McGrath on the Rank-Raglan Mythotype

Cover of Richard Carrier's book On the Historicity of Jesus. Medieval icon image of Jesus holding a codex, on a plain brown background, title above in white text, author below in white text.Yesterday I addressed McGrath’s confused critique of portions of On the Historicity of Jesus (in McGrath on OHJ: A Failure of Logic and Accuracy). He has also published a second entry in what promises to be a series about OHJ, this one titled “Rankled by Wrangling over Rank-Raglan Rankings: Jesus and the Mythic Hero Archetype” (let me know if more arrive in future). This entry is even less useful than the first. Here are my thoughts on that.

Once again Neil Godfrey already tackles the failures of logic and accuracy in the very first comment that posted after the above article. Which he has reproduced, with an introduction, in better formatting on his own blog: Once More: Professor Stumbles Over the Point of Rank-Raglan Mythotypes and Jesus.

I could leave it at that, really.

TL;DR: McGrath doesn’t understand the difference between a prior probability and a posterior probability; he uses definitions inconsistently to get fake results that he wants (instead of being rigorously consistent in order to see what actually results); and he shows no sign of having read my chapter on this (ch. 6 of OHJ) and never once rebuts anything in it, even though it extensively rebuts his whole article (because I was psychic…or rather, I had already heard all of these arguments before, so I wrote a whole damned chapter to address them…which McGrath then duly and completely ignores, and offers zero response to).

That’s pretty much it.

But now for the long of it…

What McGrath Doesn’t Get

Godfrey has also written on the Rank-Raglan scoring method before. Although I am very particular in the limited way I employ it, and I am carefully rigorous in that application, in a way others in the past have not been (advocates and critics alike). All of this is explained in chapter 6 of OHJ. I discuss the mythotype itself in chapter 5 (Element 48), but not the validity of its application. That is addressed in the following chapter. (It should also be noted that I adduce other applicable mythotypes in Elements 46 and 47.) But the kind of critique McGrath attempts simply ignores everything in chapter six. He clearly doesn’t understand the mathematical or methodological point. It does not even appear he has read that chapter. Had he, he might delete his “rebuttal” as being completely off point, and quite incapable of challenging how I use the data.

These kinds of failures are common. I corrected several before already when Chris Hallquist failed to correctly understand the method and its application. And in fact McGrath and anyone who thinks he has a point really need to read my treatment of the subject there. They won’t know what they are talking about until they do.

But my most succinct summary of the problem I voiced in my rebuttal to the hack critic F. Ramos:

[If] an above-half score on the RR criteria were not a reliable indicator of myth, then we should find many historical persons meeting that condition. That none do is therefore an objective fact we must take into account. The criteria being broad or arbitrary makes no mathematical difference (OHJ, pp. 231-32, 244-45). … Mentioning that lots of people score some of the items has no bearing on that fact. By Ramos’s reasoning, it should still be just as common for historical persons to score as high as Jesus. That it is not (despite 15 people fitting that condition), is what is relevant. And yet, while Ramos expresses his annoyance that I actually point out that Mithradates and Alexander the Great just fall short of scoring high enough to count, I actually allow that between 1 and 4 persons scoring high enough do indeed exist. So he doesn’t even understand that I am actually including the assumption of historical persons meeting the high-score condition. He has no case to make for a higher number.

And that’s the core of the debate: McGrath also has no case to make for a higher number. And quite frankly, that’s the end of the discussion. We’re done here.

You have to use my numbers. Because those are the numbers borne out by the facts. If you want to use different numbers, you’d better go find some different facts. Until you do, there is nothing left to discuss. The numbers are what they are. Cope.

But for those who want something more in depth, I’ll continue.

Primer on the Method

In my treatment of Nicholas Covington’s series on my book, I had occasion to discuss why the Rank-Raglan test changes things, in a way other markers of myth don’t (this appears under the section heading “Using Up the Gospel “Mythyness” to Generate the Prior Probability”):

I assume everything is 50/50 [whether a member of some class is historical or not] for which I have no data demonstrating it to be otherwise. I have data for the Rank-Raglan class demonstrating it to be otherwise. But, Covington could ask, isn’t that true for a few other things as well, and not just the fact that Jesus is a Rank-Raglan hero? Maybe. But…

I noted in Element 48 (in OHJ, p. 230; and also, less obviously, in my development of Elements 31, 46 and 47) that there are many other attributes the Gospel Jesus shares in common with mythical persons more frequently than historical persons besides the Rank-Raglan criteria (a fact often overlooked by people who despise the Rank-Raglan criteria and think my prior probability hinges all on that…it doesn’t). When we include them all (all those criteria, e.g. the classes Jesus falls into in Elements 31, 46, 47, and 48), Jesus falls even more certainly in a reference class over-loaded with mythical persons and under-occupied by historical persons. We just don’t have the sample size for those classes that we have for the Rank-Raglan class, so assessing priors from them, even though that certainly (even a fortiori) would always favor ahistoricity for any randomly selected member (and therefore Jesus), would still be more subjective, and thus harder to explain why it has this effect on the prior (not impossible, just harder).

