Jim Davila (and) the Hero

Jim Davila mentioned my article about the Rank-Raglan hero archetype and the historical Jesus. Jim notes that, if one includes all the Synoptic Gospels, then Jesus can be ranked higher than I suggested. When I referred to our very earliest sources, I had in mind the letters of Paul and the Gospel of Mark, which lack many of the details that are found on the Rank-Raglan scale.

Of related interest, it might be worth comparing Jesus mythicism with what one could call Homer mythicism. In the case of the latter, there is a huge difference in the evidence, and the time between the alleged era in which Homer is supposed to have lived and the first sources to mention him. But there are still interesting parallels in the arguments, including the highlighting of parallels with earlier and more widespread stories, as though any historical figure, or their teaching or compositions, ought not to reflect their cultural heritage.

POPULAR AT PATHEOS Progressive Christian
What Are Your Thoughts?leave a comment
  • http://mythicpizza.blogspot.co.uk/ Paul Regnier

    If I remember correctly, Raglan’s approach was to consider references in whole bodies of texts about the figures he studied, so it would be more consistent with his approach to consider all the gospels, not just limit it to a couple of texts. Actually, there’s no need to stop there, you should probably add in any heroic bits of non-canonical texts as well.

    • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

      When looking at a figure as depicted in the sum total of literature about them, one may do that. But a historical approach cannot ignore distinctions between earlier and later sources. And so I wonder if your comment isn’t another way of saying that the mythic archetype approach is simply different from, and irrelevant to, the work of historians?

      • arcseconds

        It certainly seems a highly relevant consideration.

        Obviously if considering the earlier material only makes Jesus not seem particularly likely to be mythical, and most of the mythical elements only appear in later sources, some weight (a lot of weight I would have thought) ought to be given to the possibility that the mythic material was added later.

        If Carrier wants to start with a Raglan-style score based on the entire literature as his prior, then somewhere along the way (the very next step would be good!) he ought to take into account that it appears that much of the Rank-Raglan criteria are met by later additions, which should of course have the same effect as if he had started with a probability based on your ‘considering just the earliest account’ score in the first place.

        Does he in fact take this into consideration?

        • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

          He does mention it, on pp.244-245. I don’t think that what he has to say there – in essence, asserting that it does not matter to his case – is adequate, never mind persuasive.

          • arcseconds

            It certainly doesn’t sound adequate. It’s essentially ignoring the consensus account and refusing to give consideration to something that ought to be considered.

      • http://mythicpizza.blogspot.co.uk/ Paul Regnier

        Yeah pretty much.

        If you took a figure whose historicity really is doubtful, like Arthur, I’d say the best you could do work be to look at the earliest available sources and make a judgement from those, based on considerations such as the date and nature of the sources.

        In the case of Arthur his RR score in the earliest references to him would be very low. That doesn’t necessarily mean he existed of course, but the fact that the legend of King Arthur later picks up classic heroic elements, while interesting, plainly doesn’t tell us anything about his historical existence. Any real 5th century British warrior-chieftain is hardly going to pop out of existence because 700 years later some French poet gives him a magic sword or whatever! So I really don’t see how RR is of much help, you still have to go back to those earliest sources and ask the same questions.

        So I don’t think that RR is all that useful for answering the basic question of “did person X exist” though personally I find it fascinating in what it might tell us about the way myth and folklore develop – and maybe how they could attach themselves to historical figures.

  • Anon-a-Mouse

    Didn’t Carrier explicitly limit the time frame to the ancient era? And all the original article shows is that it’s possible that some figures don’t line up to the pattern, not that that’s the norm.

    Possibility is not probability, it would need to be shown that the higher scored frequencies do not line up with mythicism (and vice versa for historicism), not that some exceptions exist.

    • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

      The point is that selecting some individuals to list, and choosing certain similar features to list about them, has not been shown to have any relevance to the probability of historicity one way or the other. Merely asserting that it does, then taking a hodge podge of early and later sources to show that Jesus eventually comes to be narrated about in stories featuring those details, is not going to persuade a historian unless you can first actually show that this approach has some merit to it, and answer criticisms which suggest that it does not.

