A Systematic Look at the “Witness Argument”: My Critique of David Hume (Column 672)

In the previous column I presented Hume’s arguments against the reliability of traditions about miracles and revelations. I now turn to my criticisms of Hume’s arguments.

Critique of Empiricism

My substantive critique of Hume concerns the logic of his argument, that is, the first part of the essay. First and foremost, I do not accept Hume’s empiricism—his basic assumption that experience is our only tool for handling facts. I won’t repeat here in detail my arguments for rationalism, which in large measure are rooted in David Hume himself (see Columns 87, 363 – 364, 586, 588, and many more). Briefly, all of science—and in particular Hume’s own empiricism—rests on quite a few assumptions that have no source in experience. As I have shown in my books and the columns cited above, any empiricist must concede that we possess a priori tools (intuition, the “mind’s eye,” and the like) by which we arrive at scientific claims about the world—unless he remains a thoroughgoing skeptic (i.e., he accepts no claim about the world as admissible). For example: the assumption that there is no action at a distance; the assumption of causality; the assumption that laws persist across time and space; the assumption that our senses accurately reflect reality (and that there is a reality at all); and much more. Hence, every scientific generalization—including many found in Hume himself—is not an empirical inference. The facts underlying it are learned from experience, but the general law we construct from them is not the result of observation alone; it also depends on many assumptions that are not products of experience. Hume himself tried to purge the picture of these assumptions—unsuccessfully.

Critique of Induction — A Contradiction in the Argument

The clearest example of this—again rooted in Hume’s own argument—is the principle of induction. What we call “learning from experience” rests on the assumption that what we have seen will continue to behave the same way in the future and under circumstances different from those observed; in other words, that what has been will be. This assumption itself cannot possibly have an experiential basis. One might claim that we have observed in the past that reality is indeed uniform, consistent, and persistent; but the inference from the past to the present is itself induction, and therefore cannot ground the validity of the principle of induction. Even if it could, it is clearly not a “purely observational” basis (for the directly observed facts by themselves do not yield the desired conclusion). Consider, for instance, Hume’s own example in this essay: trust in human testimony. At most this is a generalization from cases we have observed. What grounds that generalization? Assumptions such as the uniformity and persistence of human nature; that we can learn from the people we met to all humanity (i.e., that our sample is representative); that one can assess a person’s credibility (as a function of his interests), and so on. Beyond that, all the qualifications Hume himself raises about the number of witnesses, their interest in lying, etc., are the products of reasoning and generalization—not of observation alone.

It is worth noting that the source of empiricist doubt about causality and induction is David Hume himself. Hume showed that neither has a basis in observation and therefore treated them as our habits of thought rather than as claims about the world. As we shall see, this is especially odd in light of the continuation of his argument in this very essay.

After the empiricist declaration, Hume moves on to discuss the content of testimony. He argues that if the content is implausible, then highly reliable testimony is required for us to accept it. That sounds reasonable; but he goes on to claim that if the content of the testimony is a miracle—i.e., something that directly contradicts the laws of nature—then categorically no testimony could be accepted about it (there is no degree of credibility that can outweigh the assumption that such a thing cannot occur). His claim is that both sides of the equation must be assessed in light of our experience: the credibility of the witnesses and the content of the testimony. Here a striking trust appears in Hume regarding the principle of induction, which he himself presented as non-empirical. Hume asserts that if in our experience we have not seen or encountered miracles, then there are no miracles. For him it is impossible to believe that miracles occur, i.e., events that violate the laws of nature (as they are currently known to us). This seems like remarkable optimism given the skepticism he himself exhibits toward induction. If indeed our confidence that what has been will be is nothing more than a mental habit and not an empirical result, how can he ascribe to it such certainty and build on it an a priori and absolute distrust of testimonies about miracles?!

