Wednesday, January 13, 2016

Scoring the Moore/McGrew debate on miracles


With permission, I'm posting Dr. Timothy McGrew's response to my evaluation:

Yes, I meant the "net" evidence -- allowing that there may be some evidence against a proposition P, but if there is a greater weight of evidence in favor of it, then that positive evidence overbalances the negative. 
I would count moral experience as very strong, possibly decisive, evidence against atheistic naturalism. The only reservation I would have about your stronger statement is that it is not completely clear to me that atheistic moral Platonism could be ruled out. But again, as J. L. Mackie observes, moral facts in a godless universe would be very queer facts indeed. 
Regarding 3, I took that stance since (a) a large proportion of the people present would not have claimed to experience a miracle and (b) I never have (to my knowledge).  
I think your criticism 4 shows a misunderstanding of how I'm using the filter. It doesn't "preemptively exclude" things that don't pass through it, that don't, as I elsewhere phrased it, "make the first cut." Rather, it suggests that those are not promising places to make a first inquiry. Later, they may come back into focus because of their connection with other kinds of evidence, probably because they are connected to the resurrection. I did make that point in passing later in the discussion. 
On 5, there are religious environments where the religion is not established but rather newly fledged. Christianity and Mormonism are the only two examples I am aware of (with the latter clearly derivative) of large world religions founded on miracle claims from the outset. 
On point 6, without denying that such things might happen simply to meet an individual need, I'm very cautious, partly because I believe (rightly or wrongly) that I've seen some people fool themselves about private miracles, partly because I am mindful of Luke 4:25-26.

Timothy McGrew and Zachary Moore recently debated the question: "Could it ever be rational to believe in miracles?":

In this post I'm going to summarize and score their performance. As a rule I don't watch philosophical/theological debates. It's an inefficient way to present and process information on complex issues. And it's more cumbersome when I have to take notes.

I've seen the one debate once through, and I've repeatedly listened to particular statements. It's possible that I missed a key point. 

This was a three hour debate with opening statements, rebuttals, cross-examinations, Q&A (from the audience), and closing statements. There are two ways I could summarize the debate. I could offer a running summary of what was said in sequential order. That, however, would result in a very disjointed summary. In the course of the debate, Moore and McGrew stated their positions, revisited the same issues, introducing explanations, clarifications, and qualifications to their initial statements. 

It would be very choppy and repetitious to offer a chronological summary. The order would be disordered.

In the interests of coherence, I will reorganize the material to group together statements of the same kind. My summary will combine different statements on the same subject to give a compact, qualified statement of their respective positions. I will sometimes paraphrase what they said, but I will frequently use their own words. Anyone can watch the original debate to compare my summary with the verbatim proceedings.

The formal question to be debated determines the burden of proof. Winning or losing depends on how well the respective debaters discharge their burden of proof in reference to the question under review. There may be many interesting or important ancillary questions to be pursued, but a responsible debate performance will stick to the precise question at issue and resist the temptation to stray from that path. 

I. Summary of McGrew's position

1. Defining terms:

i) "Miracle": an event that would not have happened if the natural world was left to itself, as opposed to outside agency (i.e. divine intervention).

ii) Natural order: the interaction of physical agencies, as well as the actions and interactions of agents (humans, animals) with abilities much like ours. 

iii) "Rational": 

a) Follow the evidence wherever it leads

b) Seek available evidence

c) Have reasonable rules of evidence

d) A high cost of getting it wrong

iv) "Could": irrational to disbelieve in miracles no matter how much evidence there is in their support.

2. Hume did not regard this as irrational because he held there was always at least as much or more evidence for the unbroken laws of nature. Our confidence in natural laws is as certain as any empirical belief can be, based on extensive, invariable experience. That's the strongest possible evidence

Unlike testimonial evidence, the laws of nature are unfailingly true. At best, testimonial evidence for miracles can only equal, and never exceed, the evidence for natural laws. In practice, evidence for miracles is always weaker than evidence for miracles. 

However, skepticism regarding testimonial evidence for miracles boomerangs on Hume, since the evidence for natural laws is, itself, dependent on testimonial evidence. To object to miracles on scientific grounds is self-defeating, for that undercuts science no less that miracles.

3. Scientific probability is based on what nature does when left to itself. When nature functions as a closed system or isolated system, when there is no outside intervention. That's an implicit rider on scientific probabilities.

For example, a rock will normally roll downhill. If you see a rock rolling uphill, that's evidence of intervention from an external agent.