More importantly, being more certainly in a myth-heavy class is not the same thing as being in a class that is more myth-heavy. For example, the ratio of mythical-to-historical persons in the super-class of all persons conjoining Elements 31, 46, 47, and 48 may be entirely the same as the ratio in the Rank-Raglan class. But what we gain by noticing that Jesus actually belongs to that super-class is that his being in it is even less likely to be an accident than his being in the Rank-Raglan class. Thus, those other Elements (as well as the additional criteria I mention on p. 230) create high confidence that the result derived from the Rank-Raglan class alone is applicable and correct. Covington might want to argue that “surely” members of that superclass are less commonly historical than even members of the Rank-Raglan class alone are. But there isn’t really any data to establish that…and what there is, looks pretty much just like the Rank-Raglan data.

In summary, adding more mythic markers to Jesus than are included in the Rank-Raglan class, even though Jesus does indeed possess those markers, does not make Jesus any more likely to be mythical. At least not significantly (as in, enough to show up relevantly in our math). Once you are that mythical, adding even more mythic markers makes little further statistical difference. At least, I cannot see how to prove otherwise in any convincing fashion.

So we could pull a lot more data from the Gospels into generating our prior probability estimate…and it would not significantly change the prior. Which means, leaving that data in the evidence-pool instead will not generate any significant difference in the consequent probabilities, either. (Because mathematically, those two statements mean the same thing: every prior is the output of previously calculated consequents; so no change in a prior, means no difference in the consequents.)

This is what McGrath doesn’t get. At all.

Jesus most definitely belongs to a set of persons in which more members of that set are non-historical than are historical. That is a fact. It is not theory. It is not phantom. It is not opinion. It is not interpretation. It is a bare, objective, observable fact. It does not matter why or how or even when Jesus got that way. Because the only question that remains is: if you put all the people in that set in a hat, jumbled them up, and pulled one out at random, how likely will it be that the one you pull out is historical? That’s your prior probability. And the answer equals the ratio of persons in that set who are historical, to those who are not. And this is true even when historical persons are in there. And it remains true even if one of those historical persons is Jesus.

McGrath perhaps can’t understand this because he is confused by what a prior probability is. It is not the probability that Jesus was historical. But it still must be included in calculating that probability. But if that’s McGrath’s problem, he needs to re-read chapter six until he understands how the argument works. Only then can he offer any meaningful critique of it.

Literary Analysis 101

But first I’ll address McGrath’s failure to even get literary analysis right–which we must get right before we even get to how we use the data to derive a frequency of correlation with historicity as an attribute.

McGrath tries to “down-score” Jesus on the 22-point scale by sucking at basic literary analysis. For example, he says, of the criterion of a mysterious death, “crucifixion was hardly a mysterious way of dying in the first century, and so it depends what one means by ‘mysterious’.” Really.

Here is how Jesus dies even in the earliest of Gospels:

And someone ran, and filling a sponge full of vinegar [an equivalent to smelling salts], put it on a reed, and gave him to drink, saying, “So be it. Let us see whether Elijah comes to take him down.And [instead] Jesus screamed loudly, and expelled the spirit. And the veil of the temple was rent in two from the top to the bottom. And when the centurion, who stood by over against him, saw that he gave up the spirit like this, he said, “Truly this man was the Son of God!” … And [when he was informed of this] Pilate marveled that he was already dead. (Mark 15:36-39, 44)

That’s a mysterious death. It occurs mysteriously suddenly and without cause (several markers in the story are placed to emphasize this), it has visible supernatural effects, and it astonishes those observing so greatly that even an officer of Rome declares Jesus a deity. It simply makes no sense to define “mysterious” so hyper-specifically that this wouldn’t count as exactly that.

