On the Presumption of Innocence and Other Animals (Column 256)

With God’s help

If you will permit me to repeat a pathetic cliché common in our parts: these days, something momentous has happened in Israel. It has been decided to file an indictment against Bibi, or, in the golden phraseology of the aforementioned gentleman, to stage a putsch against a sitting prime minister by legal means. The discussion surrounding this issue is tainted by quite a few errors and emotions (= two sides of the same coin), and so I thought it proper to take the opportunity to try to dispel some of them. I apologize in advance that what follows is elementary to the point of banality, and I find it necessary to deal with it only because I see how unclear and imprecise these matters are in our public discourse. As Mesilat Yesharim says in its introduction, sometimes it is more important to repeat the obvious than to innovate.

Spurious Correlations[1]

The first point is the correlation between those who support Bibi politically and those who maintain that he can remain in power even under indictment. I have dealt with this more than once in the past, so here I will be brief. The question now before us is a moral one: should a sitting prime minister resign when an indictment is filed against him? There are arguments both ways (mainly one way)[2], and it is no wonder that the public is divided on this question. The question whether Bibi’s policy (if he has one) is bad or good is also an excellent question, except that it belongs to the political-security-ideological plane. Again, it is no wonder that there are arguments both ways.

What puzzles me is why the sides that answer one way on question A are also the sides that answer one way on question B, and vice versa. That is, why the groups that answer ‘yes’ to the first question also answer ‘yes’ to the second, and vice versa. These are two independent questions, and ostensibly we should expect four different groups in the public: those who answer yes-yes, yes-no, no-yes, and no-no. But there are only two: yes-yes and no-no. In previous columns I defined this as a spurious correlation, a phenomenon that reflects the fact that our discourse is biased and tendentious rather than rational. We begin with the conclusion we want to reach, and subordinate to it our answers to all the relevant (and irrelevant) questions. But rational treatment should yield independent views with respect to each such question. A public discussion that fails to distinguish among the questions is a shallow and superficial discussion.

I should clarify here that it is not by accident that I have not dealt here with the question whether the prosecution and the police acted properly or not. The answers to that question can certainly depend on one’s political opinion. Many people on the right feel that these institutions are persecuting the right (without my expressing any view on whether they are correct), and therefore even if there is a dependency between the groups that answer these two questions, in this case it can be understood (even if they are wrong, they are at least consistent). Here the correlation is not necessarily spurious. But the question whether a prime minister under indictment ought to resign is a moral-governance question, and as such it is completely independent of one’s specific political stance. Therefore, here this is a spurious correlation.

The Buzaglo Test

On Friday morning I heard on the radio an interview with two women. One belonged to the group demonstrating in Goren Square in Petah Tikva against the Attorney General. For years she has participated in those demonstrations every week. Opposite her was a protester who comes every week from Eilat (!) to the counter-demonstration. The interview was very amusing, and it illustrated very nicely my longstanding view that many people on both sides lack even the most basic capacity for thought. They can act with great devotion and dedication on behalf of an absurd and ill-defined idea, and for the wrong reasons. Again and again I discover that even opinion leaders are leading their flocks in various directions without having the ability to explain properly why this direction is right in the first place. When one hears their explanations, one understands that these are people who have not really formed a serious position, but a gut feeling suffices for them to invest admirable efforts in advancing their stance.

The interviewer asked the first protester why they wanted to pressure Mandelblit to hurry, given that this is a prime minister and his case ought to be examined carefully. She immediately reacted as though snake-bitten and pulled the expected rabbit out of the hat: the Buzaglo test. The fact that he is prime minister, she said, should make no difference. His case is like that of any other citizen. The other woman explained quite sensibly why the whole matter is baseless political persecution. (A governmental putsch, as we already said?…)

