The Casuistic Structure of the Talmud (Column 482)

With God’s help

Disclaimer: This post was translated from Hebrew using AI (ChatGPT 5 Thinking), so there may be inaccuracies or nuances lost. If something seems unclear, please refer to the Hebrew original or contact us for clarification.

In the previous column I discussed the role of nafka minot (practical differences) in Talmudic thinking. The starting point for that discussion was the relationship between study and its practical ramifications (addressed in column 479). In this column I wish to touch on a related point: why the Talmud prefers to present its discussions and principles through concrete cases rather than through general theoretical principles. In legal terminology, this is the question of the Talmud’s casuistry.

The question

In my essay on ukimtot I wrote the following about the character of the Talmud:

Science formulates its general laws explicitly, e.g., “Every … is …”. By contrast, the Sages—both in the Mishnah and in the Gemara—almost never use theoretical formulations; that is, they rarely lay down abstract, general principles. The Sages prefer to express themselves through specific cases that embody the general laws (in legal terms: a casuistic method, as in British law).

Nevertheless, as we shall see below, it is clear to the Gemara that the Mishnah intends to teach a general law and not only a specific ruling. The case cited in the Mishnah is taken as a casuistic expression of a general law. This is why the Gemara abstracts the case in the Mishnah and turns it into a “laboratory case,” thereby distilling the general law out of the Mishnah’s casuistic formulation.

I must add that even the Gemara, when extracting the general law from the Mishnah, does not formulate its conclusion as a general law. The ukimta is the formulation the Gemara uses to describe the general law. Thus the Gemara continues the casuistic manner of formulation used by the Tannaim in the Mishnah: the general law is presented in the Gemara through test cases (laboratory situations) that allow us to see which general law the Mishnah had in mind.

There are three claims here: 1) The Mishnah and the Talmudic discussion prefer to deal with cases (hence “casuistry”) rather than general principles. 2) Nonetheless, it is clear to all that behind these lie general principles. 3) Even the Gemara, which tries to extract the general principles from the Mishnah’s cases, formulates its conclusions by way of cases (the ukimtot).

This picture naturally raises the question I noted in footnote 14 there without explaining it:

Why indeed do the Sages prefer a casuistic formulation and refrain from using the language of theoretical laws (as in a positivist system)? I cannot address this important question there, and for my purposes here it suffices to establish the fact that this indeed is the Sages’ method.

At first glance this appears to be intellectual primitiveness. But I wish to argue that it is not so—that the Sages have very good reasons to prefer discussion in this form. About four years ago I was asked by Avraham (and more recently by EA) about this claim. They asked me for a further explanation, and that is what I shall offer in this column.

Casuistry

The term “casuistry” comes from the Latin casus (case). In the legal world a casuistic law is defined as a law framed conditionally: if someone commits murder, his punishment is such and such; as opposed to a formulation as a rule or theoretical principle: it is forbidden to murder. But this is a technical distinction and, in my view, not terribly interesting. What is more relevant for us is the casuistic approach (not just the wording), which advocates running the legal system through precedents and analogies to them, rather than through rules and deductive derivation from them.

Ask a person how a legal system works and he will likely answer that the legislature sets the rules (the laws), and the citizen or the judge applies them to specific cases. They deduce the legal instruction for the case before them by logical deduction. The law says it is forbidden to murder; now a person is driving recklessly toward a crosswalk where someone is crossing—he understands that this would be an act of murder forbidden by law and concludes that the act is prohibited (and, I hope, refrains from doing it). But this picture is somewhat naïve and by no means universally accepted.

The approach that views law as a collection of general principles and believes in the picture I just described is usually associated with legal positivism.[1] For the positivist, law is a system of rules and the judge’s role is passive. Admittedly, this is a simplistic and extreme description, and in practice even those classified as positivists do not truly think and act that way. By contrast, in casuistic systems the judge is meant to compare (or distinguish) the case before him with prior cases—those stated in the law or in precedents—and to decide accordingly. A principled casuistry tends to think there really are no sweeping theoretical rules, and every case should be judged on its own merits. The basic logical move in this approach is not deduction but analogy (and perhaps induction).