      • Anon-a-Mouse

        The evidence is the overall correlation of more mythical attributes associated with mythical figures, and although I don’t have the book yet I remember Carrier saying he explicitly shows this with dozens of ancient figures.

        • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

          Carrier does not show that. To do so, he would have to deal explicitly with the clearly historical figures who fit a significant number of the points on the list. It is always possible to draw up a list that focuses on examples that are cherry-picked and at times tweaked in order to fit what you want to prove. But that doesn’t count as persuasive scholarship these days, and hasn’t for a long time. In fact, it is methodologically shoddy.

          • Anon-a-Mouse

            In his comments at least, he says that he did create a large list and shows statistically why it works even if you exclude the historical figures in the book.

            I’m just saying, it’s going to look bad if he has a significant peer-reviewed correlation and all you’re showing is that the probability isn’t 100%.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            I think your comment became jumbled. When you wrote “peer-reviewed correlation,” did you mean if other scholars confirm his conclusions?

            I know Carrier claims that what he is doing is valid. I am unpersuaded. Merely telling me that he says it is valid does not make his case more persuasive. It is just reiterating what we all already know.

          • Anon-a-Mouse

            Well they at least confirmed the evidence is valid right? And if we had valid evidence that there is a large correlation, it logically follows that there is high prior probability.

            Correct me if I’m wrong but it sounds like from what you’re saying that Carrier did not use lists of historical kings or preachers etc?

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            Who cofirmed what? What are you talking about in your first paragraph?

            Carrier did deliberately choose not to situate Jesus among first-century Jewish Messianic pretenders, which would be a much better fit, because the figures of that sort we know of are historical figures.

          • Anon-a-Mouse

            The peer-reviewers confirmed the RR method and thus the results that follow (at least for now before being falsified by another study).

            I guess specifically I’m asking if he included any historical figures in the RR rankings at all.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            I thought as much. You seem to have misunderstood the role of peer review. It is a vetting of whether something deserves publication. It does not comfirm the methods and results of a study in the humanities, any more than reviewers of a scientifc paper try to replicate the experiment that has been written up before deciding whether a paper ought to be published.

          • Anon-a-Mouse

            That aside, does Carrier include any historical figures? If he doesn’t then the method is patently questionable to begin with.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            Carrier says that Alexander and Mithradates are the only historical figures who come close, and gives each a 10 out of 22 and relegates them to a footnote (pp.230-232). So yes, his method is patently questionable to begin with.

          • Anon-a-Mouse

            You mean he only calculated for high ranking figures or, like Carrier commented before, he included low-ranked historical figures but decided to only show the high-ranking correlations?

            This is really important because: if it’s the latter then he basically lied, if it’s the former, then you didn’t read that part or are misleading the issue.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            He offered dubious numbers for historical figures, claiming them as lower ranking than others have suggested they are, and relegating mention of them to a footnote.

            I think that, if you are going to assess the honesty of an author, you should read their book first. Don’t you? Obviously your false dichotomy is also an issue, but not the most fundamental issue, in this case.

          • Anon-a-Mouse

            I ask because a lot of people criticized him on his comments for what you’re describing. His response was that he showed in the book that he did include low-ranked historical figures and the correlation came out the same so he only showed the others. But you’re right in that I do need the full book.

            My advice for a good case is not just to show that it’s possible but that it’s more likely given the evidence. To do otherwise is to feed into Carrier’s narrative about the critics.

            Covington actually gives a decent summary of the debate and what historicists would have to show through literature and historical analysis to prove their case is better:

            http://www.skepticink.com/humesapprentice/2014/08/06/on-the-historicity-part-11/

    • arcseconds

      All the sources that anyone is looking at are from the ancient era. It’s not as if Jesus was attributed a rather humdrum biography up until modern times where it suddenly became incredible!