I wondered to myself what the first woman would say if I were to say that Yigal Amir should be judged like any other murderer. No special and retroactive laws, and no special treatment (a ban on pardon, conjugal visits, furloughs, and the like). So what if he murdered a prime minister? After all, the prime minister is subject to the same law as any other citizen (Buzaglo)! It is important to understand that this is not merely a formal comparison. A legal investigation of a prime minister also has major consequences (the government falls; the person elected cannot serve; and so on). It is no wonder that such an investigation is handled cautiously, if only to make sure there is substance to the matter before launching a legal proceeding with all its ramifications. If he is acquitted in the end, it will turn out that there really was a putsch here. A prime minister elected by the public will have been deposed without justification (and do not tell me he can return to politics ten years later when the trial ends and he is acquitted, by which time he will long since have been forgotten, and his replacements will not clear out a place for him even with a tractor). And that, after all, is also the main severity of murdering a prime minister. Morally, he is a human being like any other (Buzaglo), and murdering him is no more severe than murdering any other person (in my view, even less so, because it is murder committed for value-based and ideological reasons). The gravity of murdering a prime minister, insofar as it exists, lies in the consequences and not in the intrinsic severity of the act of murder. Harming a person whom the public wanted as prime minister is harming the public and its will.[3] So why is caution in legal proceedings against a prime minister that lead to injury to the public will not analogous to murdering a prime minister, which injures the public will?! Perhaps you have a distinction (I do not), but I am quite sure that the interviewee never even considered this factor. The political result she so dearly wanted (Bibi’s removal) eclipsed the full range of considerations. She harnessed them all to achieve it (and thus we return to the previous section on spurious correlations).

The Presumption of Innocence: Between the Factual and the Legal Plane

Another matter that keeps coming up in this discussion, this time from the right, is the ‘presumption of innocence.’ How dare the journalists/leftists/protesters and so on behave in an unconstitutional manner, and fail to recognize the prime minister’s presumption of innocence?! How dare they demand action contrary to the law (which allows him to remain in office)? When I hear the deep inner conviction of these speakers, I chuckle to myself. Over the course of my life I have heard quite a few absurdities, and this is one of the greatest.

Indeed, it is true that legally every person enjoys a presumption of innocence, and he may not be punished before he is tried and convicted. But the presumption of innocence is a legal rule, not a rule about facts. If I were to ask you, after it has been decided to file an indictment against so-and-so, what is the likelihood that so-and-so committed act X of which he is accused—would you answer: the same likelihood as that some anonymous resident of Australia did it? That is absurd, of course. Clearly the likelihood that so-and-so did it is far higher. True, one may not take measures against him so long as he has not been convicted, and that is his presumption of innocence, but that is not a determination on the factual plane.

Allow me to bring a few examples regarding this.

one must be concerned (one must be concerned)

The first example this reminds me of is the prohibition against believing slander that I hear. As is well known, it is forbidden to believe it, but one must be concerned (or permitted; see Niddah 61a, and in Chafetz Chaim, Laws of Slander 6:10, and Be’er Mayim Chaim there). That is, if I hear that so-and-so is a swindler, I may not believe it, but I may suspect it and refrain from doing business with him. If I do not believe it, why am I allowed to suspect it? And if I suspect it, what does it mean that I do not believe it? In the Chafetz Chaim there, this distinction is not clear (he takes pains to explain that I may not even entertain doubt about it, which of course is completely absurd).

It seems to me that the meaning of this rule can only be the following. Jewish law itself recognizes a public report as a kind of evidence. If a report circulates about someone, under certain circumstances he becomes suspect in the matter (and sometimes is even disqualified from testifying because of it). A public report is admittedly not evidence in the sense that a religious court would punish someone on its basis. But that is only a legal rule. Factually, a public report does carry evidentiary weight. For all the concerns about deliberate manipulation, such a report still has some clarifying force.

This is essentially the distinction I drew above between the legal plane and the factual plane: legally it is forbidden to accept slander, but factually it is certainly proper and correct to take it into account. It is somewhat like the well-known story about the Chafetz Chaim, who testified in court on behalf of a man he knew. After he left the courtroom, one of his students told the judge that this was a very righteous man, and related that when the Chafetz Chaim wants to send a letter and finds a messenger who will take it to its destination, he buys a stamp at the post office and tears it up (so that the post office will not lose money). The judge did not believe it, whereupon that student said to him: ‘It may not be true, but nobody tells such stories about you and me.’ When a public report circulates about someone, it has an evidentiary dimension, even if not an absolute one: it is forbidden to believe it, but one must be concerned.

"There is a presumption that whatever is in a person's possession is his."

Let me bring another example, this time from a probabilistic angle. In all legal systems, the accepted rule is The burden of proof rests on the one who seeks to take from another. (the burden of proof rests on the claimant). The person holding the money—the possessor—has the upper hand, and if no evidence is brought against him, he prevails in court. The question is why we should proceed this way. Does the mere fact that I am holding the money mean that I am telling the truth and the plaintiff is lying?

In Yeshiva Wiki, under the entry ‘There is a presumption that whatever is in a person's possession is his.,’ they write as follows:

The presumption regarding movable property (usually cited as There is a presumption that whatever is in a person's possession is his.) is a presumption inJewish law whose meaning is that ordinarily, whatever a person is holding belongs to him, unless proven otherwise.

