What Can We Infer from the Present about the Past? (2006)
In “Do No Miracles Today Imply None in the Past? A Critique of Richard Carrier’s Methodology,” Amy Sayers responds to an argument in my collection of essays “Why I Don’t Buy the Resurrection Story” (6th ed., 2006). In particular, she responds to the argument I summarize under “No Miracles Today Implies None Then” (part II of the chapter “General Case for Insufficiency: The Event is Not Proportionate to the Theory“).
First, Sayers claims that an “argument of analogy between present and past cannot work in the negative sense,” yet all negative propositions entail positive propositions, and vice versa, eliminating the merit of any such distinction. For example, I can say “if all the events I observe or know about today have natural causes, then as far as I know all events had natural causes then,” which is a positive analogy (and she grants such analogies in principle can be valid), and yet this analogy logically entails the concurrent analogy that if there are no miracles today, then probably there were none then. So her argument is illogical from the start.
Moreover, the notion that negative analogies are invalid is also unfounded. Clearly we can argue, for example, “sail-powered ships cannot travel a hundred miles per hour today, therefore they probably couldn’t then” as a basis for concluding that a sailboat could not have circumnavigated Africa in a day. Likewise, we can argue that “the earth is not flat today, so probably it was not flat then” as a basis for concluding that tales of people falling off the edge of the earth are not true, or that “human beings cannot breathe in outer space today, therefore they couldn’t then” as a basis for concluding that Lucian’s report of taking an open-decked sailboat to the moon cannot be true. Obviously, negative analogies are valid and we use them all the time. Sayers is simply wrong to claim otherwise.
Sayers gets this wrong because she commits the fallacy of false generalization: she presents one invalid negative analogy and argues from this that all negative analogies are invalid. But, in fact, the one example she gives would be invalid even if it were a positive analogy. She selects the invalid argument “there are no dinosaurs today, therefore there were none then,” but this argument is not invalid because it is negative—for the positive analogy “there is paper money today, therefore there was paper money then” is also invalid for exactly the same reason: both arguments ignore relevant facts. As Sayers herself points out, “we have evidence that dinosaurs once existed” and “we have reasonable theories explaining why they don’t any longer,” and similar observations hold regarding paper money. Sayers claims that this shifts the burden of evidence, but that has nothing to do with whether negative analogies are valid arguments. Rather, all it has to do with is when analogies are valid, regardless of whether they are positive or negative.
Consequently, Sayers completely misconstrues the argument from analogy when she presents her own deductive syllogism, which is an incorrect formulation of the argument. An argument from analogy proceeds from observations of relative probability, and is not about what actually is the case, but what we are warranted in believing. This is a crucial distinction. For example, we know for a fact that people fabricate stories of the fabulous, but we do not know that anyone could plant the vast, worldwide geological evidence of the past existence and extinction of dinosaurs (much less that anyone has actually done this). Essential to this observation is that we proceed from what we know, not from what we don’t know. Just because we don’t know of anyone who could plant all the evidence for dinosaurs does not mean no such person exists. It is a fact that all that evidence could be planted and no one would know about it. But we don’t know that this is the case, and since we can only proceed on what we know, we are not warranted in believing that the evidence was planted. And because we otherwise do have good explanations of the evidence, we are warranted in believing it was not planted—until such time as we discover new evidence that justifies changing our mind about this.
That is why the dinosaurs analogy does not compare to the miracles analogy. We can only argue from what we know, and all we know is that (1) miracles don’t happen around us and (2) miracle stories are often false. In other words, though it may indeed be the case that miracles once happened, or are happening now, we do not know that this is the case, whereas we do know that miracle stories are often false, and that no such things are happening around us. And the bottom line is that we can only proceed from what we know. So until we know something different, it is entirely valid to reason that past miracle stories are probably not true, by analogy from the present. In no way does it matter whether that analogy is negative or positive.
However, Sayers is correct to object to the blind use of this argument as an excuse to ignore evidence, and I think that is the point Sayers actually wanted to make. For it is indeed possible to have evidence of past miracles that is as strong as the evidence for past dinosaurs, and in such a case we could not argue from analogy against the past existence of miracles. I have always maintained this position. So I think what Sayers really means to argue is that an argument from analogy (negative or otherwise) only works when there is not a greater weight of evidence to the contrary. For example, we could find extensive archaeological evidence that Lucian in fact did travel to the moon: astronauts could find “Lucian was here” scrawled on the moon in ancient Greek, Lucian’s house could be dug up and found full of lunar rocks, abundant and reliable eyewitness testimony could be on record of his flight, and so on. In the same fashion, it is possible to acquire convincing evidence of miracles in past history, and I argue this very point in my book Sense and Goodness without God (2005).
But then Sayers goes too far when she misconstrues the often-heard premise “there is no evidence of miracles today” as meaning something other than “there is insufficient evidence to warrant my believing there are any genuine miracles today.” She is close to getting it right, but not quite. There may indeed be evidence available to someone else that is sufficient to warrant them believing in genuine miracles, but until the atheist is provided with that evidence, he can only reason from what he himself knows. So if there is no evidence of miracles today that is sufficient to warrant his believing there are any genuine miracles, then he can argue from that premise. To refute his argument is simple: provide the evidence that his premise is false. But beyond that contingency, Sayers cannot claim it is invalid to adopt that premise.
For instance, there could be evidence of Lucian’s trip to the moon that Sayers doesn’t know about, and she certainly hasn’t exhausted all possible research on the matter, not even what she is capable of. Nevertheless, Sayers could still rightly conclude that Lucian made no such trip based on the fact that we have no evidence today that such a trip was possible given the resources available to him. Her conclusion would be that, given what she knows at the present time, she is not warranted in believing Lucian went to the moon. But she would probably express this as simply “Lucian didn’t go to the moon,” leaving it as understood what she actually means by this. So, too, for those who argue by analogy against miracles.