      The point is that considering, say, all four Gospels in their current form you can justify a highish score. Carrier thinks it’s 20, Jim Davila gives 16.

      But if we look at the earliest material (I’m guessing, but I’m imagining McGrath has in mind Paul and the Gospel of Mark, possibly without the ending that is believed to be a later addition) he doesn’t get a high score, but a rather low one.

      So it looks like most of the mythic elements were added later. How can that justify thinking he’s probably mythical on the basis of this score? If I tell a story about Abraham Lincoln and make him conform to the Rank-Raglan scale, would that start to persuade you he wasn’t a historic figure?

      As I said earlier, in your Bayesian rational reconstruction of consideration of the evidence, you could start off with ‘scores high on the Rank-Raglan scale’ and therefore start with a probability of 0.8 of being mythical or whatever, that’s fine(ish), but you had better include ‘most of the high scores come from later sources’ in your analysis and immediately drop it to close to whatever you think a score of 4 tells you…. which isn’t very much.

      And it sounds like Carrier just fails to take this into consideration.

      • Anon-a-Mouse

        I bring up the point because the article featured people like Kim Jung-Il and Harry Potter who obviously aren’t ancient.

        Because the outcomes suggests that, in the ancient world, there’s a low chance for a historical person to be mythologized to that level regardless. McGrath is right to point to people like Mithradates but that’s an outlier, most at that high rank weren’t historical.

        The question of whether Paul/Mark were positing an earthly being to begin with is obviously a separate issue.

        • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

          How did you determine that Mithradates is an outlier, rather than counter-evidence to the claim that scoring high on this scale makes it unlikely that one is historical? You would need to do what Carrier has not done, namely (1) prove the ahistoricity of some figures he merely presumes to not be mythologized historical figures, and (2) survey at least some category – let’s say royal figures and claimants to thrones or positions of leadership – in a very comprehensive manner, to determine what the percentages are. But even so, if done in this more rigorous manner, the whole approach would nevertheless be open to objections, since myths regularly draw on elements from the mundane and the real in storytelling, and so it would be foolish to presume that real-life elements that are popular in myth are therefore not real-life elements.

          • Anon-a-Mouse

            The two aren’t mutually exclusive but what you’re trying to prove is that the counter-evidence makes up 50% or more of the figures, and that’s only one example. The point is a greater overall probability.

            And yes, from what Carrier said, that is the methodology he used and found the strong correlation for RR. This means that the one example is only a outlier to the general trend. Again, I can’t confirm it for certain.

            Wouldn’t the criteria be whether the mythical people themselves are real? Obviously any fictional character would have elements of how real people act.

          • arcseconds

            Does he detail this process, i..e. explain the methodology in detail, including exactly who he looked at and how he selected them, and how he assigned the scores? Is this selection process robust, comprehensive, and unbiased in terms of what is being tested?

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            There is no reason why Jesus cannot be an outlier, even if one accepts the legitimacy of approaching the subject in this way. The point is that the Rank-Raglan scale can fit a historocal person perfectly and so provides no counter-evidence to a person being historical.

          • Anon-a-Mouse

            But that’s not how the argument works, the point is to see which is MORE probable, and if the evidence points to one side, then that is the better argument, and the one we should use.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            And so Mithradates is someone who, like Jesus, could be judged less likely to be historical, using this approach. But he is historical, and so clearly this kind of assessment of probability poorly fits the way historical study works and the kinds of data it provides.

          • Anon-a-Mouse

            No that’s for the PRIOR probability, but we know he is historical because of overwhelming POSTERIOR evidence. This is fundamental for the entire debate.

          • arcseconds

            I agree, but I would have to wonder about this way of setting the prior probability.

            Carrier’s defenders indicate to us that he doesn’t take this RR stuff super-seriously anyhow, it’s just a somewhat arbitrary (although in his eyes presumably somewhat defensible) way of getting an initial prior, and he (apparently) points out that it doesn’t matter much how he does this, considering further evidence should rectify the situation.