Set Theory 101

I already address in OHJ the issue of being so specific in your meaning as to try and avoid obvious conclusions of literary theory. Hence I corrected vaguenesses in the Rank-Raglan wording that didn’t match the way they applied the criteria (and then applied the criteria consistently), so that the criteria would be meaningfully specific but not absurdly specific (p. 228, n. 187; pp. 230-31, n. 191; pp. 231-32, n. 193). This is actually a Christian apologetic tactic, of deploying a ridiculous “hyper-specificity” as a means to deny something belongs to a category that in fact it does. They similarly end up saying ridiculous things about resurrection doctrines (OHJ, p. 62) and mystery cults (OHJ, p. 97, n. 72), by defining terms so hyper-specifically as to exclude numerous unmistakable resurrection doctrines and mystery cults. The same tactic was even attempted on messianism, so as to deny Jews had any concept of a messiah before Christianity! (OHJ, p. 61, and pp. 71-72, n. 27.)

This is rhetoric, not a respect for facts.

In actual fact, it does not matter how broadly you define a criterion. Because if you define it too broadly, so many people will fall under it as to render the criterion useless, whereas if you define it too narrowly, you will fail to see existing patterns entirely, while if you define it somewhere in the middle, since all members are tested against it equally, if members are just as likely to exist as not exist, then this is what the data will show. The only way to regard a criterion as useless is if no matter how narrowly or broadly you define it, all resulting members are just as often historical or not. But if there is some definition on which there are still too many members to be a chance accident and yet a clear disparity in how many are historical, that is a fact of reality you must account for in your math. You can’t dodge that fact with semantic tricks. (And I actually mathematically prove this point on pp. 238-44.)

McGrath has no response. He doesn’t even mention any of this.

If you are not going to interact with the book, you cannot be regarded as critiquing it. You are just engaging in more warblegarble.

Bayes’ Theorem 101

That McGrath has no idea what he is talking about is illustrated by one paragraph in which he says, “In addition to knowing that some historical figures rank highly on the scale,” a fact I not only fully acknowledged but even accounted for (in turn a fact McGrath never acknowledges or even seems to understand), “it must also be pointed out that in many instances we do not know whether a myth is based, however distantly, on some historical figure. And so using this scale as though it could demonstrate non-historicity is actually assuming what needs to be proved.” Indeed. If someone did that, they would be committing the fundamental error of using the prior probability as the posterior probability. Which I very explicitly don’t do.

What he just said is in fact why I don’t use this prior probability (and hence the Rank-Raglan data) “to demonstrate non-historicity.” I use it to establish a prior probability. Which is not sufficient to demonstrate non-historicity. As I explicitly explain on pp. 238-39 and 252-53. So McGrath is actually just repeating what I already explained in the book, as if I didn’t know this, and didn’t take it into account. And somehow he thinks this is a valid criticism. That bespeaks a befuddling failure to pay attention to the very book he claims to be reading.

The final proof that McGrath isn’t really reading my book is that he waxes on about the example of Mithradates from the work of Adrienne Meyer. Evidently wholly unaware that my refutation of her is in the book: OHJ, pp. 231-32, n. 193. He says nothing in her defense. Evidently, because he didn’t even know I refuted her. Because he hasn’t actually read the book he claims to be reviewing. (For Meyer, the facts simply don’t support her scores, and her definitions are too absurdly broad.)

Christian Apologetics 101

That McGrath is engaging in Christian apologetics and not honest scholarship is proved by his using a standard apologetic tactic: the equivocation fallacy. He uses absurdly hyper-specific criteria to down-score Jesus, then uses absurdly hyper-broad criteria to up-score later historical persons. You can’t play that game. You have to apply the criteria consistently.

It’s just worse that his examples are not even applicable comparands because they are actually borrowing the Christian narrative, in a dominant Christian world culture, a thoroughly different cultural context. As I noted in my treatment of Hallquist making the same mistake:

I should also point out that, except when dealing with cross-cultural universals like human biology or economics or social dynamics, we can no longer use modern examples as relevant to the reference class for Jesus (because without a sound reason, we can’t use modern frequencies of anything distinctly cultural as the ancient frequency of it: I explain this mistake in PH, pp. 18, 245). But even if we played that game, Hallquist would lose. Because Superman also fits the Rank-Raglan profile, as do Anakin Skywalker, Optimus Prime, Aragorn, and Captain Kirk: all score above 10. But it would be as invalid of me to use those heroes to argue Jesus is less likely historical, as it would be of Hallquist to use Kim Jong-Il to argue the reverse. The context of all these examples are no longer sufficiently applicable. Although note what would happen if we did what Hallquist wants, and counted all modern examples scoring above 10. Do the math. If you know what a ‘probability’ is, you’ll laugh.

You also can’t just re-define the criteria and count your resulting scores as definitive. If the scores come out high for a large number of members on any definitions, you must examine the resulting reference class for any imbalance in membership, because if there is one, you must use that reference class. This is not only a consequence of the Rule of Greater Knowledge (cf. Proving History, index). It is a necessary fact of logic. That imbalance will be corrected, if corrected it should be, by including other data (that’s why this is just a prior probability, not the posterior probability). But you can’t just ignore data.