From another angle one can put it thus: precisely as an empiricist, Hume should prefer the direct encounter with a miracle as a factual source of knowledge, for that would be an immediate cognition we had—a directly experienced event. By contrast, the conclusion that, in light of our experience, miracles do not occur (that the world operates by the laws of nature) should be rejected, for it is a generalization from observed facts, not a directly observed fact. That generalization is contradicted by our current direct observation. An empiricist should clearly prefer direct observation to the laws of nature. It seems odd to argue, in the name of empiricism, for the priority of the generalization over a direct observation. To sharpen: I do not mean to claim that hallucination or mistake is impossible when we encounter a miracle. Of course it is possible. But a priori—so long as it has not been proven that it was an illusion and we are weighing the two possibilities against each other—an empiricist should start from the presumption that the direct observation is more credible. It merely reveals to us that induction—already suspect on its own terms—failed here.

Moreover, even if for some reason we decide to cling to the a priori (and non-empirical) assumption that the world operates according to rigid laws of nature, who can guarantee that the laws known to us now are the correct ones? Time and again we have discovered that the laws of nature are not what we thought. Newtonian mechanics was corrected by relativity and quantum mechanics. We discovered dark matter. We discovered that light travels at a constant speed that does not obey Galileo’s velocity-addition rules, and more. Perhaps the “miracle” we now experienced is merely a correction to the laws of nature as we conceived them. Is every deviation from the laws as we currently know them impossible? I would not expect empiricists to take such a stance. Note: I am not arguing that a miracle is a scientific observation that should revise the laws of nature. It does not meet the scientific test of repeatability (the ability to reproduce the event in any lab by anyone under similar conditions). My claim is twofold: (a) Precisely from an empiricist standpoint, the “miracle” enjoys priority over the opposing laws of nature, because it is the product of direct observation. (b) The laws of nature do not possess the very high certainty Hume ascribes to them, since they are the result of induction—and, as noted, Hume casts doubt on induction.

Critique of Induction — Begging the Question

If you look carefully you will see that Hume weaves into this argument, with virtuosity, both a contradiction and begging the question (see, briefly, Ben Elkanah’s question here). I dealt with the contradiction (vis-à-vis his view of induction) in the previous section; here is the begging of the question in that very argument.

If we assume, in light of our experience, that there are no miracles, then no testimony about miracles can be accepted (for Hume’s reason), and—presto!—we discover that we have no reliable historical testimony about miracles. Therefore, as a “lesson from experience” there is no possibility of accepting their occurrence (again, for Hume’s reason). A neat circular hocus-pocus. It is worth recalling the examples Hume himself brings in the second part of the essay, concerning various miracles, where—even with allegedly perfect credibility—he dismisses the reports out of hand as blatant lies, without any indication for doing so. No wonder, then, that in his view no reliable report of miracles has ever been accepted; on his terms, such a report cannot be accepted.

To sharpen: as a card-carrying skeptic, I am not claiming here that these miracles actually occurred (see more below). In most cases I, too, tend to reject such testimonies. My claim here is logical: Hume’s argument begs the question. It is important to understand that, as I have written more than once, begging the question is not always a fallacy. Any position grounded in a logical argument presupposes what it aims to establish (see, for example, my essay here). But that is acceptable when one advances an argument to defend a position—i.e., to ground it in the speaker’s own premises. By contrast, when an argument is presented as an attack on another position, begging the question has no force. Reuven cannot attack Shimon’s view on the basis of Reuven’s own premises that Shimon does not accept. In our case: you can assert that you yourself are unwilling to accept the existence of miracles and entrench yourself, circularly, in that stance. This is a legitimate and consistent position, and I have no principled problem with it. But clearly it cannot serve as a refutation of someone who is willing to accept them. Hume of course intended his argument as a refutation, not merely as a statement of his personal stance. Begging the question cannot constitute a refutation.

Thus far I have presented one attack on the logic of Hume’s argument: regarding induction he both begs the question and contradicts the empiricist foundations of his own outlook. I now present several more attacks on his logic.

Critique for Non-Falsifiability

The circularity I described above effectively turns Hume’s claim into a thesis that cannot be falsified. On his view, any report of a miracle is to be rejected out of hand as a hallucination, even without a shred of evidence for that—merely by virtue of being a report of a miracle. How, then, could one refute this thesis? Note that the refutation would have to be empirical, since Hume is operating within an empiricist framework. If we were to experience a miracle directly, we would supposedly have to reject it. And if we were to receive a report of a miracle, we would have to reject that too. How, then, could we ever show that Hume is wrong? This is a theory immunized against criticism (especially empirical criticism). Advancing such a claim in the name of scientific empiricism seems somewhat self-contradictory, doesn’t it?

