top of page
Image by Annie Spratt




Between Similarity and Congruity: Exploring the Tertium Quid of Relevance

*Harshit Goyal


Imagine that it’s the year 2050, and despite all the interpretative controversies, the government in Hart’s world has still put up the board saying, ‘No vehicles allowed in the park’. Further, suppose that this has led to a case in which the court is asked to decide whether an eco-friendly self-driving hover car is allowed inside the park or not. Since the law is vague, and as the technologically advanced nature of the issue has left the court without any direct precedent, the court finds deductive reasoning difficult and decides to use analogical reasoning.

But what should be the starting point of such reasoning? Since a hover car is comparable to many objects in different aspects, which similarities and dissimilarities should the judge focus on? What should the judge do when it resembles one vehicle that has been disallowed in the park and with another that has been allowed? The objective of this piece is to answer these questions by exploring the criteria of relevance in analogy.

This article has been divided into two parts for a systematic analysis. The first part will set the context by analyzing the need to discuss relevance. The second part will then discuss a three-stage process that can help judges gauge relevant similarities and dissimilarities.

Since this topic is wide, it is also important to mention what the piece does not cover. Admittedly, the article is not exhaustive; it does not—as no article can—cover all the different modes of interpretation. Legal interpretation and analogical reasoning can be approached in myriad ways. Instead, what the article aims to provide is a heuristic device to analyse different aspects involved in analogical reasoning. While this process may not be detached from judicial discretion or other forms of reasoning, it will offer a working methodology regarding how to approach this subject.

Similarity, Congruity, and the Need of Relevance

Analogical reasoning is usually defined as a process of comparing similarities between a settled and a contested situation to derive certain conclusion regarding the latter. But what kinds of similarities do we focus on in such comparison?

It cannot be just any similarity because every object is similar to every other object in infinite number of ways. It is also a scientifically proven fact that our brains find some pattern and similarity in all the situations that we encounter.   Hence, as similarities are inevitable, we need something more to find correct answers. 

Such similarity cannot also be congruity because: firstly, as argued by the prominent philosopher Wittgenstein, no two situations are completely congruent, and there exist at least some differences between the two; and secondly, even when such congruency is possible, such cases would most likely be settled outside the court, and it would not need to use analogical reasoning for the same.

Hence, we necessarily need a third criterion that exists between the scale of similarity simpliciter and complete congruency. This third element is the standard of ‘relevance’, which enables us to navigate through the infinite number of similarities and to reach a conclusion even when the facts situations are not completely congruent.

Various scholars have offered different methods of finding this standard of relevance. Rudolf Carnap suggests a probabilistic account, in which the higher the number of similarities, the more relevant is the analogue. This theory is not helpful as it begs the question regarding what similarities should be chosen. Without any standard of gauging materiality of similarities, it creates a risk of selecting characteristics that might be unhelpful in reaching the solution.

On the other hand, some philosophers are pessimistic about finding any such specific method for relevance and instead argue that relevance depends on the experience of a rational judge. Scholars like Frederick Schauer place excessive trust on the ability of the judges and argue that for a judge the ‘connections will jump out immediately’. As the subsequent sections will show, judges do resort to bad analogy. Hence, this theory is also fallacious. 

Lastly, one of the most prominent theories of relevance has been that suggested by Scott Brewer, who argues that an analogy must follow the following steps:

While it seems satisfactory, this theory also jumps the gun by not providing first standard of relevance. Since every fact situation will have innumerable characteristics, one is left with no guidance on what properties must be picked for comparison. Hence, while Brewer formulates that A will have (x, y, and z), in reality:

Hence, unless we have a way of determining which ‘n’ to pick, or which of ‘p, q, r, s, t’ is relevant, we cannot reach the step 2 of Brewer’s process.

Therefore, as the standard of relevance is important and as the prominent theories have failed to provide a satisfactory metric, we need another standard. The next section will seek to provide a three-stage process to fill this lacuna.

The Trilogy of Analogy: Three Stages for Securing Relevance

Stage 1: Similar problem

The analysis from the previous section shows that we need some guiding principle for choosing which factual characteristics should be used for comparison. For this, Bartosz Brozek's theory can be useful. Instead of looking into the factual similarities, it starts by gathering cases for the same problem. These cases can be used to derive principles, which can provide criteria for selecting which factual characteristics to use.