            But why use this method? It’s not a usual historical technique for assessing historicity, and nor does it resemble the arguments normally made by historians. It wasn’t invented to assess historicity. It’s blind to the distinction between later and earlier sources. It’s also a bit of a traditional mythicist approach, to look for parallels with other myths and to conclude on that basis that the Jesus story is just a repeat, and Carrier tends to want to distance himself from the rank-and-file of mythicism.

            Carrier thinks biblical scholars are by and large a lost cause, but he does say he wants to engage historians more widely, so why is he using something that’s only likely to endear him to mythicists at this point?

          • Anon-a-Mouse

            I generally agree but I also don’t see any better methods for evaluating priors since all methods will have to use some kind of description and compare that to a certain class of people. It just seems convenient both in that it has all those elements, can include unfavorable figures and that it has been used academically before.

          • arcseconds

            Has the Rank-Raglan scale been used by historians to argue for the historicity of any figure? The fact it is used in comparative mythology or studies of folklore to look at similarities between stories hardly suggests its applicability here.

            There seem to me to be many other options.

            There’s a paper around that looks at the use of phrases like ‘brother of X’ or ‘father of X’, etc. in the 1st century. The conclusion is that these phrases, when used in the third person, much more usually indicate a literal relationship rather than a metaphorical one. You could use the frequency of literal rather than metaphorical useage to determine a probability for ‘the brother of the Lord’ to indicate a literal relationship of James with Jesus.

            Or you could look systematically at the founding-figures of cults.

            Or, you know, you could ditch the whole Bayesian approach as it’s not clear how to get it started.

          • Anon-a-Mouse

            As a direct historicist argument by a historian specifically there are no others and a lot of what you mentioned are separate thorny issues.

            But I don’t think you need to ditch Bayesian methods, especially when most of the argument is not on RR.

            What was the brother interpretation paper as a reference?

          • arcseconds

            I really don’t understand the first half of the first sentence.

            Are you complaining that there’s a problem with starting with a good piece of evidence for historicity? I can’t see how you can possibly object to that. As Carrier says, it shouldn’t matter where we start.

            The paper is “Literal and Extended Use of Kinship Terms in Documentary Papyri,” Mnemosyne 57 (2004): 131-76. I confess I have yet to read it.

          • Anon-a-Mouse

            No I’m saying the debate over that evidence is a completely different issue, all evidence is added up in the end but how we determine each individual piece of evidence is methodologically different. The “brother” evidence requires textual analysis, mythical traits frequencies etc. requires something like RR.

            And thanks for the study, iirc Carrier actually gave 50/50 to higher for historicists on the brother issue, his main piece of evidence was for what “rulers of the age” meant, which based on language uses made it highly likely for Jesus to be killed by demons.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            I find that this way of applying Bayesian reasoning to historical figures is inherently problematic. It is like talking about a particular individual chef in the 1920s as having been unlikely to or likely to be male because of generalities about that category in that time period. It simply doesn’t matter what the norm was. It matters what the evidence indicates about that individual, and Carrier seems to be trying to use the Bayesian approach to introduce a way of talking that poorly fits history.

          • Anon-a-Mouse

            Yes that’s why it all gets combined at the end anyway, the question now is about priors.

          • arcseconds

            What is wrong with what I suggested just now? We accept they’re probably male on the basis of the frequency, but acknowledge that assessment is not very robust in the face of further evidence. Even just one direct reference to them being female might be enough to make us think they’re probably a woman instead. (Such a reference might even defeat much more direct evidence of maleness, like male names, other references to being male, we might think that the one reference is only really explicable by them being female, and the rest of the evidence can be explained by them adopting a male guise.)

            Historians want to focus on more direct evidence, and of course there’s good reasons for doing so, but it doesn’t seem to me that ruling out all considerations of background probabilities is out of order.

            You argue for this sort of thing yourself when you say that miracles are inherently unlikely. I don’t think you’d be too impressed by ‘it doesn’t matter what the norm was, it matters what the evidence says happens’ in this case?