Learn Some Basic Math Concepts. Please.

McGrath closes with a quote of Dundes (emphasis mine):

The fact that a hero’s biography conforms to the Indo-European hero pattern does not necessarily mean that the hero never existed. It suggests rather that the folk repeatedly insist upon making their versions of the lives of heroes follow the lines of a specific series of incidents. Accordingly, if the life of Jesus conforms in any way with the standard hero pattern, this proves nothing one way or the other with respect to the historicity of Jesus.

McGrath evidently thinks x “does not necessarily mean” y means x cannot make y more probable. Sorry. No. That’s not how math works. Dundes said nothing about a high score not being able to up the probability of non-historicity. He only said it does not entail non-historicity. Which is exactly what I say myself. Repeatedly. In the book. Which McGrath is starting to show little sign of having read.

If x increases the probability of y by 10% or 50% or even 80%, it would still be true that x “does not necessarily mean” y. So quoting someone saying x “does not necessarily mean” y tells you nothing about the one thing you actually want to know: whether x increases the probability of y. Even when Dundes says Jesus scoring high (and Dundes fully agrees he does, by the way, contrary to McGrath) “proves nothing one way or the other with respect to” historicity he is simply repeating the same point: it does not prove Jesus did not exist. He can’t mean anything more than that, because he uses the logical connector “accordingly,” so he is saying the second statement follows from the first, and is not a separate additional point being made.

But even if Dundes made the same error McGrath just did, and mistook x does not entail y as meaning x cannot increase the probability of y, in actual fact, and indeed perhaps unknown to Dundes (who did not study Bayesian reasoning or how reference classes work, or set theory or probability theory), when we look at the available data, the Rank-Raglan profile does prove “something” with respect to historicity: it proves high scoring members are more often not historical.

And that means just what I explain in the book:

Just as ‘the prior probability that Jesus was raised from the dead by a supernatural agency is the same as the prior probability that a supernatural agency raised Romulus from the dead, or Asclepius [etc.]’, so, too, the prior probability that Jesus is historical is the same as the prior probability that Romulus is historical, or Moses or Hercules or anyone else in the same class. Again, as with the resurrection claim, the evidence in the case of Jesus can be much stronger than for any of the others; but that is accounted for with the consequent probabilities. Here we’re only talking about the prior probability. We’ll get to the evidence later. The prior probability is not the posterior probability, and only the latter is the probability that Jesus existed. But we have to start somewhere, and this is the best starting point, because here the frequency data are sufficiently clear and there is no narrower reference class we can say the same for. (p. 239)

It’s clear McGrath did not read this. Or did not understand it. Because he has no response to it.

McGrath does not attempt to show the ratio of historical to non-historical persons in the high Rank-Ragan set is different than the ratios I propose. Nor does he then apply his ratio to see how that changes the probability that Jesus existed.

So McGrath actually doesn’t know whether the ratio is any different than I say, or if it is, whether that even makes the historicity of Jesus likely, or how likely it makes it. Which means McGrath has nothing to contribute to this debate.

[McGrath also has a third article that I address here.]


  1. John MacDonald March 6, 2015, 7:37 am

    I made the comment on McGrath’s blog that:

    One good point Carrier seems to makes is in pointing to Phil 2.5-11 and explaining that it paints a very different picture than the usual miracle performing Jesus we are used to:

    “Have this mind [of humble love] in you, which was also in Christ Jesus, who, existing in the form of God, did not decide to seize equality with God, but emptied himself, taking the form of a slave, being made in the likeness of men, and being discovered as a man in outward form, he humbled himself, becoming obedient to the point of death, a death of a cross. (Phil 2.5-11)”

    Carrier comments on this passage that:

    “Key things to notice here are that again no mention is made of Jesus having a ministry, teaching anything or performing any miracles. To the contrary, having ‘emptied’ himself of all he was and ‘humbling’ himself completely to the status of a ‘slave’ imply he would have had no supernatural powers at all (Richard Carrier, On The Historicity of Jesus, pg. 535).”

    MCGRATH RESPONDED THAT: I am continually struck by the fact that the one point at which mythicists don’t question the scholarly consensus among conservatives is the the point at which their view is most open to criticism as reflecting undue influence of dogmatic doctrinal positions. The notion that we are dealing here with divine self-emptying is the stuff of what is known as “kenotic Christology” in the realm of systematic theology. But it is unfortunate – if unsurprising – that Carrier uncritically accepts the reading of that back into Philippians. Even Robert Gundry, a conservative Evangelical himself, sees in this language an echo of Isaiah 53:12 rather than a reference to incarnation (The Old Is Better, p.239 n.35 and p.284).