For example, the positivist will say there is a general prohibition on lying, and that in every case we merely apply that prohibition. The casuist, by contrast, may hold that there is no general prohibition on lying in every situation, and that each case must be judged on its own. In some cases lying is permitted; in others not. On the face of it this is a rather silly formulation, since clearly there is a general prohibition on lying, except that in certain cases one may deviate from it. In other words, deviating from it requires justification—not the other way around. It is true that the legal instruction varies across cases and is hard to formulate as a sweeping rule; but that does not mean there is no rule. From the perspective of pure casuistry, there is ostensibly no need for legislation and no rules at all; one could run everything by accumulating comparisons to precedents.

These are of course extreme formulations. I know of no positivist who truly believes a general rule applies to every case and that there is no room for common sense and interpretation that would carve out exceptions. I do not think there exists a positivist who sees the judge as a mechanical logic engine deriving outcomes from the statute book. Conversely, there is also no jurist who thinks we have no need for laws and legislation—that judges suffice and legislators are unnecessary. A heap of cases without any general rules is not a legal system one can operate (how are the cases to be evaluated? If there are no rules, what are the guiding lines?). The two pictures I sketched can perhaps serve as theories and/or general approaches to law—setting an overall direction—but in practice one cannot truly run a legal system that way. Pure positivism is too rigid and fossilized; actually running a legal system requires some flexibility and attention to the particulars of each case. Extreme casuistry is likewise impossible as a method for a judge to reach decisions and, indeed, for legal certainty (that citizens or judges can know what is permitted and what is forbidden and what the law says). These are but simple examples of the problems with both approaches.

As noted, at first blush positivism looks like the cleanest, purest approach—the natural picture. The legislator sets rules and the judge applies them. Without this, we fall into the controversial, complex, and irritating maze called “judicial legislation,” that is, situations where a judge’s interpretation of the law effectively becomes a new piece of legislation—seemingly beyond his authority. But on the other hand it is clear there is no way to create a legal system composed solely of general laws and give the judge the role of a passive deductive engine. Such an approach leads to terrible distortions in many cases. Hence, nowadays no one holds either picture in its pristine form.

A milestone in sobering up from legal positivism was the Nuremberg Trials. There something inconceivable happened within accepted legal thought: individuals were prosecuted for obeying the laws of their own country (Nazi Germany, in that case). The Nazis’ defense was that they were following orders or the law, but the claim was not accepted. Complex and more flexible legal constructions were developed that did not view the law and the rules it sets as the last word. One must understand that there are situations (cases) where the law (the rule) does not apply, or at least there is no obligation to obey it. More than that, obedience to the law is not a defense and may lead to conviction. A less extreme continuation of these doctrines is the construction of the “manifestly illegal order,” which likewise instructs a person not to carry out a directive received from a legal authority where it contradicts fundamental principles (even if they are not explicitly written in the statute book and were not enacted; for example, moral principles).

Despite the convergence of the two pictures I described and the retreat of various legal systems from the extremes, in the final analysis there remain two types of legal systems today: those that lean more toward positivism, i.e., legislation by rules (e.g., German Bundes law). These systems view the judge’s role as deciding the case before him by deductive derivation from statutes. And there are legal systems that advocate adjudication primarily through comparisons to precedents (British common law), which see the judge’s role as analyzing similarities and distinctions in the precedents. As stated, neither system truly resides at the extremes I described. Both progress steadily toward a compromising middle. Both have enacted laws that are binding, not just precedents; and both make room for non-deductive analogies and distinctions, and for attention to the special circumstances of different cases.

Halakhic rules and Ptolemaic astronomy

We have seen that adopting a positivist approach to law yields a rigid, fossilized product that cannot truly capture lived reality in all its complexity. The reason is that no law can genuinely cover all cases and their differences.

A good analogy here is Ptolemaic astronomy (of the Greek astronomer Ptolemy), which took the heavenly orbits to be circular. The circle is a perfect, aesthetic, harmonious path, and therefore it seemed clear to them that this is how the world must operate (a kind of aesthetic version of Occam’s razor). Of course this thesis did not stand up to the facts and was contradicted by various observations (our stubborn world refuses to be aesthetic—at least not according to the aesthetics within us). They therefore added epicycles and deferents to reconcile the circular-orbit assumption with reality. Epicycles are smaller circular paths superimposed upon the larger circular orbit, and one can then add further circles and spheres within the previous ones, and so on, to adjust the model and bring it closer to empirical observations. In the end Tycho Brahe and Kepler came along and showed that the orbits are actually elliptical. After taking account of all the epicycles—circle upon circle upon circle—and all sorts of small circles, the whole thing is an attempt to describe an ellipse by means of a big circle with two smaller circles at its ends, and then to fill in the picture with more and more little circles until one arrives at an ellipse.