From another angle: imagine a scientist reporting a surprising discovery— a fact that does not fit the laws of nature as we know them. How could he convince us, or himself, that he did not hallucinate? Hume’s proposal instructs him to abandon his discovery immediately, since hallucinations happen from time to time whereas miracles—violations of the known laws of nature—certainly cannot occur. Had we followed Hume’s counsel, we would be stuck with the science of Adam, with no ability to change a thing.

One might respond that a scientific experiment is repeatable—others can check the fact and confirm the observation—which is not the case with testimony about a miracle. That does strengthen the credibility of the testimony; but recall what Hume himself said: even if there are multitudes of witnesses, we cannot accept testimony about a miracle, since a deviation from (known) natural law is impossible, whereas the hallucination or error of many witnesses certainly is possible. When the probability on one side of the equation is 0, we must accept the other side, however small its probability.

In other words, Hume’s argument leads to absolute conservatism—that is, to rejecting any testimony or report of a deviation from the known laws of nature (i.e., of the implausible). This is precisely one of the charges leveled against Occam’s razor (see Column 426, §4).

Critique for Ignoring the Credibility of Testimony

To present my next critique, I will bring a story that happened to someone I know (also cited in Chapter 15 of Truth and Unstable). It concerns my good friend Amnon Levav, whom, after many years of acquaintance, I can attest is trustworthy and rational.[1] Amnon has worked professionally for many years in creative thinking and cognitive biases, and he is a very skilled facilitator of ideation groups in business, social, and other fields. Amnon prefaces the story by saying he hasn’t told it often, because he feels ashamed of his own part in it and because many do not believe him; but he swears it happened exactly as he describes.

The story goes roughly like this. His company provided pro-bono services to a city in the American Midwest. The goal was to find ways to improve relations among the city’s different populations, which had deteriorated to the point of violence. A group of sixteen people was assembled by the local organizers as a reasonable representation of all segments of the city’s population (on the basis of race, religion, sex, sexual orientation, and other parameters). It was a very diverse and heterogeneous group, all active and constructive citizens who cooperated well. The first day went as planned. At the start of the second day, everyone sat in a circle of chairs except for one person standing in the center. That person was asked to recall something from the previous day, and then to invite someone else to stand in the center in his stead and say something from yesterday, and then sit down, and so on. Each person, upon finishing, looked around, found someone who had not yet spoken, and invited them to the center. The process continued until the person in the center turned to the facilitator (Amnon) and said, “That’s it—we’re done. Everyone has spoken.” Amnon says he began to summarize and draw conclusions. His first remark (remember, this was a creative thinking workshop) was: “Why did you all assume that each participant should speak only once?”

Suddenly, one of the participants, an African-American woman, interrupted and said: “Wait a minute—none of the Black participants were invited to speak.” Everyone (including Amnon himself) looked around in astonishment, and it turned out she was entirely right. There were six Black participants sitting in the circle, and not one of them had been invited to speak. Until that woman pointed it out, those six themselves remained silent. It turned out that the participants—who came to foster reciprocity and consideration—simply did not see roughly a third of those sitting with them. Amnon concludes that this became the main lesson of those two days: how starting assumptions and vantage points can blind us to what is right before our eyes.

Amnon is secular and I am religious; for years we have been accustomed to debating questions of faith, especially about rationality, and arguments like Hume’s are our daily bread. When Amnon sent me this story, I replied as follows (from here on I am nearly quoting):

Indeed, it is hard to believe. But my conclusion differs from yours. There is another option: perhaps it didn’t actually happen, and you simply do not remember correctly (i.e., you are rewriting history—intentionally or inadvertently—or overlooking details that would change its meaning). In Hume’s terms, I would say that there are two interpretive possibilities for your story: (1) you all really failed to notice that roughly a third of the circle did not speak—i.e., your outlandish story records reality; or (2) you reconstructed the past in a tendentious way (innocently or deliberately), and the event never happened. Now I ask myself: why should I believe your present memory? Perhaps it never happened, and you built this memory or story because of contemporary ideological and moral fixations. How convenient to have such a story to illustrate an important moral lesson. There’s nothing like a good story (and in the absence of identifying details, it cannot be checked or verified). It is worth a thousand sermons.

