social parity. (paragraph 262)

This is a crucial observation. What Chandrachud J. is saying is that once it is established that exclusions will occur – leading to deprivation of rights – then the State bears the burden of first resolving these issues before rolling out the project. Once again, the contrast with the Majority could not be starker. The Majority notes that Aadhaar is an ongoing project, and “glitches” must be ironed out as they present themselves. To this, Chandrachud J. replies: “you cannot be ironing out the glitches when Articles 14 and 21 are at stake.” In other words, people – especially the most vulnerable – cannot be used as experimental subjects for improving the efficiency of technology, as that would violate every constitutional principle in the book:

No failure rate in the provision of social welfare benefits can be regarded as acceptable. Basic entitlements in matters such as foodgrain, can brook no error. To deny food is to lead a family to destitution, malnutrition and even death. (paragraph 263)

There is, of course, a bitter irony here. In the passive euthanasia judgment, it was the Chief Justice who had first articulated this principle, when he asked whether the individual was to be turned into some kind of a guinea pig for an experiment. Unfortunately, it would take only a few months for him to forget, and it would be left to a dissenting opinion to remember.

A Dissent for the Ages: Individual, State, Identity 

Justice Chandrachud’s dissenting judgment recognises what the Aadhaar case was truly about: it was the first time in its history that the Court was called upon to answer serious questions about the interface between technology, the relationship between individual and State, and the Indian Constitution. The judgment is shot through with a keen awareness of this fact. Indeed, in paragraph 3, Chandrachud J. notes that “our decision must address the dialogue between technology and power.”

These are not careless words. The Aadhaar case was all about the relationship between the individual and the State, and how technology was altering – and even potentially inverting – that relationship. It was about how power worked itself through technology, through algorithms, becoming the arbiter of peoples’ rights and entitlements. And it was, at its heart, a question about what our Constitution had to say about that.

This is something that the Majority, in its techno-utopian celebration of greater efficiency and unique identification, misses entirely: the other side of the story, the contrapuntal notes. It is the side that comes across with particular clarity at various points in the dissenting opinion: when Chandrachud J. discusses how unique data sets can lead to “perpetuating of pre-existing inequalities” (paragraph 10), or when he refuses to play-off civil rights and socio-economic rights against one another, or in his attention to how biometric systems are “most aggressively” tried out with welfare recipients (paragraph 120). And it comes across most vividly in a brief discussion about identification and identity, an issue that plagued the hearings throughout. Consider, for example, the dissent’s discussion of how the concepts of “identity” and “identification” are being merged with the advent and technology (and compare, once again, with the Majority’s celebration of a “unique identity”):

Identity includes the right to determine the forms through which identity is expressed and the right not to be identified. That concept is now “flipped” so that identification through identifiers becomes the only form of identity in the time of database governance. This involves a radical transformation in the position of the individual. (para 185)

A finding that immediately leads to the following conclusion:

The submission which has been urged on behalf of the petitioners is that an individual entitled to the protection of the freedoms and liberties guaranteed by Part III of the Constitution must have the ability to assert a choice of the means of identification for proving identity. Requiring an individual to prove identity on the basis of one mode alone will, it is submitted, violate the right of self-determination and free choice. (para 185)

In other words, in an age when identification has subsumed identity, the individual must be granted a choice in the means by which she elects to “identify” herself to the State. It is a simple enough concept, but radical in its application.

They say that Minerva’s Owl takes flight at dusk. And so, you may well ask: what is the point of this dissent, when the Supreme Court spent six years busily allowing Aadhaar to become a fait accompli, and then legitimised it through a Majority opinion anyway? To that, perhaps, there is only one answer: the great cases are always 4 – 1 in favour of the State.

Until the 1 is resurrected. And becomes the pathway for a future that is still struggling to be born.

(The writer assisted Mr. K.V. Viswanathan, senior counsel for one of the Petitioners challenging Aadhaar).