Ever since, “epicycles and deferents” have served as an analogy or expression for a fossilized outlook that insists on speaking an a priori language to a recalcitrant reality. This forces those who hold such a view to complicate matters endlessly to preserve the language, instead of changing the language and acknowledging that reality owes nothing to our a priori categories. Sometimes an openness to a new language and a willingness to abandon the old reveals that describing reality is much easier than we thought.

Back to the Talmud

Returning to the Talmud and halakhah, the rules there behave quite similarly. I will illustrate with an example. There is a clear Talmudic rule for deciding disputes between Abaye and Rava: the law follows Rava in all disputes except for six cases, remembered by the acronym YaL KaGaM (see Kiddushin 52a and parallels). Seemingly a very clear rule fixed already in the Talmud, uncontroversial. Yet Maimonides rules in several additional places (beyond those six) in accordance with Abaye. For example, regarding “lo titgodedu” (see Yevamot 14a; and Rambam, Laws of Idolatry 12:14) and regarding “i aved lo mehani” (according to some interpretations; see Temurah 4b; and Tzitz Eliezer IV §3, who discusses this at length; and see here). It is no surprise that devotees of rules begin to add epicycles and deferents to the Talmudic rule. Some propose a sub-rule: in disputes where Abaye and Rava argue about the view of other sages (Tannaim or earlier Amoraim), the rule does not apply. Whence this sub-rule? To my mind, only from their fevered imagination. If more exceptions are found (places where Abaye and Rava disagree over others’ views and the law still follows Rava), another epicycle and deferent will be added—for example: if the dispute concerns Tannaim of the third generation and later, then in fact the law follows Rava; or in such a case it necessarily follows Abaye; and so on.

The reason for these convoluted moves is very similar to our astronomical example: the assumption that the rule is rigid and sweeping does not stand up to the facts, and in order to continue holding on to it, small auxiliary assumptions are added. But this assumption is manifestly mistaken. The Talmud, like every legal system, shows remarkable recalcitrance toward the positivism of the “rule-mongers,” though to their credit they do not despair. This is probably also why this literature is entirely neglected—not only by students but even by halakhic authorities and decisors.

My claim is that this is why the Talmud does not believe in rules and focuses on cases. Rules are too rigid; cases always refuse to submit to them. Therefore it sees dealing with cases and analogies as a more effective, efficient tool than formulating rules and deriving deductively from them. No wonder the Talmud rarely formulates sweeping rules and prefers to deal with cases (and sometimes present the rule through them). The belief embedded in this structure is that—like English law—the Talmud does not believe in rules but in examples and more flexible thought. It looks looser and less efficient, but ultimately it yields far more accurate, correct, and less fossilized results. Cases convey the halakhic message better than sweeping rules. Examples will follow.

The hermeneutic rules of “general and particular”

In the second volume of the Talmudic Logic series we dealt with the hermeneutic rules of “general and particular” by which the Torah is expounded. We showed there that even the Torah has no general laws and prefers formulation through examples. For instance, regarding the redemption of ma’aser sheni, the Torah does not tell us the rule—what one may redeem it for and what not. It gives examples (“herd and flock, wine and strong drink”). We are meant to generalize from these examples to understand what is permitted and what is forbidden.

The Gemara sees in the verses describing the redemption of ma’aser sheni a structure of “general, particular, particular.” In the hermeneutic rules there are two further related structures: “particular and general” and “general and particular.” We explained that this is halakhah’s alternative to the positivist structure. The hermeneutic rules instruct us that when the Torah formulates a verse with a general expression followed by some particulars and then again a general expression, it intends us to take the examples and generalize them to various radii determined by the formulation: “general–particular–general” → a medium-radius generalization; “general–particular” → a minimal generalization; “particular–general” → a maximal generalization. And in one place in the entire Talmud we also find a “particular–general–particular” structure—a different medium-radius generalization.