Outside intervention changes the way nature behaves. So the probability of miracles depends on whether we have good reasons to believe the system was not left to itself in that instance. 

4. The scientific method is like a metal detector: very good at what it was designed to uncover, but the fact that a metal detector can't find a lost contact lens does nothing to prejudge the existence of the contact lens. 

5. The regular course of nature is a necessary backdrop for the recognition of miracles. 

Although McGrew didn't explicitly say so, the implication is that natural laws, far from being incompatible with miracles, are a prerequisite for miracles. Miracles require that point of contrast to stand out as divine signs. 

6. Due to the multiplicity of reported miracles, we need a filter (rules of evidence) to isolate and identify the best candidates:

i) Distant in time. If the first report of the alleged event isn't at or near the time of the event, its credibility is diminished. 

Belated reports are hard to check up on if the reporter can't contact a witness on the ground. 

ii) Distant in space. If the reporter of the alleged event wasn't at or near the place where it reputedly occurred, its credibility is diminished. 

Was the reporter close up to the event in time and space, either directly or via access to eyewitness testimony?

iii) Statistical noise. Events that are consistent with either a natural or supernatural explanation. 

iv) Trivial events. Allegedly weird events that serve no rational purpose.

v) Self-serving events. Does the reporter have something to gain (e.g. Joseph Smith)? 

vi) Events that confirm a preexisting belief system. Adherents are predisposed to believe it. They aren't motivated to verify it even if it never happened. 

By contrast, testimony to the Resurrection took place in the teeth of the Judeo-Roman establishment. 

If a reported miracle makes the first cut, it graduates to the next criterion. If it fails at any stage along the way, it merits no further consideration. That's where your preliminary investigation ends. 

7. Assuming a reported miracle survives (6), it must meet additional criteria:

i) A public event. Multiple witnesses.

ii) An observable event. 

To function as a divine sign, it must be observable.

iii) An early record (e.g. memorial) of the event.

iv) Distinguish optimal eyewitness testimony from unreliable eyewitness testimony. 

8. These are not sectarian criteria, but generic, common sense criteria. 

9. McGrew's filter screens out most reported miracles. Screens out all ecclesiastical miracles, and many or most Biblical miracles. Indeed, McGrew said only about 5-6 candidates survive. 

That's not a problem since, at this stage of the argument, the objective is not to determine the prevalence of miracles, but to determine whether any miracles occur. To achieve that modest aim, a fine-mesh filter is adequate.

10. Although he didn't flesh it out, at one point in the debate he suggested that once you establish certain anchor miracles (my term), you can use that frame of reference to go back and render other reported miracles more credible, even if they were caught in the filter. 

11. Given the multiplicity of reported miracles, the question is where to start. The purpose of the filter is to narrow down the search parameters to manageable proportions. What reported miracles are good candidates to establish whether that kind of event ever takes place? What reported miracles furnish a good starting-point in your investigation? That's the purpose of the filter. 

The filter intentionally eliminates many candidates that may indeed be bona fide miracles. The purpose of the filter is to establish a lower threshold, not an upper threshold. (That's my interpretation of McGrew's position.)

12. To function as divine signs, miracles must be rare. 

By the same token, it is reasonable to demand greater evidence for reported miracles than ordinary events. 

13. According to Scripture, God hasn't salted miracles across history, waiting to be discovered. Rather, they cluster around three different periods in time and place: the Mosaic era, the ministry of Elijah and Elisha, the ministry of Christ and the Apostles. 

14. In assigning background probabilities, we must use the same reference class rather than mixing reference classes. To say the Resurrection has low prior probability due to the base rate for dead men coming back to life is confused. In that case, the base rate has reference to what happens when nature is left to operate on its own. But the Resurrection is predicated on outside intervention.

You can calculate the trajectory of a cannon ball based on natural laws. You can't project the trajectory of a human agent based on natural laws. As a personal agent, where he goes isn't dictated by natural laws. 

15. It's a methodological error (my phrase) to stipulate  a rule of evidence that walls you off from reality. If you were wrong, you will never discover it because you refuse to accept a certain kind of truth even if it really is true. That puts you out of reach of evidence. You block it out by definition. In the words of William James, a rule of thinking that guarantees we won't wind up in certain places. 

16. In his opening statement, Moore ran through several kinds of reported miracles. However, none of these constitutes a counterexample to McGrew's position because his filter screens out Moore's examples. They just aren't germane to McGrew's position. 