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            I think that there are different sorts of probabilities. What Carrier is doing with the Rank-Raglan scale and the historical Jesus seems to me to be akin to the erroneous reasoning one gets from someone like Bill Dembski. It isn’t enough to talk about improbability in a general way, because it is misleading. It may be improbable that any one historical individual was a lottery-winner. But using the ratio of non-winners to winners in order to set the prior probability vanishingly small, in an attempt to claim that someone mentioned in a newspaper as having won the lottery probably didn’t do so, seems to me to be cooking the books in a dubious manner. Surely the likelihood of someone winning the lottery is about the odds prior to it happening, and not about the historicity of it having happened after the event?

          • arcseconds

            I think the problem is not that there’s different probabilities (that seems a mathematically dubious claim to me) but rather that probabilities don’t actually contain all the information about your attitude towards the data you have.

            The example I’ve employed before is that you and I can both give a probability of 0.5 to heads on the next toss of a coin by a coin-tossing machine to start with. And then we can both get the same information after that: 20 heads in a row. I conclude the machine is rigged, and the 21st toss is most likely to be a head. You, however, stick with 0.5. You designed and built the machine, and you’ve just watched it toss a million tosses where the ratio of heads to tails is very nearly 0.5, and you conclude that the 20 heads are a 1 in a million event that you’d expect to see once in a million times.

            That’s because our initial assessments at the start of the tossing are based on different evidence. I’m essentially just guessing (on the basis that someone building such a machine would probably be trying to make it fair, maybe), but you’re basing it on quite a bit of knowledge. That means I’m going to find any data whatsoever on the machine’s actual behaviour quite persuasive, whereas for you it’s only a small amount of the overall data, so you’re much less willing to give up the 0.5 probability.

            As far as the lottery case goes, if you know someone has bought a ticket and you don’t know the results of the draw, of course it’s right to conclude that there’s a very small chance of them having won. But on seeing a newspaper article you should up the probability to something much higher, although probably not too close to 1, because newspapers can and have got this sort of thing wrong, so maybe 0.9995 or something.

            So having a very small initial probability doesn’t need to and shouldn’t bias one horrendously in one direction, so long as one is appropriately swayed by better evidence, and most things we’re wont to call ‘evidence’ are going to sway you an awful lot away from an initial guess.

            (However, it depends on the source, of course. If they merely tell you that they won the lottery, I think there is quite good grounds to doubt them, particularly if it was, say, 10 years ago. The proportion of people people lie about this sort of thing maybe isn’t great, but I reckon an awful lot more people tell tall tales than win lotteries, so here the background probability seems quite relevant. If it had been a more common event, like a car crash, it would be easier to believe. )

            We’ve had an example of someone who refused to be swayed from an initial background probability: the guy who argued that because (allegedly) only one in a thousand households had six children, the reference to Jesus having four brothers and at least one sister has a 0.999 chance of being a lie, basically assuming that statements about someone’s family are completely independent of the facts about that family.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            Thanks for finding a better and more mathematically accurate way of making the point that I was trying to!

          • arcseconds

            No charge :-)

            We’re told that Carrier is just after an initial prior probability for Jesus’s existence, and this doesn’t matter too much. This is in principle correct. It shouldn’t really matter what the initial probability is, so long as it isn’t 0 or 1 (in which case Bayes’s theorem will never tell you to change it), it should get ‘washed out’ by subsequent evidence. There are actually formal proofs to that effect.

            Although the ‘washing out’ proofs don’t tell you how much evidence will be required to do that. Someone with a massively skewed prior probability (someone here mentioned 1 in a quadrillion recently) might not actually end up moving terribly much by the available evidence in this particular case, so really extreme values would also be problematic, I suppose.

            But 0.3 – 0.6 or whatever Carrier thinks his Rank-Raglan scale gives him isn’t extreme at all, so he and his supporters are kind of correct when they say ‘it isn’t too important, it’s just an initial prior’. So even if he’s starting with a slightly skewed starting point it shouldn’t matter if he’s treating the more substantive evidence correctly.