In light of Hume’s argument about miracles, I added, I even have a decent argument for favoring interpretation (2): in option (1) the event is wildly implausible. The very fact that all six were not invited to speak—at an event meant to promote harmony among all groups—is highly implausible. In addition, there was a collective blindness of everyone else, every last one (including you yourself, the vigilant and experienced facilitator), who did not even notice that this occurred. This is downright unbelievable—almost a miracle. In option (2), by contrast, the event itself is perfectly plausible (all sixteen actually did speak and there was no blindness), and all we have is a case of forgetting or selective recoloring of memory (or even deliberate fabrication), and that in only one person (you). That can certainly happen to any of us. Which interpretation should I prefer?

One might object, in favor of option (1), that it offers an explanation for the forgetting: racial bias inducing blindness. But that seems doubtful to me; in my experience, I am not aware of such a sweeping blindness resulting from such biases. Yet even if I accept that such a bias exists, I can propose another bias that explains option (2) equally well: moral bias. You remember or describe the story that way, wittingly or unwittingly, because it helps you demonstrate a moral principle that is very important to you (and to me, of course). Again, alternatives must be compared. Even if something seems implausible, one must always compare it to the alternative; perhaps the latter is even less plausible. One may not dismiss an alternative that seems wrong before checking that the other alternative is superior.

What would you say about such an argument? This is, of course, not a miracle that violates laws of nature, but still a highly implausible event. It seems to me quite natural to apply Hume’s logic here. And yet, personally, I do not accept it. I do believe Amnon. What justifies that? Hume’s logic seems rock-solid, does it not? Biases, forgetting, or misrepresentation do occur from time to time; but such an event is very implausible.

It seems to me that here we must factor in our trust in the reporter. If we trust him, we will accept his report even when its content is very implausible. This is precisely how we accept scientists’ reports of surprising, paradigm-shifting experimental results; we trust their reports, and that trust outweighs our trust in the (current formulation of the) laws of nature. Is such a stance irrational? I think it is highly rational. Were a highly reliable and very critical person to stand before me and report a miracle he witnessed with his own eyes, I would investigate as best I could, but I would indeed consider believing him. Trust in people is not irrational, even if it is hard to quantify probabilistically. Here too I do not accept the inductive picture Hume selectively adopts. Our trust in people is nourished by our assessment of the specific speaker—not merely by statistics about how often people have lied or told the truth in the past. Trust in testimony and reports is an important factor; it can indeed compete with very low probabilities assigned to the reported content. Contrary to Hume’s assumption, that trust is not the product of empirical induction but of an immediate impression of the reporter. In Chapter 16 of my book Truth and Unstable I discuss the complexity of assessing a person’s credibility and show that it is not merely the result of cumulative experience. Most importantly: given some assessment of credibility, all our probability calculations must change. Probability calculations always occur within a certain information frame and concern the evaluation of alternative possibilities for the missing information. When the information we possess changes, the probability calculus must change accordingly (see more on this below).

In sum: we have seen that Hume’s argument begs the question by turning the thesis of “no miracles” into a non-falsifiable thesis. It leads to absolute and unreasonable conservatism that prevents us from drawing new conclusions and changing paradigms. The explanation for this flaw lies in our trust in the reporter. That is the way out of the loop in which Hume has imprisoned us. It is certainly possible that I will accept a report of a miracle—so long as the reporter is trustworthy. Contrary to Hume’s assumption, trust in human beings can outweigh our trust in the laws of nature and in experience—and this trust is not merely empirical-inductive.

Another Look at the Role and Meaning of Testimony

In Columns 226 and 228 I discussed the status of probabilistic evidence in law. Consider two witnesses testifying to a murder. Their testimony is generally accepted unless there are indications that they are lying or mistaken. One might wonder: why is such testimony accepted? The chance that the witnesses are lying or mistaken is not zero—perhaps it is, say, 3%. But the opposing chance—that a person committed murder—is much less than 3%. There are far fewer murderers than liars (in general, it is plausible that the number of offenders is inversely proportional to the severity of the offense). So why accept their testimony? Hume’s recommended comparison of alternatives would have us reject it.