Likewise, McGrew allows for the possibility of demonic miracles.

17. In answer to a question from the audience, McGrew said it's not incumbent on Christians to explain how  supernatural agents can interact with the natural order. That's not an issue unique to miracles. As a general philosophical issue, causation at a direct level is puzzling. Eventually, any causal explanation will bottom out where one thing just does cause another with no further level in-between them. But because some things apparently do cause other things, and because that has explanatory value, we grant that assumption. It allows us to account for why some things happen. 

II. Summary of Moore's position

1. Different religions report the same kinds of miracles. Moore attempted to draw parallels between Jacob's ladder and Muhammad's midnight journey (on a winged horse) to Jerusalem, a eucharistic miracle involving a skeptical 8C Brazilian monk compared to a Hindu miracle about Ganeshi, Balaam's donkey compared to a Hindu miracle about a water buffalo reciting the Gita, a Jewish miracle (Honi the circle-drawer), Hanukkah, resurrection miracles attributed to St. Nicholas, Lourdes, and Fatima (the "miracle of the sun")–witnessed by thousands, and transubstantiation. 

2. Every new religion makes miracle claims. 

3. McGrew's filter is an ad hoc filter.

4. Roman Catholicism has its own filter. Which one is right?

5. There is no objective standard of comparison to distinguish credible miracle reports from incredible miracle reports. No positive control. Nothing like the one kilogram platinum ingot that's the base unit of mass for the international metric system. 

6. Reported miracles suffer from a type 1 error: too many false positives. The error rate is overwhelming.

7. Reported miracles suffer from a type 2 error: too many (undetectable) false negatives. 

8. The background probability for miracles, even assuming God exists, is vanishingly low. 

9. How, in principle, could we even detect supernatural agency? That's inaccessible to sensory perception.

10. A cumulative case strategy hits a wall. Multiple lines of evidence require you to evaluate each piece of evidence separately. That successively lowers the overall probability because you have more things to independently prove. 

11. He accused McGrew of committing the post hoc fallacy and/or sharpshooter fallacy. Here's a definition:

An analysis of outcomes out of context that can give the illusion of causation rather than attributing the outcomes to chance. The Texas sharpshooter fallacy fails to take randomness into account when determining cause and effect, instead emphasizing how outcomes are similar rather than how they are different.

III. Assessment of McGrew's position

Needless to say, I'm far more sympathetic to McGrew's position than Moore's. It was that way going in, and that way coming out. To be won over by Moore's position was never a live option. For one reason, I have my own considered position on miracles. 

It's very praiseworthy that McGrew is bringing his expertise to the general public on this all-important topic. That's beneficial to believers and unbelievers alike.

That said, I have some reservations about a few things McGrew said. Of course, given the constraints of the debate format, he had to keep many things in reserve. 

1. "Following the evidence wherever it leads" is a good rule of thumb, but I don't think that's absolute. There are times when it's rational to believe something despite evidence to the contrary. For instance, there may be conflicting evidence. Perhaps McGrew's statement is shorthand for the "preponderance of evidence" or something like that.

2. In addition, I rule out atheism in advance. If a position subverts moral realism, subverts human value, and/or subverts human reason, then I don't take that seriously. My investigation would be confined to finding out if the position in question has those catastrophic consequences. If so, I look no further in that direction. 

It's like que sera sera fatalism. Suppose someone hexes me. Tells me I will die a horrible death before the age of 30. 

Well, that will either happen or not. If I'm doomed, there's nothing I can do about it, so there'd be no point proving I'm doomed. There'd no point writing a philosophical defense of fatalism. Believe it or not, if fatalism is true, it makes no difference what I think or do or refrain from doing about it. So that's a waste of intellectual energy. There might be some value in disproving fatalism, because that would make a difference–if, indeed, fatalism is false. 

Same thing with comparing atheism to Christian theism. These aren't symmetrical alternatives. Not even close. 

3. The debate was conducted as if everyone's source of information about miracles is secondhand information. The only evidence for miracles is testimonial evidence.

Certainly that covers a major subset. But in discussing the rationality of belief in miracles, you have many people who say a miracle happened to them, or happened in their presence. We need to distinguish between what's rational to believe in the case of firsthand experience and secondhand information. 

For instance, suppose I have a dream. At the time it seems like an ordinary dream. But then it comes true. The dream was very specific. It's highly unlikely that it was just a coincidence. In that case, the dreamer is warranted in believing his dream was a premonition.