We explained that these structures arise from the very problem that modern legal positivism tries to cope with. We illustrated it through H. L. A. Hart’s (the positivist’s) discussion about bringing vehicles into a children’s playground. A general law banning “vehicles” tells us nothing about what exactly counts as a vehicle: bicycles, a toddler’s ride-on toy, a war-memorial tank, a tractor? Context can help us understand the intent (purposive interpretation: preventing noise or the risk of running children over), and we can also do this through examples: “No vehicles may be brought into the playground, such as a motorcycle, a tractor, or a car, and similar.” Note the “general–particular–general” structure, whose goal is the understanding that, on the one hand, the law is broader than the specific examples, and on the other, the examples are the basis that gives us the direction for generalization. This is a good illustration of the limitations of a rule compared with examples, and of the advantages of combining the two.

Another example: “This one benefits and that one does not lose”

In column 431 I discussed the development of lamdanut (analytical learning) over the generations. I illustrated its sophistication through the sugya of “zeh neheneh ve-zeh lo chaser”—“this one benefits and that one does not lose.” The backdrop was the tension between a positivist formulation of a sweeping analytic rule like “this one benefits and that one does not lose” and a casuistic discussion of a specific case: “One who lives in his fellow’s courtyard without his knowledge—must he pay rent?” The Gemara usually discusses cases, but in that unusual sugya the Gemara itself tries to extract the rule from the case; in most sugyot this is done by the commentators. It is worth seeing that column for the full picture; here I will focus on our point.

In the course of the discussion, Rav Chisda enthusiastically reports to Rami bar Chama about the debate in the study hall regarding “this one benefits and that one does not lose.” Opposite Rav Chisda’s evident enthusiasm, Rami bar Chama dismisses it with indifferent scorn, saying that all their pilpul is pointless because the law is set by an explicit Mishnah. After Rav Chisda insists that he identify the Mishnah, Rami bar Chama’s surprising answer arrives:

“If you derived benefit, you must pay what you benefited.”

The case concerns an animal that ate vegetables in the public domain. Its owner is exempt from paying for the damage because “tooth” damage in the public domain does not incur liability. But the Mishnah adds that if the animal derived benefit (it was sated and he saved the feed he would have had to give it), he must pay for the benefit.

This is a fascinating and very puzzling answer, especially coming after build-ups that lead us to expect something sensational. After all, that case is one of “this one benefits and that one loses,” since the owner of the vegetables has lost his produce. How did Rami bar Chama think to resolve the question of “this one benefits and that one does not lose” from that case? One can hardly avoid the sense that the Gemara deliberately creates an expectation of brilliance so that, when the source is revealed, we think it nonsense (as indeed Rava did) and are disappointed.

The Gemara itself asks:

“But that is a case of ‘this one benefits and that one loses,’ whereas ours is ‘this one benefits and that one does not lose’!”

And it explains:

“Rami bar Chama held that produce placed in the public domain is treated as if ownerless.”

One who places produce in the public domain is deemed to have abandoned it.

This cannot be taken literally. If the owner abandoned it, then indeed there is no loss; but then why should one pay him for the benefit? The produce is not his. It is thus clear that the Gemara means that in such a situation it is as if he abandoned it; i.e., he bears contributory fault for placing it in the public domain, and therefore the payment is for benefit as if there had been no loss.

Rava mocks the proof from the Mishnah and is certain it is foolishness. Why? Because he examines matters in a positivist way (see my columns there regarding the usage of talmid chacham). The sweeping rule is that where there is benefit and there is loss, one must certainly pay, and that differs from a situation of benefit without loss. By that rule it is obvious the Mishnah’s case is “this one benefits and that one loses.” Rava is indeed correct. But Rami bar Chama wishes to teach him—and Rav Chisda—that when one examines the case on its own terms one discovers that the rule misses something here. True, there is loss, but there is also contributory fault; thus de facto it is as if there were no loss.

Now ask: is there a sweeping rule for all “no-loss” situations? According to the Gemara’s phrasing—yes, for it speaks about the collection of cases described by the term “this one benefits and that one does not lose.” But now we can see this is only phrasing. Our case is exceptional and the rule should not be applied to it. There is a very good reason, of course, but from the bare phrasing of the rule we could not have known it. One must use common sense and examine our case—not only derive mechanically and deductively from a sweeping rule.