This connects to my earlier critique about Hume’s ignoring the credibility of testimony, but it is sharper here. Compare it to another case discussed in those columns: in a city with two bus companies, one has 97% of the buses and the other 3%. One day a bus runs someone over and there are no witnesses. Is the statistical consideration sufficient to convict the larger company? In law, clearly not. In such a case there is no way to compel payment. Note that the probability supporting conviction is identical in both cases: 97%. And yet in the first case we convict on the basis of testimony, and in the second we do not convict on the basis of statistics. Why? In those columns I offered several explanations; but above all, regarding witnesses there are specific credibility considerations pertaining to the witnesses before us. It is not merely a question of the general distribution of liars among human beings. In the bus-company case, by contrast, we are merely comparing probabilities.

I add that on appeal, higher courts do not revisit the lower court’s assessments of witness credibility; they review only legal judgment. The reason is that immediate impression is considered the most important tool for determining credibility, and only those standing before the witnesses can assess it. We learn, therefore, that cases involving testimony should not be handled by mere probability comparisons. The assumption that the general distribution must determine our attitude toward specific witnesses in a specific case is not necessarily correct.

The Significance of Rarity — Another Look at Begging the Question

In my conversation with Jeremy he repeatedly argued that when a person goes to the sea on an ordinary day, he does not see the sea split in two; and when he approaches some mountain he does not see God revealed to him. Therefore, from experience, seas do not split and deities do not reveal themselves. These are miraculous events, and thus all reports of them should be rejected on the strength of our accumulated experience.

I replied that we are also familiar with rare astronomical and cosmological events. For example, the Big Bang occurred only once in the history of our universe. In addition, there are rare eclipses, stellar explosions (see, for example, here), supernovas, and so forth. Can one argue against a report of such an event that since I have never seen a star explode I should not accept reports of such an event? Since when is rarity a measure of plausibility? There are rare events that do occur, and were we not to accept testimony about them, we could never know of their occurrence. Rarity is not a clear indicator of low probability.

The conclusion is that rarity is not an argument. To say that some event is implausible or even impossible, one must bring direct—not merely statistical—support. The fact that God revealed Himself only once in history (or not at all, according to some) is not, by itself, an argument against revelation. The question is whether such an event is indeed implausible.

But whether an event like that is plausible depends on your assumptions. For my part, I have concluded—on philosophical (not empirical, and so Hume may dislike them) grounds—that there is a God who created the world, as detailed in the first four dialogues of my book The Prime Existent. Moreover, it seems to me quite plausible that He would reveal Himself to tell us what He wants of us (I explained that there as well and, briefly, also in my debates with Yadan and with Jeremy Fogel). Now a report reaches me that He revealed Himself and delivered a message about what is required of us. Why is that a miraculous event whose report I must reject? Because I have never seen God revealed? So what? I have never seen a supernova either. To the contrary: such an event is expected in advance, and when it occurs there is nothing more natural. True, on my view this happened only once; but, as noted, rarity is not by itself a refutation. We do not doubt the occurrence of a rare astronomical event because it is expected to occur; we can even compute that it will. The conclusion that such an event is expected, and the calculations themselves, can sometimes be based on observations and reports—and I will accept those, because the reporter’s credibility can outweigh the unusual, implausible content. Thus, divine revelation and the miracles He performs are not problematic for the believer. Their rarity is not, in itself, a refuting argument. Again, I am not equating revelation with a scientifically explained astronomical event; I am only exposing the weakness of Hume’s counter-arguments and showing that rarity is not necessarily a measure of low likelihood.