What if he tells his best friend about the dream? Well, that depends. If he tells his best friend about a funny dream he had last night, then a day later, both of them witness it come true, then I'd say the dream has the same evidential value for the second party (if the dreamer gave his friend a detailed description).

What if he tells his friend after the fact, or his friend isn't there to witness the dream come true. In that case, the friend might be justified in believing the dream, but on a weaker basis. By the same token, he might be justified in withholding judgment. It didn't happen to him, and it wasn't a veridical dream for him. 

4. There's potential tension between McGrew's filter and his objection to methodological atheism. He faults atheists for discounting certain kinds of evidence or certain kinds of truths in advance, yet his own filter preemptively excludes reported miracles which don't meet the criteria, even though–by his own admission–that may screen out many bona fide miracles. 

Likewise, for reasons I've stated on other occasions, I don't think reported miracles have a higher burden of proof. But I won't repeat myself here. 

Perhaps, though, he'd just say the point of the filter is to establish a foothold for miracles. It's intended to eliminate reasonable doubt by its focus on some index miracles (my term). 

5. One of his criteria seems to be in tension with his statement that if there were any real miracles, we'd expect them to occur, or occur with greater frequency, in a religious environment. That's where we'd expect them to happen if they happen at all. Yet his filter screens out reported miracles that confirm established opinions. 

I'd add that this particular criterion of his filter is quite similar to the notorious dissimilarity criterion in Gospel criticism. 

6. I agree with McGrew that when a miracle functions as a general sign, it must be evident. But in principle, miracles can function as signs for individuals. We might distinguish between a public miracle to attest the Christian faith, and a private miracle to give an individual guidance or encouragement. 

Likewise, verification is not the only function of miracles: they can simply meet a personal need. Verification might be a fringe benefit.

7. We might compare veridical/inverdical miracles to verdical/inverdical NDEs. The existence of veridical NDEs will establish that this kind of event occurs. 

If that can be established, then it raises the probability that some inverdical NDEs are true. After all, veridicality isn't what makes it true. Rather, veridicality furnishes independent evidence. But veridicality depends on a particular setting, particular circumstances, which are incidental features of NDEs. It's rare that NDEs would take place in that setting. So the mere fact that most NDEs are inverdical isn't prejudicial to the reality of the experience. 

And it's subject to the same degrees of certainty or uncertain as my example regarding premonitory dreams. 

IV. Assessment of Moore's position. 

1. His position reminds me of the celebrated debate between Clifford and James. Is the priority to believe fewer errors at the risk of believing fewer truths, or to believe more truths at the risk of believing more errors? 

2. A besetting problem with Moore's performance was his systematic failure to adapt his argument in light of McGrew's filter. Moore did nothing much to advance his original argument. He kept reciting the same talking points despite the fact that McGrew's filter, if valid, moots nearly all of Moore's talking points. Moore was caught off guard by McGrew's position, and he had nothing to fall back on. It reminded me of Bart Ehrman's ill-fated debate with W. L. Craig, where Ehrman walked right into an ambush, and was bleeding to death for the rest of the debate. Likewise, Moore had nothing in reserve when McGrew preempted his prepared argument. After McGrew's opening statement, much of what Moore said in response seemed to be stalling for time. 

3. Apropos (2), Moore treats McGrew's criteria, or any criteria for miracles, as arbitrary. But McGrew's criteria for miraculous events aren't essentially different from criteria for historical events generally. The main difference is that his criteria are more stringent in some respects, but an opponent of miracles should hardly find that objectionable.

Most of McGrew's criteria are stock criteria for assessing the credibility of eyewitnesses and sifting testimonial evidence. Much of this is what a historian would apply to accounts of Caesar crossing the Rubicon. It isn't something McGrew cooked up just for miracles.

Moore acts as though, unless criteria for scientific or historical knowledge fell from the sky, they are arbitrary. But the criteria for scientific and historical knowledge are necessarily philosophical criteria which humans devise. There's nothing inherently suspect about that enterprise. And it's unavoidable. The alternative is radical skepticism regarding the possibility of scientific and historical knowledge. And that's counterproductive for Moore's own position.

4. He thinks to be any good, a criterion must be objective in the sense that the platinum ingot is an objective standard. But that's a poor example, for that's an arbitrary social convention. It isn't even a criterion for truth. It's simply convenient. 

But you can't extrapolate from an artificial standard for weights and measures to historiography. Realty is independent in a way that the metric system is not. Our criteria must be suited to the nature of reality, and not imposed on reality.