This brings us to her specific discussion of my presentation of the argument, where she commits all the same errors. Right from the start, she incorrectly formulates my argument as the following deductive syllogism:
- If dead people are not resurrected today, they were not resurrected in the past.
- Dead people stay dead today.
- There is no evidence that a God exists who would or could perform a miracle that interferes with the dead’s state of staying dead.
- Therefore, Jesus, who died “back then,” stayed dead, too.
This is not at all how my argument proceeds. A correct formulation of my argument would look something like this:
- If I have insufficient evidence to believe miracles like resurrections happen today, and if I do not have better evidence for miracles in the past than I have for their absence in the present, then I have insufficient evidence to believe miracles like resurrections happened in the past.
- I have insufficient evidence to believe miracles like resurrections happen today and I do not have better evidence for miracles in the past than I have for their absence in the present.
- Therefore, I have insufficient evidence to believe miracles like resurrections happened in the past.
This argument is valid and sound, and its conclusion would logically encompass all particular cases like Jesus.
Clearly, Sayers has gotten my argument wrong in many ways. She mistakes assertions about what I am warranted in believing with assertions about what is the case apart from what I know. Indeed, contrary to her allegation, in the relevant essay I explicitly deny that I need evidence of resurrections specifically, and instead seek evidence of any comparable agent or phenomena that would make resurrections both possible and more likely than other explanations of the same evidence. Sayers also gets wrong what I said about God: I did not mean “there is no evidence that a God exists who would or could perform a miracle” as if there could never be, but rather that I myself, at the present time, have no evidence that there is such an agent. Obviously, if I get some such evidence, then I might very well change my mind about the resurrection, or the miracles of Asclepius, Mohammed, or Krishna, or any number of other past miracle claims. Sayers also gets wrong what I said about God: I did not say that “there is no evidence that a God exists who would or could perform a miracle.” Rather, I said that I myself had no evidence that there is such an agent. Maybe someone else has that evidence, or maybe there is some such evidence out there waiting to be found, but until I have it, I can only reason from what I know. And from what I know, when I ask myself “Which is more likely: that God stopped parting seas and raising the dead, or that these stories are, for various historical reasons, fictions?” I am compelled by what I know to conclude that the latter is more likely. I could be wrong. But until someone proves me wrong, I cannot warrant believing otherwise. That is my argument. And Sayers says nothing against it.
As I conclude in the relevant essay, “we do not believe stories that come to us second-hand which contradict our direct experience, because each fact [the second hand story, and our own direct experience] presents us with two possible realities, the only evidence of one is a story, the only evidence of the other is direct observation.” We know stories are more often mistaken or untrue than our direct experience, and our direct experience provides no support for these sorts of stories. If our direct experience instead provided some support, then we would have an argument from analogy that would support rather than count against the case for past miracles. But since our direct experience produces the opposite information, it produces the opposite support—and that is where things will remain until such time as our experience changes.
But that isn’t the only way to overcome the argument. For we also know of conditions when stories and other evidence are less often mistaken or untrue than our direct experience, so we can also construct an argument from analogy in support of a past miracle whenever those same conditions obtain. The most common example is the heliocentric theory, which flatly contradicts direct perception since the Sun “obviously” moves while the Earth sits still. Though anyone can directly perceive all the facts that entail the truth of the heliocentric theory (and many a scientist and enthusiast has done so), by far most people only know the facts supporting the theory through testimony (of experts, teachers, TV shows, authors, etc.). Thus, it is clearly reasonable for testimony to overcome direct experience when that testimony is sufficiently overwhelming. But the amount, quality, and degree of testimony required for this reversal is vastly greater than what we have for any miracle claim in history, and thus there is no parallel.
But there could be. If we only had the same scale of testimonial evidence, in quantity, scope, and reliability, for any miracle claim, then that claim would be as believable as any other amazing but credible story. Hence I agree we could still overcome the inference against miracles with a substantial quantity of evidence from the past, as long as it was sufficient to justify concluding that our present direct experience is abnormal. As in my example above, we could indeed prove beyond a reasonable doubt that Lucian went to the moon, or we could at least raise adequate suspicions that dinosaur bones were planted. But as my other chapters demonstrate in “Why I Don’t Buy the Resurrection Story,” and as my concluding point about Thomas emphasizes in “No Miracles Today Implies None Then,” we do not have that kind of evidence from the past, certainly not in the case of any miracle claim.
In conclusion, Sayers ends with an argument that some skeptics are irrational, which is certainly true. But that has no bearing on what would persuade me, and therefore it can have no bearing on anything I have argued. Nor can it pertain to any other reasonable person. How I would respond to irrational people is not relevant to whether I believe the evidence is sufficient to convince a rational person. The fact remains that I don’t. And Sayers has not said anything that justifies changing my mind about that.
 See my essay “Proving a Negative” (1999).
 Fabulous stories are fabricated for many different reasons, including honest mistakes, innocent exaggerations, symbolic exposition, and even outright deception or delusion.
 See my most complete discussion of the subject of miracles in “Review of In Defense of Miracles” (1999). Note that I disagree with interpreters of Hume who think it is logically impossible to have sufficient testimony to accept any miracle, largely because Hume did not define “miracle” correctly.
 The proper formula for the argument from analogy is Bayes’ theorem, which is too complex to explain here. But for an informal treatment, see my Digression on Method in “Why I Am Not A Christian” (2006) and The Problem of Uncertainty in “Was Christianity Too Improbable to be False?” (2006).