Hence, in our introductory problem, instead of deciding whether a blue self-driven and eco-friendly hover car is more materially similar to a blue motorcycle, a car, an aeroplane, a self-driven robotic grass-cutter or an eco-friendly bicycle, the court should first look into cases on a similar problem, i.e., what vehicles have been allowed. In case they reveal a principle that being eco-friendly is important for being allowed in the park, the court can conclude that instead of it being blue, self-driven or being like a car and an airplane, the relevant factual feature is being eco-friendly.

To be sure, looking at the same problem necessarily involves some generalization. For instance, in the hover car example, rather than looking for principles on the very same problem (i.e. entry of hover car), court will look into generalized problem of what vehicles can be allowed. For such generalization, the court looks into the essence of the problem at hand. But as argued by Sunstein, such generalization should be limited to ‘immediate’ or a ‘low level’ of generality, as invoking extreme generality can enable comparison with problems that have different policy reasons, and are thus irrelevant for each other.

This point can be illustrated by the case of Zaheerudin v. State, in which the Pakistani Supreme Court was deciding whether anti-blasphemy laws violated the constitutional right of Ahemdiyas. The court held that like Coca-Cola would not allow passing off their products, Sunni Muslims cannot allow passing off their faith. 

Hence, instead of invoking a second order or immediate generalization (such as when the constitutional right of freedom to religion is violated), the court invoked an extreme generalization (such as when an aspect of your identity can be used by others) and used the principle from the very different problem involving patent rights. It was a different problem because it involved different policy reasons (such as protecting investment of commercial entities through IPRs) than the reasons for constitutional right to freedom of religion. Hence, it drew analogy from a principle that was irrelevant.

This can be juxtaposed with United States v. Gratkowski, where the court constrained itself to immediate or second level generalization. In this case, the accused was charged for using Bitcoin for buying child pornography. When the investigating authority sought discovery of transactions, the accused contended the same on grounds of right of privacy. The court framed the problem as immediate generality, i.e., when can discovery be sought in such cases. This led the court to a decision in which bank transactions were held to be discoverable, and hence, the court concluded that being a public financial transaction is a relevant factual characteristic.

Similarly, in Ritesh Sinha case, the Supreme Court was asked to decide whether an accused can be lawfully compelled to produce a sample of voice for comparison with other recording. The court answered this in affirmative by forming problem as what can be lawfully compelled in such investigation, and got answer by comparing it with Kathi Kolu Oghad case, where it was held that accused can be compelled to produce physical specimens.

Hence, forming the problem as immediate generality can enable the court to find principles for finding which factual similarities are relevant.

Stage 2: Similar factual characteristic

As discussed above, stage 1 can reveal principles regarding what factual characteristics are relevant to reaching the conclusion. Hence, the process will be incomplete unless one sees whether those characteristics are present in the case at hand.

For instance, in the introductory example, looking for cases of the same problem might reveal that a statue of a boat was allowed as a symbol of cultural decoration of the park and a bicycle was allowed for being eco-friendly. Using the case of boat would not enable the court to reach the conclusion as the hover car does not share the factual characteristic of being such a symbol. Hence, unlike what Brozek believed, the court will have to necessarily analogize facts as well. In that sense, note that this is akin to using Brewer in reverse. Since the court will have the Brewer’s Step 3 rule from cases of previous problems, it can see which of ‘n’ is relevant. From this, the court will then see if those ‘n’ are present in the case at hand. Hence, unlike Brewer’s sequence, court will move from his Step 3 to Step 1.

Very often, courts fail to recognize this stage, and by stopping the analysis at stage 1, they apply examples irrelevant to the problem. An instance of the same is Commonwealth v. Jones,  in which accused was alleged to have used his phone to sexually harass the complainant. When the investigating authority sought password to his phone, he contended against it by using the right against self-incrimination. To analyze this issue, the court used the case of Fisher v. US, which held that if authority knows about the files to be discovered, the question of self-incrimination does not arise. However, this case did not share the relevant factual characteristic because while Fisher case involved physical files, Jones involved a mobile phone, which stores innumerable kinds of files that are impossible to know beforehand. On account of this, giving of password allowed access to the files that police could not have known. As this risk is not present in physical files, court applied an irrelevant case by failing to analogize facts.