Another example: liability for damage caused by one’s property

The general rule in tort law is that a person must pay if his property causes damage. The condition for liability is that the damaging object was owned by the defendant. From this rule it follows that if the damaging item was not his property, he is exempt. This raises a question: what if an animal is in my custody but is not mine—am I liable for its damages? It turns out that I am. Thus, for example, a bailee or a thief—although not “owners” in the full legal sense—are liable for damages caused by the animal in their possession. This does not sit well with the rule that liability arises only when the damaging item is owned by the defendant, but common sense dictates that in such cases he should be treated as an “owner” for the purposes of tort liability.

In my Bava Kamma lectures (7–11) I discussed the law of one who places his fellow’s animal in another fellow’s field, comparing it to a thief and to a bailee. In all these cases the defendant does not own the damaging object, yet there is logic to imposing liability and responsibility for the damage; and indeed the Rishonim debate whether this is liability as “a person who damages” (in which case ownership is unnecessary), or a special liability the Torah innovated, or perhaps it can be derived from the rule that “whenever one’s property causes damage one must pay,” even though here it is not one’s own property. At least according to some views in some cases, the debate is precisely over the positivist question. All agree that one who takes his fellow’s animal and places it in someone’s field, and it eats, must pay. But Tosafot and the Rashba there dispute whether this is “a person who damages” (since it cannot be “property that damages,” as the placer is not the owner) or “property that damages” despite the lack of ownership—he is “as if” the owner for this purpose. The same goes for the thief and the bailee. Regarding a bailee whose deposit caused damage, for example, some hold the bailee’s liability derives from the laws of bailees and not from torts, since in torts he cannot be liable because he is not the owner of the damaging animal. According to that approach he is liable to the depositor, and the victim must sue the depositor (practically we may simplify this due to shi’abuda de-Rabi Natan). Others hold the liability comes from torts, despite the lack of ownership—he is treated as an owner for this purpose.

Again we see that a positivist formulation that hangs everything on ownership does not stand up to specific cases. It is too rigid and fossilized to handle reality’s complexity. Yet the Talmud does not forgo the general formulation, and still ties liability to ownership. It relies on our interpreting matters with common sense and applying them according to circumstances, not merely by technical deduction.

Another example: delivery of a bill of divorce

A final example is the delivery of a get. The Talmud’s attempts to define what counts as delivery of a get are extraordinarily extensive. Scores of examples are brought (primarily in tractate Gittin): a get written on something prohibited for benefit; placing the get in her courtyard; giving it into his own courtyard and transferring that to her; a get in her hand with a string in his hand; a get in the hand of her sleeping (and bound) slave; “this is your get but the paper is mine”; and many, many more.

The commentators struggle desperately to extract from this welter of examples a rule that will define “delivery of a get.” Such a rule would be sweeping, so that all the Gemara’s cases would be instances of it (and one could deduce the law deductively in each case). The Ketzot HaChoshen tries to define delivery as either a transfer of ownership or a physical handing over, and rejects both possibilities (prohibitions of benefit undermine transfer of ownership; physical delivery is undermined by transfer via courtyard). The Ketav Sofer devotes four long, very tiring sections to this (§§14–17), offering tortuous formulations that attempt to fit all the examples—and come up empty-handed.

The obvious question is: why didn’t the Gemara itself give us the rule, sparing us the frustration and futile labors? Seemingly, formulating a rule would reduce the chance of error in application to various cases, since we would use deduction from the rule to know the law in each case. In my Bar-Ilan weekly-parashah essay for Ki Teitzei (year 5755), briefly mentioned as well in my essay on Zeno’s arrow, I argued that the reason is that there is no simple way to formulate such a rule. Any formulation would lead to mistakes and misapplications in various cases. Therefore the Gemara preferred to give us a long list of cases, assuming that studying them would clarify the law of delivering a get in the best way. Surprisingly, it is precisely the casuistic presentation that yields the most accurate result and minimizes errors in application. Rules and logical inference are reliable, absolute, and precise—but their very strength is their weakness. Analogy and comparison to cases—tools that are weaker and less precise—produce better outcomes.

Incidentally, in learning systems today it is already known that the best way to teach a system or endow it with some function is not through a general definition of the function so that the system deduces applications, but rather by training through examples and inducing tentative rules that improve as examples multiply. In my view this is precisely what the Gemara did in the sugya of get delivery.