This presumption has several names and appears in many places in the Talmud under different formulations: There is a presumption that whatever is in a person's possession is his., which is the presumption itself; another expression is We do not presume people to be thieves., which is in fact the reason and source of the presumption, stating that an ordinary person is not presumed to be a thief and should not be suspected as such, and therefore when a person is holding an object we should assume it is his property; and a third formulation on this subject is We are as certain witnesses that whatever this one is holding belongs to him., meaning that the fact that what a person is holding is his is so clear that we ourselves are considered witnesses to the matter.

At first glance, the presence of the property in a person’s hands is evidence that it is his. Although the later authorities discussed (see there in Yeshiva Wiki) whether this is a clarificatory presumption or a rule-of-conduct presumption (a legal presumption). Ostensibly, this is a dispute about reality. These views disagree over whether most objects in a person’s home really are his. But on second thought there is no doubt that it is not clarificatory on the factual plane, and no one could dispute that.

This presumption concerns ordinary objects found in some home, such that if we were to gather statistics about them, we would discover that most belong to the homeowner. There is no doubt that if we conducted such a survey, that is what we would find. Does this mean that when such a claim comes before a court, the presumption is that the plaintiff is lying and the possessor is telling the truth? Clearly not. The collection of objects involved in a legal dispute between plaintiff and defendant is a special subset, and regarding it there is no reason to assume that this presumption applies. As I said, why assume that the plaintiff is a liar and the possessor truthful? Think about the set of all objects in the world. As noted, with respect to it there is a majority that are found in their owners’ homes. But within it there is a subset of objects whose ownership is under legal adjudication. Regarding this subset there is no reason to assume that most are found in their owners’ homes (for that covertly assumes that most plaintiffs are liars. Even the Talmud describes a presumption: A person does not make a claim unless he has a basis against the other party.).

To clarify the claim, I will present it using Venn diagrams. Suppose that the ellipse in Figure 1 is the set of all objects in the world. Its right-hand side (the smaller area) consists of objects that are not in their owners’ homes, and the left-hand side, which is the majority, represents the objects that are in their owners’ homes:

Figure 1

Now let us take an ellipse that represents the objects under legal adjudication. The question is where it is located within this larger ellipse. The assumption that the majority exists for this subset as well assumes that this ellipse is divided similarly to the whole ellipse, as in Figure 2:

Figure 2

But this ellipse can also be positioned differently, for example as in Figure 3:

Figure 3

In this model, one cannot infer from the majority that exists in the general set to a majority in the subset of the smaller ellipse.

This is an error that has been studied quite a bit by researchers of cognitive fallacies (see, for example, in my article, ‘The Representativeness Fallacy in Jewish Law,’ and in column 144, on the halakhic and legal implications of this fallacy). Those who suffer from it apply to a special subset the distribution that is true of the general set, even where the subset may be biased (and in our case it is even likely that this is indeed the case).

An Example from the Ketubot Passage: Most women marry as virgins.

The Talmud itself notes this fallacy in Ketubot 16a-b, when it qualifies the majority rule that ordinarily women marry as virgins by the countervailing majority that most women who marry as virgins have a public report to that effect. You can follow the diagrams I presented above, where the large ellipse is the set of all women who marry, most of whom (the side to the left of the line) are virgins. The smaller ellipse represents those among the women who marry about whom no public report went out that they were virgins. You can then immediately see that within the subset of women who married and about whom no public report went out that they were virgins, there is not necessarily a presumption that most are virgins. In fact, this is an even more extreme case than the previous one, because among those who marry as virgins, most do have such a public report. That means we know something about the position of the smaller ellipse (its left side covers less area than the rest of the area to the left of the line and outside it). Even so, we have no way of knowing the ratio between the two parts of the smaller ellipse, and that is the question that interests us here (whether the woman about whom no report went out is a virgin or not).

Let us put numbers into the picture. Think of a situation in which 80% of the women who marry are virgins (that is, the left side in Figure 1 is 80% of the area). Among the virgins, 75% have a public report that they are such. It follows that out of every 100 women who marry, there are 80 virgins, of whom 60 have a public report and 20 do not, plus 20 non-virgins. Therefore, when a woman comes before me and all I know about her is that no public report went out about her, she may belong either to the group of non-virgins or to the group of virgins without such a report, and those two groups are equal in size. Therefore, in such a case I have no way to determine whether she married as a virgin or not (20 versus 20).

