Judicial Reasoning, not Calculation

3. Rules conflict. Two provisions, or a provision and a principle, point opposite ways. Article 19(1)(a) protects speech; Article 19(2) permits restriction in the interests of public order. Nothing in the text weighs one against the other. The weighing is reasoning.

4. The rules run out. The residue of cases where no authority governs is the subject of [The Gap in the Law].

What kind of reasoning it is

Deductive reasoning does appear, and is the form in which judgments are written: the rule is the major premise, the facts the minor, the decision the conclusion. But the premises are not given; selecting and formulating them is the real work, and that is not deduction.

Reasoning by analogy is the characteristic legal form. This case resembles that one in the respects that mattered there, so it should be decided the same way. Analogy is not calculation because the respects that matter must themselves be identified, and reasonable judges differ.

Reasoning from principle operates where rules fail. A principle, unlike a rule, does not dictate a result; it states a consideration with weight, to be balanced against others. The basic structure doctrine works this way.

Consequentialist reasoning asks what result the welfare of society requires. Cardozo ranked this highest, and Indian constitutional adjudication uses it constantly.

Why the mechanical picture is attractive anyway

It makes law predictable. A citizen can be told in advance what will happen. If reasoning intervenes, prediction becomes harder, and certainty is a real value.

It makes judges accountable. If the process were mechanical, a wrong decision would be an arithmetical mistake, demonstrable and correctable. Reasoning admits of honest disagreement, and disagreement is harder to police.

It removes the judge's personality from the outcome. This is the deepest attraction. If the answer is calculated, it does not matter who calculates it.

The honest reply is that these are reasons to want the mechanical picture to be true, not reasons to think it is. What the legal system actually does is accept that reasoning is involved and then discipline it: by requiring reasons in writing, by appeal, by precedent, and by the collegiate bench.

The Indian evidence

Shayara Bano v. Union of India, AIR 2017 SC 4609, shows reasoning at work more clearly than any exposition could.

Facts. The petitioner sought a declaration that talaq-e-biddat, by which a Muslim husband may divorce his wife by pronouncing talaq three times at once, abruptly, unilaterally and irrevocably, was void and unconstitutional, and that it was not protected by section 2 of the Muslim Personal Law (Shariat) Application Act 1937.

Held. By three to two the practice was set aside, and the majority did not reason alike. Two judges held it manifestly arbitrary and therefore violative of Article 14, applying the arbitrariness doctrine from Royappa. A third held that it was not an essential religious practice and was bad in personal law itself. The two dissenting judges held that it was protected personal law under Article 25 and that the remedy lay with the legislature.

Why it matters here. Five judges, one record, one set of rules, three different routes and two different results. On a calculation model that is impossible; on a reasoning model it is exactly what one expects, because the premises had to be selected and the selection was contestable.

The Indian answer MU wants can be put in three propositions. First, the same text has yielded opposite results at different times: compare A.K. Gopalan v. State of Madras, AIR 1950 SC 27, with Maneka Gandhi v. Union of India, AIR 1978 SC 597, on an unchanged Article 21. Second, dissents are common and are often vindicated later, which could not happen if the answer were calculable. Third, the Court has openly created rules the text does not contain, such as absolute liability and the basic structure.

A worked example

Section 498A of the Indian Penal Code punishes cruelty to a wife by her husband or his relatives. A husband refuses for two years to speak to his wife. Is that cruelty?

The mechanical attempt. Find the definition, apply it, output a result. The definition speaks of wilful conduct of such a nature as is likely to drive the woman to suicide or to cause grave injury to her life, limb or health, mental or physical. Silence is conduct, and mental health is covered, so the machine answers yes.

Why that is not enough. Whether two years of silence is conduct "of such a nature as is likely" to produce that consequence is not readable off the section. It requires a judgment about this marriage, this woman, this duration, informed by a view of what the provision is aimed at, namely the harassment of women in the matrimonial home. A judge who thinks the section aimed at dowry harassment may decide differently from one who thinks it aimed at cruelty generally, and both will write a deductive-looking judgment.

The point. The deductive form of the judgment conceals a choice made before the deduction began, in fixing the major premise. That is Cardozo's proposition exactly.

Criticism

The realist objection. That Cardozo did not go far enough: if reasoning rather than calculation decides cases, then what really decides them may be the judge's temperament, background and politics, with the reasoning supplied afterwards. This is the American realist position and it is a serious one, since it converts the study of the judicial process into the study of judges.

The positivist objection. That the account blurs the line between law and what the judge thinks the law ought to be, and so weakens the claim that there is any law binding the judge at all.

The practical objection. That it offers a lawyer no guidance. If the process is reasoning, advice becomes prediction, and prediction depends on the bench.

Quick revision

1. Cardozo, set by MU in December 2024: the process of the law is not a mere calculation, a mechanical adding and subtracting of established rules; it is a process of reasoning.
2. Calculation fails at four points: characterising facts, the edges of general words, conflict between rules, and the exhaustion of rules.
3. Forms of legal reasoning: deduction in the written form, analogy in substance, principle where rules fail, consequence where welfare is in issue.
4. The mechanical picture is attractive because it promises certainty, accountability and impersonality; the system secures those instead by written reasons, appeal, precedent and collegiate benches.
5. Royappa, AIR 1974 SC 555: arbitrariness violates Article 14, a result not calculable from the text.
6. Shayara Bano, AIR 2017 SC 4609: five judges, three routes, two results.
7. Gopalan against Maneka Gandhi: the same Article 21, opposite readings, twenty eight years apart.

Test yourself

1. Give two points at which the mechanical model of adjudication breaks down, with examples. The characterisation of facts, since whether a motorised toy is a "vehicle" cannot be calculated and must be judged by reference to the rule's purpose; and the edges of general words, since terms such as arbitrary or reasonable have a penumbra of doubt that no procedure resolves, as Royappa shows for Article 14.

2. Why does Shayara Bano prove Cardozo's proposition better than an exposition of it would? Because five judges applied the same materials to the same facts and produced three distinct lines of reasoning and two results. If deciding were calculation, that could not happen. It happens because the premises had to be chosen, and the choice was contestable.

3. What does the legal system substitute for the certainty the mechanical model promised? Discipline of the reasoning rather than elimination of it: reasons given in writing and open to criticism, correction on appeal, the constraint of precedent, and decision by benches of more than one judge.

4. State the realist objection to Cardozo. That if reasoning rather than calculation decides cases, the real determinants may be the judge's temperament, background and political outlook, with the legal reasoning constructed afterwards to justify a conclusion already reached. On that view the study of the judicial process becomes the study of judges rather than of doctrine.