Following a rule

The later Ludwig Wittgenstein, in his Philosophical Investigations, dwells at length on the problematics of learning by rules. His claim is that, in essence, we have no absolute way to follow rules; therefore giving a rule can never ensure that the learner has understood and applied it correctly.

Take, for example, the sequence: 3, 5, 7…. What is the next number? Many of you, I assume, will say: 9. That is of course a possibility (the assumption being that this is the sequence of odd numbers), but equally it could be 11 (assuming the sequence of primes). So who is right? No one—or, actually, both. Moreover, it is easy to show that any number you propose as the next term can be justified. If someone answers −27.43, I can easily justify that as well. How? Suppose the sequence is generated by:

F(n) = a + bn + cn² + dn³

We have four fixed coefficients (a, b, c, d). If we require that for n=1 the function F yields 3, for n=2 it yields 5, for n=3 it yields 7, and for n=4 it yields −27.43, we get four equations in four unknowns. Solve them, and now we have an expression that justifies the answer −27.43 as the next term. And so for any number you like.

Now you wish to teach students the sequence of primes, so you write out the primes up to, say, 103, and assume they can now continue on their own. But, as noted, a student might understand that up to 103 these are primes and from there on it’s odd numbers; he will answer that the next is 105. You will never succeed in teaching him the rule, for learning a rule is always based on demonstrating it on a finite number of examples.

Wittgenstein concludes that rules have no independent standing. Their meaning depends on the examples and on the contexts of the learner who applies the rule learned. There is no way to transmit a rule from one to another without anchoring it in examples, and the examples can always be extended in whatever manner occurs to the learner. The continuation depends on his way of thinking and habits; there is no single “true” or “correct” continuation. From here we can see why the positivist method rests on an illusion—that acting by rules would minimize error and yield certainty. Not only is acting by rules not recommended; it is impossible. One cannot act by rules without illustrating them through specific examples, and those will in turn project back onto the rule’s meaning.

The conclusion is that positivism is not only ineffective; it is sheer illusion. There is no such thing as positivism. One cannot produce a law that is merely a set of rules and assume they will unambiguously determine the legal outcome for every new case before the judge. The judge will always have to use analogies and common sense; it is thus clear he will not always hit upon the legislator’s intent. This is also the idea behind the “general and particular” hermeneutic rules, as shown above. A rule has no meaning unless accompanied by examples that guide its addressee’s interpretation.

A principled conclusion

The conclusion from the analysis thus far is that one cannot base a normative system on a set of sweeping rules. On the other hand, we clearly cannot function without rules. By what do we decide the law in the case before us? What will guide our analogies to other cases? What are the relevant lines of similarity that determine whether this is similar or different? These are criteria that presuppose some rules. Recall Wittgenstein’s sequences: behind every continuation I proposed stood a rationale. If the next term is 9, the rationale is “odd numbers.” If the next is 11, the rationale is “primes.” If the next is −27.43, the rationale is “the function F with the coefficients we solved for.” So too in all the examples above. The assumption that one who places his fellow’s animal in another’s field is liable is based on a rationale of owner responsibility, from which we extract situations of responsibility even absent ownership. But we cannot jettison ownership entirely from the criteria for tort liability. The prohibition on lying is certainly sweeping, even though there are cases in which it is permitted. It is not true that there is no rule. There is a rule—but it is delicate and not amenable to positivist formulation.

Our path forward in legal and halakhic interpretation proceeds in a continual back-and-forth between examples and rules. We must form a rule out of the examples, return and test it on further examples, and then form an updated, more precise rule—again and again. This is a delicate tango between examples/applications and the rule, and we will never have certainty in the result nor a single, clear, universal outcome. Hence there will always be disagreements among interpreters; we will never rid ourselves of the nuisance of judicial legislation (see on this in the final part of my book Ru’ach HaMishpat). Both rules and cases (examples, applications) are needed in this tango; one cannot do without the other. Positivism assumes one can rely solely on rules and logic—an illusion. Casuistry assumes one can rely solely on precedents and examples—also an illusion.