Another Look at Probability Calculus and Comparing Alternatives

From a different angle: probability calculations are always made within a given knowledge-frame. Within that frame we compare possibilities regarding what we do not know and compute probabilities accordingly. Therefore, before calculating, it is very important to make explicit the knowledge-frame in which the calculation is being done. Suppose someone claims that he rolled a die and got a 6 a hundred times in a row. Assuming a fair die, Hume would tell us that it is very reasonable to assume he is lying or mistaken, because the event is extremely implausible. By contrast, suppose we know from another source that the die showed the same face a hundred times in a row, but we do not know which number it was (was it one hundred 6’s, or 2’s, or 3’s?). Now if someone claims it was 6, we have no reason to reject his claim. Moreover, in such a situation we should not say that the chance he is right is 1/6, even though there are six possible outcomes. No—the chance he is right is almost 100%. Why? Because we have no reason to assume he is lying. If we knew nothing about the outcome and someone merely guessed “6,” then indeed we would say his chance is 1/6. But if a witness testifies that it was 6 (given that we already know there was some astonishing run of a hundred identical results), we accept his statement straightforwardly (i.e., with probability near 1). Here is the importance of testimony-credibility versus the probabilistic computation of its content.

Similarly, when we know that God exists, and then someone testifies that God revealed Himself to him or performed a miracle for him, we have no reason to doubt his testimony. True, a miracle or revelation is a rare event; but given the prior information, the rarity is irrelevant. In such cases, and within the information-frame as we have it (according to the believer’s assumptions), Hume-style probability comparisons are inappropriate. Hume can reply that he rejects God’s existence and therefore rejects the credibility of such testimony; but that merely states his (legitimate) position. Such question-begging cannot serve as a refutation of a believer’s position.

Consider someone who testifies before you that he saw a rare eclipse. The probability of such an eclipse is minuscule. Should we reject his testimony? Not at all. Assuming such an event can occur, and if the person is reasonably trustworthy, we accept his testimony even though its content is implausible. It is not appropriate to compare the probability of the content of testimony with the probability that the witness is lying or mistaken.

A nice example appears in a Talmudic case (cited in Column 226). Reuven sells an ox to Shimon; Shimon goes out to plow with it but cannot, because the ox is wild and gores. He claims against the seller that the ox was found to be a gorer and demands his money back. Reuven responds that he sold it for slaughter, not for plowing, so its being wild is irrelevant. Who is right? The Amoraim dispute this, since most sales are for plowing, not slaughter. Nevertheless, the law follows Shmuel that the majority does not determine the ruling in this case, even though in cases of doubt halakhah usually follows the majority.

R. Shimon Shkop explains: indeed, in most cases an ox is sold for plowing—but sometimes it is sold for slaughter. That is not impossible (it is unlikely, but not miraculous). If so, when someone claims he sold for slaughter, why should we doubt him? If I tell you that my height is 1.95 meters, should you reject my claim because that height is very rare? Clearly not, since there are people of that height. I am telling you I am one of them; there is no reason to doubt me—and certainly not to conclude I am lying or mistaken because the probability of that height is low.

When I believe in God, a miracle or a revelation is not impossible. On the contrary, there is no principled obstacle to its occurrence, even if such events are rare. When testimony of such occurrences reaches me, I have no a priori reason to doubt it—unless the testimony itself seems suspicious. Here I am dealing with Hume’s argument rather than specific claims about the credibility of particular traditions. Hume, who approaches this via probability-comparisons and assigns probability 0 to the occurrence itself, begs the question—or at least reasons within a different informational frame than the believer. His claim may be acceptable on its own terms, but it cannot refute the believer’s position.

Critique of Hume’s “Distance of Testimony” Claim

Hume claims that reports of miracles always come from regions of primitive peoples with poorly developed critical thinking. In more educated regions such reports are very rare, and when they appear they stem from traditions that began in more primitive locales. In my view, this claim contains a problematic assumption—and may actually point to an explanation that refutes Hume’s thesis itself.

We saw above that Hume begs the question and in effect posits a thesis that cannot be put to tests of refutation. A person who thinks like Hume will never accept testimony of a miracle, for he will always dismiss it out of hand as a report of something impossible, and at most will rationalize it (some error that led the witness to think he saw a miracle). But that, in turn, undermines Hume’s claim. Now I will advance a similar point against his “distance” claim: perhaps this is precisely why we do not find miracle reports in cultures with developed critical thinking. In such cultures, even if such reports emerge, they will be rejected out of hand because of an empiricist mindset like Hume’s. No wonder, then, that we do not find such reports there.