  2. David Ashton March 6, 2015, 3:26 pm

    Very interested in the Indo-European & Semitic theme of the Cosmic Battle between the Divine Hero & the Evil Reptile.

    For the Christian believer, similarities with hero and/or risen deity myths confirm the “universality” (in the west) of their Son of God beliefs. See e.g. Victor White OP on this.

  3. David Bridges March 17, 2015, 6:01 am

    Greetings from Ireland Richard on this St. Patricks day. Another brilliant piece of scholarship and writing demolishing the awful Alistair McGrath. I love your style and erudition, even if I am a bit out of my depth understanding it all and I am going to buy a copy of your book On the Historicity of Jesus to read on holiday. I live in Crossgar, a few miles from Downpatrick where St. Patrick is supposed to be buried (HA). Did you know that Alistair McGrath went to school in Downpatrick?
    All the best……its getting near time to drown the shamrock!

  4. I think your blog would be more readable if you kept more to the arguments and used less ad hominems. I have found myself more and more put off by the general tone on your blog. it’s like you’re writing for a set of rabid fans rather than focusing on substance. On this matter I generally agree with you in the historicity debate, but I happen to think McGrath has some good (and some bad!) objections to the R-R argument. So naturally I would like to see your response, but it seems to me that you’ve missed actually responding to his better arguments, being so busy trying to come up with vitriolic statements. Now I don’t mind some ‘frank’ language once in a while.. but generally speaking, politeness is a good thing.

  5. John MacDonald June 9, 2017, 12:26 pm

    Otto Rank, in 1909, developed a Hero pattern that was very much based on Oedipus’s legend. Lord Raglan, in 1936, developed a 22-point myth-ritualist Hero archetype. Oedipus scores the highest on this scale, at 21 out of 22 possible points. Rather than claiming, as Carrier does, that Jesus scoring well on this scale is evidence in favor of Jesus being mythical, a far more simple explanation is that some legendary material about Jesus was added to his biography to pattern his story after the model of Oedipus (just like Dr. Dennis R. MacDonald argues in his new book “The Dionysian Gospel: The Fourth Gospel and Euripides,” that the fourth Gospel is modelled after the pattern of Dionysus in Euripides). Folklorist Alan Dundes has noted that Raglan did not categorically deny the historicity of the Heroes he looked at, rather it was their common biographies he considered as nonhistorical. Furthermore, Dundes noted that Raglan himself had admitted that his choice of 22 incidents, as opposed to any other number of incidents, was arbitrarily chosen. This could all just as well mean legendary embellishment of the historical Jesus, as it could mythical genesis. So, the Rank-Raglan mythotype seems irrelevant to the mythicist/historicist debate about Jesus

    1. If Jesus’ portrayal is in fact imitating Oedipus, Jesus scoring high on the Rank Raglan scale is no more evidence of mythicism than is Matthew portraying Jesus as the New Moses.

      1. It is evidence supporting mythicism.

        Because indeed: Moses is also a Rank Raglan hero. And also didn’t exist.

        Hence, the more Jesus looks like non-existent persons, the higher the prior probability he was a non-existent person. Just like anyone else.

        What that prior probability is, will equal how often persons who look just like Moses and Jesus and Osiris and so on, didn’t exist.

        If you want to claim that frequency is higher than 1 in 3, you need evidence that it is higher than 1 in 3.

    2. All addressed in OHJ.

      But in short, it can’t be random, because random attributes would generate a high frequency of historical members, when the set exceeds six. We have fifteen. The probability of so many factors corresponding to such a low probability of existing by arbitrary accident is astronomically low. Therefore, the set is indicative.

      It is therefore evidence. It doesn’t matter how arbitrary the discoverers of the pattern think it is. The pattern is so weirdly predictive, it cannot be a chance accident. That’s why it’s useful. Had it not been that way, we wouldn’t be using it. Or if we did, there would be so many clearly historical members, it would give us a high prior of Jesus existing (just as the case with Spartacus, for example, who belongs to no low low-frequency historicity classes, but definitely belongs to at least one high frequency historicity class).

      Random attribute assignment can either only generate few set members, or generate a large set most of whose members are historical. To get a set of fifteen none of whose members are likely historical, is effectively impossible. Unless having those attributes does indeed correspond to a low frequency of existing. Hence, that’s what it entails.


Add a Comment (For Patrons & Select Persons Only)