            I actually wonder whether he’d not be better off selecting an even more arbitrary starting point.

            0.5 for example, perhaps on the basis that we want to be fair towards both options (some people (experts) would argue that it should be 0.5 as there are two options and that’s all we know, but I side with the experts who think this reasoning is very dubious, but I think a defense could be mounted for this particular case).

            Or maybe even 0.9 for Jesus’s existence, on the basis that it’s the consensus view, or merely to tilt the field in favour of historicity, and say “see, even if someone comes into this thinking he probably does exist before considering any of the substantive evidence will be persuaded by the evidence for mythicism”.

            Or do both: it might be interesting to see at what point being skewed initially against the proposal versus being initially neutral ends up with the priors having being largely ‘washed out’.

            (And actually I think skewing it against what you want to prove makes a lot of sense. After all, if there’s a strong argument for Jesus’s existence, you’d hope it would be convincing also to people who for some reason started off their investigation biased fairly strongly towards mythicism but were otherwise rational and treated the evidence appropriately. )

            And if he had started with a more obviously arbitrary starting point, we wouldn’t need to quibble about his use of the Rank-Raglan scale.

            (Unless he tabled it as evidence, but really, it’s hard to see how this could count as very strong evidence even granting Carrier quite a lot of ground as to the applicability at all)

            Speaking of which, I have a further quibble which does make for some comparison with assumptions about black criminality, but that’ll have to wait until later.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            I would be very interested in seeing someone work through the Bayesian process using a more generally accepted view of the evidence. I wish I had the time – if no one else does it before my next sabbatical, who knows, I just might try…

          • arcseconds

            Well, as I think I’ve said before, if one was to do this properly, one would engage the help of an expert (a bayesian epistemologist, say, there are other options but this is the most directly relevant), familiarize oneself with some of the ‘rational reconstructions’ done using Bayesian epistemology in the philosophy of science, and as an initial project start off with something in history which is now non-controversial, but where historians initially had quite a different view but were persuaded by an accumulation of evidence, and show how it can work in that case.

            (of course, one could say that Jesus’s existence is non-controversial, but if there’s any idea the treatment could one day persuade mythicists or people with mythicist sympathies, then rhetorically a case that they already accept mainstream findings on would be better)

          • arcseconds

            I also wouldn’t be all that surprised if someone had already done something like this…

          • arcseconds

            Regarding my quibble, I’m having a little difficulty in expressing this clearly, which suggests to me I don’t fully understand it.

            But I suppose I’m concerned that the Rank-Raglan scale doesn’t ‘carve reality at the joints’, so to speak. Rank, given his Freudian background, was almost certainly thinking in terms of some deep-seated unconscious thingy that caused stories to be told like this. I’m not too sure about Raglan, but it seems to me that he may be cut from the same cloth as Campbell or maybe even Jung, so again there might be some kind of picture of an archetype or something working its way through human narrative.

            And if that’s true, then the RR story picks out a genuine phenomenon, and if for some reason that tracks mythic figures rather than historical ones, then just as smoke track fire, RR tracks mythicism.

            But it may be there’s no genuine phenomenon here at all, and it’s just a grab-bag of different myths told for different reasons that have superficial similarities with one another. Maybe all of the factors mentioned on the scale vary largely independently of one another, and all we’re seeing is the result of the fact that if enough myths are written, some figures are going to have all of those traits simply through random distribution.

            This receives support perhaps from the fact that Jesus seems to have been attributed a lot of this stuff as a deliberate attempt to (a) make him look messianic, and (b) make him look like Moses. Which is presumably quite a different sort of causal path to take to get here than Mithridates or Osiris.

            So it might be a bit like trying to use some cultural classification like ‘food animal’, which doesn’t respect biological kinds, to guess at some property that biological kinds would be more appropriate to use. For example, we might conclude from the fact that some animal is a food animal that it probably has wings, but obviously we might be very wrong and it’s actually a cow. We’d be better off knowing that it’s a member of Aves, and dealing with its biological taxa.