In L. Frank Baum’s The Wonderful Wizard of Oz, a strong wind carries Dorothy and her house from Kansas to the Land of Oz, full of wizards and witches. When she lands there she meets a woman who introduces herself as the Witch of the North. A fascinating conversation ensues about cultural differences:

“Aunt Em told me that the witches were all dead—years and years ago.” “Who is Aunt Em?” asked the little old woman. “She is my aunt who lives in Kansas, where I came from.” The Witch of the North seemed to think for a time, with her head bowed and her eyes upon the ground. Then she looked up and said, “I do not know where Kansas is, for I have never heard the name of that country before. Tell me, is it a civilized country?” “Oh, yes,” replied Dorothy. “Then that accounts for it. In the civilized countries I believe there are no witches left, nor wizards, nor sorcerers, nor magicians. But, you see, the Land of Oz has never been civilized, for we are cut off from all the rest of the world. Therefore we still have witches and wizards among us.”

In my book Two Carts and a Hot-Air Balloon (Part II, Gate V, ch. 3) I quoted this passage to explain the advantage “primitives” have over the cultured when it comes to a willingness to accept unusual and supernatural phenomena. This naturally comes with a disadvantage—too ready an acceptance of such reports—but at least they do not slam the door on them as Hume and his empiricist heirs do.

Perhaps this is what Churchill meant in the words I placed as the epigraph to Chapter 16 of Truth and Unstable:

The best lesson in life is to know that even fools are right sometimes.

At the very least, we ought to keep open the possibility that they are right.

A Personal Note

This brings me to a personal question. The typical reader already knows me and knows that I tend to doubt reports of mystical events. I share Hume’s skepticism. So why do I accept the reliability of the tradition about Sinai or the biblical miracles?

My answer is that indeed, sometimes fools are right—but they remain fools. My initial inclination is skeptical, and in many cases I can offer an explanation that drains such reports of their “miraculous” content. But that does not mean I will dismiss every such report out of hand, because I remain open to the possibility that my skepticism is mistaken—at least in some cases. My thesis is open to refutation. If a reliable tradition reaches me, passes through various tests, and my impression of the witnesses is that they are credible, I will accept their testimony. A mass testimony transmitted carefully and critically, surviving many generations, accompanied by a book given at that very event, with no discernible interest in inventing it, and with evidence pointing to the antiquity of the events—this inspires my trust, especially against the background of the assumption that such a revelation would be expected even had no tradition reached me.

Of course, one can argue (and, knowing the cast of characters, I am sure the comments will do so), but I will not elaborate here, as that is not the topic. In this section I only wished to explain that my words against Hume do not contradict my skeptical stance toward mysticism and miracle reports in general. I am not locked in any direction. I am simply skeptical.

Conclusion

A common claim among defenders of tradition is that it is very hard to deceive multitudes of witnesses and implausible that they would all transmit a false testimony to their children. But that only says that the probability of such a disruption is very low. One must not forget that, against it, stands an alternative whose probability is 0 (since the events contradict the laws of nature). Therefore Hume argued that we should nonetheless adopt the second alternative.

In Chapter 15 of The Science of Freedom I placed as an epigraph a saying of Sherlock Holmes from The Sign of Four that captures precisely Hume’s argument:

Once you eliminate the impossible, whatever remains, no matter how improbable, must be the truth.

That is exactly Hume’s claim: if one of two alternatives is impossible, then the second—despite its implausibility—must be true.

At the beginning of that same chapter I placed another epigraph, this time from Lewis Carroll’s Alice in Wonderland:

Alice: “There’s no use trying—one can’t believe impossible things.” The White Queen: “I daresay you haven’t had much practice. When I was your age, I practiced for half an hour a day. Why, sometimes I’ve believed as many as six impossible things before breakfast.”

This, I think, is a fitting Zionist response to Hume’s argument.

In the next column I will address Hume’s own doctrine—what conclusion he himself drew from the arguments discussed thus far.

[1] The story appeared under the title “Blind” on his company’s website, SIT (dated April 27, 2010): http://www.sitsite.com/blog/2010/04/blind/#more-262. Since then it has disappeared from there for some reason.