Thus, once we have defined a special subset within the general set, it is not necessarily correct to apply to it the majority that holds for the general set.[4] There are many additional examples of this from all areas of life. True, if we have no special information about it, we may infer that the distribution within it is similar to the distribution within the general set. But when we have a special consideration about it (indicating that it is biased relative to the general set), for example that when two litigants come before a religious court both enjoy a presumption of propriety (and therefore among the objects under legal adjudication it is reasonable to assume equality between those that belong to plaintiff and defendant), or that among women who marry as virgins most have a public report (and therefore one may assume equality between virgins without a public report and non-virgins).

Note that the case of non-virgins is more problematic than that of the movable-property presumption. In the movable-property presumption we have a positive rationale in favor of an equal relation between the two groups (both sides have a presumption of propriety), and therefore the smaller ellipse is positioned so that the dividing line passes exactly through its middle. By contrast, in the marriage case we have no positive rationale in favor of inequality within the smaller group, and therefore the smaller ellipse can be positioned in several ways, some of which preserve the majority and some of which do not (this depends on the numbers I presented above. They could of course have been different, in which case a majority would indeed have emerged), as in Figure 4:

Figure 4

In such a situation we assume de facto equality between the two possibilities, because the majority that existed in the larger group has been undermined. When we have no information, we assume that the first majority has been undermined, and then the assumption of equality is negative (because there is no necessity that there be a majority for the other side), and for purposes of the halakhic-legal discussion the picture is like Figure 3 above, though of course factually that is not necessarily the case.

Back to the Presumption of Innocence

Despite everything I have said above, when the two litigants come before the court, the legal presumption is that the money belongs to the defendant until proven otherwise. Although factually there is no basis for this whatsoever, there is a legal presumption that whatever is in a person’s possession is his. This assumption has many varied reasons, some of them self-evident. But these are all legal reasons, and they have absolutely nothing to do with the facts (whose object it really is).[5] Factually, I would say that the likelihood that the plaintiff is lying is more or less similar to the likelihood that the defendant is lying. And yet legally there is a great deal of sense in behaving as though such a presumption exists.[6]

Exactly as we saw in the examples above, in our case too most people whose guilt has not been proven in court are indeed innocent. This is certainly true factually if we look at the entire body of citizens (the large ellipse above; its left side consists of those who are not guilty). An overwhelming majority of citizens have not been accused in court, and therefore their guilt has not been proven, and they indeed are not criminals. But among those who actually are brought to trial, there is not necessarily such a majority.[7] Those who stand trial are a subset of all people (that is the smaller ellipse in the diagrams above), and regarding them there is no reason to assume that most are innocent. Which of the above diagrams represents the situation?

Note that even if we had no information, there are various possibilities (depending on conviction rates and on the reliability of the legal system), as in Figure 4. But as we have seen, even when there are several possibilities the majority has already been undermined and we assume Figure 3, and therefore one cannot apply the majority to the subset. But here we are dealing with people who have undergone a police investigation and against whom it was decided to file an indictment, and especially in the case of the prime minister, who also underwent a hearing before the Attorney General and was equipped with first-rate lawyers, and yet it was decided to file an indictment against him. In such a situation, if I had to bet, it is clear that the more reasonable bet would be that Bibi is not innocent—that is, that he will be convicted.

In this case, the relevant figure is the following:

Figure 5

You can see that in this case, within the smaller ellipse not only is there no majority on the left side (presumption of innocence), there is actually a majority on the right side (those who are guilty).

That is on the factual plane (whether he is guilty or not). On the legal plane, however, it is correct to grant even Bibi a presumption of innocence, that is, not to assume he is a criminal until the matter is proven in court. But that does not mean that this is in fact the more plausible factual possibility. On the contrary, it is more plausible that those against whom indictments are filed are guilty. Exactly as we saw regarding the presumption that whatever is in a person’s possession is his. The legal plane does not necessarily express the factual plane, and yet each has its own justification and it is right to act in accordance with it in the relevant context.

Implications for the Demand for Resignation

What I have described so far does not necessarily mean that one must demand the prime minister’s resignation. On the one hand, on the factual plane it is reasonable to assume that he is guilty and will be convicted, at least on some of the charges (the fact that there are several charges obviously increases the likelihood of conviction). But legally he has a presumption of innocence. One may now discuss whether the demand that a prime minister resign ought to be based on the presumption of innocence (which of course stands for him as well) or whether it should be based on the factual consideration (according to which he is most likely guilty).