            (one way we could characterise this relationship is that if we know an animal is an avian, and it doesn’t have wings, then we can still bet it has feathers: an unexpected outcome there doesn’t mean we need to radically revise our other expectations. But if we know a food animal doesn’t have wings, we probably have to also abandon the notion that it has feathers and two legs. )

            In this respect it could well be a bit like knowing someone is black and guessing as to their criminal past or lack thereof.

            A recent paper I had an occasion to peruse (but I haven’t gotten around to reading seriously, alas) did seem to have pretty good evidence that what really tracks criminal behaviour is extreme deprivation (they looked at several measures, including poverty), and it just so happens that disproportionally more black people live in extremely deprived situations. And there’s other results that support this contention.

            So the way this might be working is that it’s extreme deprivation that has the robust causal connection with criminal convictions, and that the association with black skin is accidental, in the sense that being black on the one hand and being extremely deprived and having a criminal background on the other have no real causal connection to one another at all, it’s just that historically more blacks have been in extremely deprived situations and that has continued to the present day. So race might end up (at best) being used as a not very good proxy for class.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            I would love to see someone with relevant expertise in the method run the calculations, adopting a more mainstream approach, and to then write about what was different in their assessment of the evidence and interpretation of the data compared to Carrier, and how it led to their different result.

          • arcseconds

            I think I’d have to agree a little bit with Anon-a-Mouse here, actually.

            No method would give certain proof of Jesus’s existence or non-existence, and you yourself have argued and admitted this plenty of times with respect to particular cases, e.g. acknowledging that ‘brother of the Lord’ does not prove conclusively that Jesus existed, but just makes it highly probable that he did so.

            Assuming the legitimacy of this approach, including that Carrier has done his frequency analysis correctly, then a high Rank-Raglan score should incline one towards mythicism. If it’s the first piece of evidence one considers, then taking the frequency of mythical figures as the initial prior probability makes sense.

            However, what I would say about it is that it’s very weak evidence, in the sense that better, more direct evidence should quickly sway one towards a different conclusion.

            As you’ve got much better evidence for historicism, it’s right to be pretty unimpressed by a high Rank-Raglan score. If this was somehow entirely new information to you (and ignoring the fact that the early accounts don’t support a high score) perhaps it might sway you a tiny amount towards mythicism… but really not very much so.

            And of course you’ve already taken this largely into account in your assessment, because you already knew about the mythological extravagances of the later accounts, and we should beware of Bayesian-styled arguments tricking us into double-counting evidence.

            I feel a crime story analogy coming on. We might liken this to prior criminal convictions. If we start our murder investigation with 10 suspects, all of whom (as we think initially) have at least an opportunity to do the murder (they’re all trapped by snow in the same Agatha Christie bed-and-breakfast or whatever). Let’s say the first thing we discover is that one of them has a criminal record. People who have a criminal record are vastly more likely to murder again than the general population, so this might set our probability that this person is the murderer quite high, say 0.6 or something. But if we find better and more direct evidence that he’s not the guy, like some else’s fingers on the murder weapon and a plausible alibi, then we should immediately drop the probability a great deal, possibly to below 0.1 as the person with the fingerprints has now become the main suspect.

            If we encounter the evidence the other way around, it probably shouldn’t affect the posterior probabilities, so this means we start with 0.1 (prior probability of one of the 10 committing the murder), reduce it immediately below that when we get an alibi, reduce it still further once we find the fingerprints (our main suspect is now definitely someone else), and when we find this fellow has murdered in the past, we might raise it just a tiny bit. He’s now the best alternative suspect purely on the basis of his criminal past, but the person with the fingerprints has direct incriminating evidence.

            My point here is that at the point we take account of the evidence it might not have the same effect. Initially it might have a big effect, but as we become more convinced he’s not the murderer on the basis of other evidence, the effect may fall into negligibly.

            Whereas something like fingerprints is always going to be somewhat impressive.