The upshot is that one cannot interpret without rules—but one must beware of clinging to them. The Talmud speaks in cases and relies on us to infer rules from them. Occasionally it also mentions rules, but even then we must not take them too seriously. We must not adhere to them too tightly. We must always weigh their fit to the case before us. The Talmud says the law follows Abaye only in YaL KaGaM—yet there are additional cases where the law follows him. It is a guiding, approximate rule; we must beware of cleaving to it too tightly.

This does not mean there are no rules at all, or that rules are never precise and truly correct (and are only approximations). It is possible that there exist precise, sweeping rules and the problem lies in our formulations. Perhaps somewhere “in heaven” there is a set of general, precise rules that fully define halakhah; yet even so, our formulations may fail to capture them. Therefore, even if someone assumes that there are precise rules out there (i.e., that “pure halakhah” is a collection of exact, sweeping rules), we must always suspect that the current formulation of a rule is imprecise—just as in scientific research that strives for general laws of nature but assumes that almost always we are dealing only with approximations.[2]

The Talmud’s attitude to rules

From here we can understand the Talmud’s guarded attitude toward rules. Beyond what we have seen so far, I will bring two more examples.

As is known, the Mishnah rarely brings rules; rather, it deals in examples or cases. At the beginning of Bava Kamma, the Mishnah departs from its norm and brings a rule:

“There are four primary categories of damages: the ox, the pit, the mav’eh, and the fire. The ox is not like the mav’eh, and the mav’eh is not like the ox; and these two, which have breath of life, are not like fire, which has no breath of life; and these, which go forth and cause damage, are not like the pit, which does not go forth and cause damage.”

“The common denominator among them is that they are prone to cause damage (Rif’s version: ‘and they are your property’), and their guarding is upon you, and when they cause damage the damager is liable to pay the best of his land.”

After specific examples are cited, the Gemara presents their common denominator. Seemingly, if the Mishnah gives the rule, then the earlier examples are superfluous. Indeed, we might expect the Mishnahs always to give the rule rather than cases—it is more precise and efficient. Well, at least here the Mishnah did us a favor and gave the rule.

Yet the Gemara (6a) asks:

“What does ‘the common denominator among them’ come to include?”

A rather astonishing question. The Gemara sees redundancy in the Mishnah, for it brought both a rule and examples. It already senses something is unnecessary; but what it proposes to remove is the rule—not the examples. It does not understand why the rule was brought if there are examples. One would have expected it to ask the reverse: why give the examples if there is a rule? But for the Gemara, the examples are primary and the rule is secondary and peripheral.

This already points to a far-reaching conception. Not only is the rule not preferable and not necessary; even when given, it is superfluous. The examples do the work on their own; the rule is simply unnecessary.

A second example is in Kiddushin 34a (and parallels):

“Which is a positive commandment not time-bound? Mezuzah, parapet, returning lost property, and sending away the mother bird. And is that a rule? Behold, matzah, rejoicing [on festivals], and hakhel are positive commandments that are time-bound, and women are obligated! And furthermore: Torah study, procreation, and redemption of the firstborn are not time-bound positive commandments, yet women are exempt! Said Rabbi Yochanan: One does not learn from general principles, even where it is stated ‘except for…,’ as we learned: ‘One may make an eruv or shituf with anything, except for water and salt.’ And is that all? What about truffles and mushrooms? Rather, one does not learn from general principles, even where it is stated ‘except for…’.”

Note the Gemara’s startling innovation. Granted, if there is a general rule that women are exempt from all time-bound positive commandments and we find some exceptions, we can say “the Tanna taught the rule and omitted some exceptions.” But when the rule includes a detailed list of exceptions, one would expect there to be no further exceptions. It turns out that even where the rule states “except for…,” we should expect exceptions. This is overt disdain for rules—and precisely where the Gemara brings a rule. The lesson: even when using rules, we must treat them with limited trust and with suspicion: honor it, but doubt it.

It is no wonder that Maimonides—who knew the rule that the law follows Rava except for YaL KaGaM (“except for YaL KaGaM”)—rules like Abaye in additional cases, for he knew just as well the rule that “one does not learn from general principles—even where it is stated ‘except for…’.”

The relationship between practice and theory: halakhah and science

Halakhah deals with cases; theory (lamdanut) deals with rules. The relationship between them is complex. The positivist approach sees halakhah as deductive consequences of theory, but that is simplistic. In many instances the situation is reversed: theory is constructed as generalizations distilled from the insights gleaned from cases. Intuition precedes theoretical conceptualization.