The conclusion of this discussion is not unequivocal. One must understand that what is really at stake here is the question of what kind of prime minister we would want for ourselves. Would we want a person who is most likely a criminal (a determination that has passed through several investigatory and legal instances)? In my view, very likely not. Alternatively, there may perhaps be those for whom this does not bother them for some reason (at least so long as it is Bibi). But one thing is clear: there is no place to tie this to one’s political worldview (right or left). This is a moral question and a question about the form of government we would want. And here again we return to the spurious correlations.

A Technical Argument

One might perhaps raise the concern that a policy requiring a prime minister to resign already when an indictment is filed against him enables those who wish him ill to pin unfounded accusations on him in order to get rid of him (what is called ‘a putsch by legal means’), and therefore one must not allow this to happen.

But in my opinion this is a rather weak argument. We are dealing with a demand that arises after several instances have dealt with the matter and said their piece. Moreover, with regard to a minister in such a situation, there is broad agreement that he indeed should resign already at this stage (and so the law also states). Is there no concern about tendentious harassment when it comes to a minister? And again, if these are indeed weak arguments, the correlation between those who raise them and those who support Bibi politically raises grave suspicion (I am using brutal understatement here, of course) that this is a spurious correlation.

We are dealing with a person who, factually, is most likely a criminal. Would we want, because of formal-legal considerations of presumption of innocence, such a person to stand at the head of our state and have control over our fate and the fate of our public coffers? That such a person make the fateful decisions for all of us? I can say that at least I most certainly would not want that. And all this before we have even touched on the difficulty of running a country while also managing a complicated trial, and the division of attention required in such a situation, which itself constitutes a significant consideration in favor of demanding resignation.[8]

Public and Legal Discussion

I think this is what is meant when people call on us to distinguish between the discussion on the legal plane, where every person enjoys the presumption of innocence, and the discussion on the public plane. The point is usually not sharpened, but I think what is meant is the difference between the factual question and the legal question. The claim is that public discussion ought to be based on the facts and not on legal presumptions.

[1] See on this in columns 1, 41, 92, 146, 151, and others.

[2] See my comment here, and the discussion that begins here, and further on in the column.

[3] See columns 242 and 244.

[4] Therefore those commentators are mistaken who see this as one majority against another. The second majority nullifies the first as though it were not there. See the summary of my Ketubot lectures in the recordings here—lesson 8.

[5] This shows that there are reasonings that are not probabilistic and yet are not ‘scriptural decrees.’ I discussed this in the appendix to Kuntres HaMigo, which deals with ‘legal reasonings.’

[6] I think that even the later authorities who maintain that this is a clarificatory presumption do not really mean that it is a factual presumption. They only claim that on the legal plane one should treat this presumption as though it were factual (and not merely a rule of conduct). The same is true of other disputes among the later authorities regarding rules such as majority or presumption—whether they are clarificatory rules or rules of conduct. Usually the factual truth ought to be agreed upon (whether there is clarification here or not), and the dispute is how to treat this rule on the legal level: whether it is viewed as a rule of conduct or whether legally one treats it as though it were clarificatory. A clear example of this is the dispute discussed in Shev Shema'tata and Sha'arei Yosher regarding the rule Does the mother's presumption help the daughter? (see for example here). Suppose a woman had relations with some man, and the question is whether he was of a disqualifying status. If so, he disqualifies both the mother and the daughter born from that union from marrying a priest. The mother has a presumption of eligibility (for before relations with him she was eligible to marry a priest), but the daughter does not, because she was born from that very union. The claim is that if this is a clarificatory presumption, then if we validate the mother because of her presumption, we should validate the daughter as well. If the man involved was not of a disqualifying status, then the daughter too was not disqualified. But if the rule is one of conduct, one can validate the mother and disqualify the daughter. There is no doubt that on the factual plane this rule clarifies nothing (the fact that the mother was born eligible to marry a priest does not mean that the man with whom she had relations was eligible in that respect). Therefore it is clear to me that the dispute is only on the legal plane: whether legally one should regard this rule as a clarificatory rule (even though it does not really clarify) or as a rule of conduct.

[7] This bears on the question of what percentage of defendants are convicted. There are critical claims according to which in Israel that percentage is very high, but just now my friend Judge Menachem Finkelstein wrote an article in which he refutes that claim (showing that it is based on data that are analyzed incorrectly).

[8] See a bit more detail here, and the discussion that begins here.