This can be compared with the case of Google India Pvt. Ltd. case, in which the complainant alleged intermediary liability of Google on the ground that even when it was notified about the defamatory remarks in the Google Group, it failed to take any action. For this, the complainant drew an analogy with a case that had held a clubhouse liable for not taking down a defamatory remark from the noticeboard. The court noted that this was an irrelevant analogy because, in that case, the company was held liable because it had the right, power, and ability to take down the post. As these factors were found to be contentious in the case of internet-based social groups, the analogy was inapplicable.

Hence, even when the previous case related to the same problem (intermediary liability for not taking down defamatory post), it was irrelevant because the material facts of that case were not analogous to this case. 

Stage 3: Resolving conflicting analogies

While stage 1 gave method of determining relevant factual characteristic and stage 2 implemented the same by comparing the characteristic of source and the target, the process is not complete because stage 2 might give two conflicting analogies.

For instance, in the introductory example, a bicycle could have been allowed in the park on account of it being eco-friendly and a car could have been disallowed because it might run at a high-speed, which could pose risk for the children playing in the park. Since our hover car shares both of these characteristics, stage 2 is not enough to tell which of these conflicting analogies should be finally chosen.

To understand the final stage of relevancy, let us assume a situation in which a competition regulator is deciding whether say Amazon’s restriction on the right to data portability (which enables the users to transfer their data from one platform to another) is anti-competitive or not. In stage 1, court looks for principles of same problem, i.e., what is anti-competitive. In the stage 2, it narrows down to cases sharing relevant factual characteristics. In this, it finds Hemant Sharma and SV Sabam  case, where locking-in of customers was held anti-competitive as it denies market access and imposes an unfair condition on them. But at the same time, it finds another case of Neeraj Malhotra, in which court allowed such activity because negating the same would have harmed financial strength of the entity.

Now, Stage 2 would term both of these analogies as relevant because restricting data portability involves locking-in and allowing the same poses financial burden. To solve this conflict, Brozek suggests that we should adopt a weight formula, in which relative importance of principle A over B can be measured as:

Using this formula in the present case, let us assume that principle A terms restriction on portability an unfair condition on consumers, and principle B allows such restriction on account of financial well-being.

In case B is upheld, there will be ‘High’ interference with A because such locking-in makes consumers vulnerable to the abuses by the platform and it jeopardizes the right to have control over one’s internet profile, which is becoming increasingly important and has also been characterized as a part of fundamental right. Similarly, as consumer welfare is the ultimate and the primary goal of Competition Act, the abstract value of this principle is also ‘High’.

Au contraire, upholding A would lead to just low interference in B because the costs for data portability will be very low compared to Amazon's financial strength. Further, since the financial health of a company has not been termed the ultimate goal of competition law, its abstract value is low or moderate at best.

Putting these values in the weight formula, we get:

As both of these equations give values greater than 1 (4 and 2, respectively), the court will conclude that principle A is more relevant than B.


As the technological advancement is revolutionizing many laws, analogical reasoning is becoming increasingly important. But as shown in the first section of the paper, this reasoning needs a criterion for determining what similarities are relevant, which is not provided by the existing theories.

To fill this lacuna, the second section suggested a three-stage process for determining relevance. In this, the court first looks for cases on similar problem to get principles regarding what factual characteristics are relevant. In the second stage, it uses these principles and compares whether the source and the target situations share these material characteristics. In the last stage, it deploys the weight formula to determine which of the conflicting analogies should be more relevant for the case.

Hence, instead of one single criteria of relevance, relevance of analogy can be conceptualized in the following way:

In this graph, x-axis, y-axis, and z-axis represent three stages, respectively. Point A is irrelevant because like Brozek and the case of Commonwealth v. Jones, it uses only stage 1. Point B is irrelevant because like Brewer, it uses only stage 2. Similarly, point C is irrelevant because this principle is neither related to the same problem nor has similar factual characteristics. Additionally, points D, E, and F are not relevant because they use only two of the three stages. Au contraire, point R is the relevant analogy because it lies on all three axes, i.e., it compared similar problems, juxtaposed similar fact characteristics, and was adjudged to be more relevant out of the conflicting analogy.

Therefore, in this manner, the three-stage process can be used as a working methodology for gauging relevance in analogical reasoning.


Harshit Goyal is a Law Clerk-cum-Research Associate at the Supreme Court of India.


Recent Posts

See All




bottom of page