Take the difference between experimental and anecdotal evidence. It's often useful to study things in a controlled setting where you can eliminate irrelevant factors and reproduce results. But some people mistakenly make that the ideal of scientific knowledge or knowledge in general.

Yet most of what we know about the nature world is based on field work rather than lab work. A geologist studies volcanic action as it happens in nature. 

Likewise, a biologist may study animal behavior in the wild. Indeed, to study animal behavior in a controlled environment may be misleading precisely because animals often behave abnormally in unnatural settings. To understand animal behavior, you generally need to study them in their natural surroundings. 

The same holds true for historical events in general or miraculous events in particular. We must take them as they come to us. When, where, and how they occur isn't something we can ordinarily dictate. 

5. An indetectable agent can produce detectable effects.

6. Moore never absorbed the crucial distinction between natural processes and personal agency. Take an automated car. Presumably, this will be common in the future.

Some automated cars will be involved in fatal accidents. In most cases, that will be due to mechanical error.

However, some of them may involve murder. The killer hacked into the computer system and caused the fatal accident. Statistically, that might be improbable, but a homicide detective can't just go by statistics. Indeed, the rarity of fatal mechanical error will itself raise suspicion of foul play.

Suppose the wife was having an affair with the automechanic who serviced the car a week before? Suppose she recently took out a life insurance policy on her husband?

7. Moore's objection to a cumulative case argument is wildly counterintuitive. He acts as though, the more evidence you have, the less credible the claim. Multiple lines of independent evidence make the claim less probable because each piece of evidence must be individually evaluated. To some extent that's true, but so what? Surely he doesn't really think having more witnesses, more circumstantial evidence, actually weakens rather than strengthens the case for a given event. It's hard to believe he thought that through. 

8. Consider the burden of proof. Atheism demands a universal negative in reference to miracles. Which means in principle that an atheist must disprove every single reported miracle while a theist must only prove one single miracle. 

Keep in mind, too, that if miracles occur, most go unreported because most people aren't famous. Their experience never makes it into the history books. 

9. Every new religion doesn't make miracle claims to launch itself. Islam didn't, Buddhism didn't, Hinduism didn't. 

10. In drawing examples from comparative religion, Moore makes no attempt to date sources. 

11. Moore's case is overly reliant on controlling metaphors like "false positives," "false negatives," "playing with the same money." At best, these are impressionistic analogies. They are not substitute for definitions and arguments.  

12. To my knowledge, Hindu and Buddhist "miracles" don't typically attest beliefs. They are just fantastic stories. 

13. Let's run though some of his examples:

i) The Bible doesn't rule out the possibility of pagan miracles. Indeed, the Bible arguably grants the existence of some pagan miracles. 

ii) Jacob's ladder is a dream. That's not physical teleportation. It's not comparable to Muhammad's alleged midnight journey.

iii) Regarding Balaam's donkey, before you attempt to compare that to other stories of talking animals, you have to consider how that functions in the narrative. 

iv) Were there Brazilian monks in the year 700? Doesn't the presence of Roman Catholic monks and priests and in Latin American depend on the introduction of Catholicism by the Conquistadors and their missionaries? Unless I misheard or misunderstood what Moore said, his grasp of church history and relative chronology leaves much to be desired. 

v) In the nature of the case, transubstantiation, even if true, is indetectable. 

vi) I've discussed Lourdes and Fatima:


  1. steve,

    I know you and other tbloggers have written rather extensively on supernatural phenomenon, but have you written about the possible origins/nature of the Egyptian magicians' replications of the first sign performed by Moses/Aaron as well as the first and second Exodus plagues?

    According to the ESV Study Bible "Egyptian texts are filled with descriptions of priests performing extraordinary feats, including turning inanimate objects into animals."

    Just curious about your thoughts on the subject.

    1. The mere fact that ancient Egyptian books have spells that claim to turn inanimate objects into animals is not, for me, a reason to to believe they could. However, I've read enough about the practice of witchcraft to think's possible. If so, I assume the source is demonic.

    2. I agree that the Egyptian texts in and of themselves bear little weight, and could be easily and reasonably construed as false and propagandistic, but not so with the Exodus account which seems straightforward and direct about the effects of the Egyptian magicians' secret arts recorded therein.

      Like you I also think a demonic source is the best explanation, and these historical echoes mesh nicely with many of the apparently demonically/satanically energized eschatalogical events described in Revelation.