I must admit that in the past I recoiled from practical halakhic decision-making. It seemed to me technical and boring—at best the application of analytic principles. Since then, I have matured. The complexity of real-world practice and its very ungentle fit with abstract, general theories is fascinating. Without analyzing cases with common sense, theories are not worth much. They receive their full meaning when they are set against cases. The theory that “ownership is required to be liable for damage caused by one’s property” is not worth much if one does not examine it against cases of placing someone else’s animal in another’s field, of a bailee whose deposit caused damage, of a thief, and so on.

In passing I will note that even in science, theory by itself—without being applied to cases—remains vague, unclear, and imprecise. Experiment does not only test the theory; it clarifies it and illuminates its implications. In the maturation process I described, I discovered that experimental science has a complexity that theoretical science lacks, with a beauty no less captivating. Their combination yields optimal understanding of the world—just as in halakhah the combination of cases and rules yields optimal understanding and the best decisions.

A concluding note

I have argued here for the Talmud’s casuistic character and explained the rationale underlying the decision to build it that way. Was this conscious? I am not at all sure. I do not know whether the Talmud’s editors consciously thought about how to edit it and decided to leave it in the pattern of give-and-take rather than bottom lines. But it seems to me a very wise decision, and I do think that at least intuitively (even if not fully consciously) this was their rationale. It would have been natural to build a binding halakhic code as Maimonides did, rather than like the Talmud. I do not believe this choice stemmed from primitive thinking or ignorance of the drawbacks of an open text. In my view, at least intuitively they understood the words of Nahmanides in the introduction to Milchamot Hashem:

“And you, the reader of my books, do not say in your heart that all my answers to our teacher Rabbi Zerachiah, of blessed memory, are in my eyes decisive proofs that compel you to concede, and that you will boast if you can raise a doubt against one of them—or that you will strain yourself to enter the eye of a needle to cast off the force of my proofs. It is not so. Every student of our Talmud knows that in the disputes of its commentators there are no absolute proofs, and that most difficulties are not conclusive—for in this wisdom there is no demonstrative proof like arithmetical calculations and physical experiments. Rather, we will put all our delight and suffice, in every dispute, to distance one of the opinions by weighty arguments; we will press the sources; and we will give greater credit to the side that better accords with the plain halakhot and the sound sugyot, together with the agreement of right reason. This is the limit of our ability and the aim of every wise man and God-fearing person in the wisdom of the Gemara.”

The Talmud’s editors only drew the obvious conclusions from this regarding the proper nature of such a work.

I have often noted the genius in the decision of the Sages of Israel to accept the Talmud as the canonical text defining halakhic discourse and decision-making. This is not in spite of its being an open text, but because of it. A closed text with rules and bottom lines would not have withstood changing circumstances and times, nor could it have provided unified answers for all shifting contexts and cultures. It is precisely an open text that enables a conversation among sages and communities of very different cultures, with all attempting to apply flexible principles drawn from Talmudic cases to the cases before them. Many disagreements and much non-uniformity arise—but this is to the Talmud’s credit, not its detriment. A closed, rigid (positivist) text would have shattered amid changing circumstances; today we would not live within a single halakhic culture, but in countless disparate halakhic systems with no possibility of dialogue between them. This is another cultural-historical expression of the decision to prefer casuistry over positivism.

Again, I do not mean to claim this was a conscious decision or that these were the reasons. In retrospect it turns out that the collective thinking that led to this choice was brilliantly effective. Without it, we would not be here today. Thanks to it we have a single halakhic system within which a conversation unfolds among divergent views, approaches, and positions—yet all can speak to one another, challenge and answer, and even sometimes be persuaded and decide. Without Talmudic casuistry, I doubt any of this would have happened.

[1] The connection to the usual meaning of “positivism” is not straightforward. Typically the term “positivism” describes the approach that separates law from morality (as against natural law). Here I use it somewhat differently. There is a connection between the two meanings (to explain it we would have to enter the topic of philosophical and scientific positivism), which I will not pursue here. Here I use “positivism” to describe the view of law as a collection of principles by which the judge is to decide using deductive logic.

[2] See on this in my weekly “Midah Tovah” essays for Parashat Ha’azinu, 5764–5765: essay 53 and essay 101.