          • arcseconds

            I guess all that is to say that yes, he certainly could be an outlier, and we certainly have evidence that suggests strongly that he is, but the Rank-Raglan score could still be counter-evidence nevertheless, and the fact it can fit a historical person is no more argument against it than ‘the brother of the Lord’ could be a ritual title of some sort is an argument against it being strong evidence of historicity.

            But even granting the Rank-Raglan evidence a lot of concessions for the sake of the argument, it seems to me at best it would be extremely weak counter-evidence once all the evidence is considered.

        • arcseconds

          I’m not talking about whether Paul or Mark were positing an earthly being.

          Assuming it’s a decent way of approaching the question of historicity at all, which I think is highly debatable, if all the source material up until say 80 A.D. gives a Rank-Raglan score of 4, then on the basis of looking at that material through the Rank-Raglan lens it’s not correct to determine anything much about historicity at at all, and the fact the later material gives a higher score is very plausibly explained by this material being added later (the consensus view is that this is by far and away the most likely story, and I agree with that).

          This has nothing to do with what Paul or Mark were doing or thought they were doing. Mark could even say on page 1. ‘I’m telling a story about a fictional guy I made up with Paul called Jesus!’ — if Mark started like that then we might well all conclude Jesus is a fiction, but we wouldn’t be making that conclusion on the basis of the Rank-Raglan score, but rather that Mark explicitly tells us so.

          We can’t conclude anything about historicity, even provisionally, with a Rank-Raglan score of 4, and that’s all the consideration of the earliest sources gives us.

          You, however, seem to be quite intent on ignoring the distinction between earlier and later sources, a distinction which I would have thought to be of critical importance to history.

          Do you have any good reason for ignoring this distinction?

          • Anon-a-Mouse

            I’m not ignoring it, in fact, I’m saying you need to use different methodologies for this distinction. What the mythicist theory proposes is that Jesus went from fully celestial to an earthly myth, a change which requires looking at two different kinds of evidence.

            For RR you need all literature for the ancient era (which will exclude the earliest evidence and focus on the synoptics). For the earliest evidence you need to determine whether that was celestial or not, separately.

            My point is, you’re conflating different things.

          • arcseconds

            No, I’m not conflating different things. I’m saying that insisting on using the Rank-Raglan score in a synchronic way is ignoring the historically important diachronic dimension. Raglan’s interest wasn’t even in determining historicity of the figures he looked at, so for his purposes ignoring the diachronic development of stories might be fine, but for determining historicity it is very important.

            If you think it’s not important, then once again, I’d like to know why. You oughn’t just accept the method as Raglan proposed it uncritically.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            And I think the probability calculations in what should be a deductive method based on evidence is inherently problematic. Can you imagine if someone proposed a Bayesian approach to criminology, claiming that juries ought to start by calculating a defendant’s prior probability of guilt based on his belonging to the category of black male and the comparative number of convictions of people in that category? But in fact we do much the opposite, seeking out jurors who will not think in such terms. And so either something is wrong with our approach to justice, or (more likely) something is wrong with Carrier’s approach.

          • arcseconds

            I think the reason we don’t do that is not because there’s something wrong with the approach per se, but because we’re not actually very good at assessing evidence in this way.

            The chances of any arbitrary black man, say, having committed a particular crime (let us say, in the same city, to not make it too absurd) are still extremely low, even on the most pessimistic view on black criminality. The chances of them having committed a similar crime in the same year is still very low.

            And, of course, it’s still the case that the vast majority of black men aren’t criminals at all.

            So even if one subscribes to some highly racist beliefs about black criminality, it still doesn’t justify thinking the black guy in the dock is significantly more likely to be guilty than if he had been white. Taking into account the race of the defendant appropriately is almost the same thing as ignoring the race of the defendant.

          • http://youcallthisculture.blogspot.com/ VinnyJH

            Good point.

            One of the first things a detective will do when investigating a crime is to look for suspects with a record of similar offenses. Nevertheless, prior bad acts are not admissible in court as proof of a defendant’s guilt of the crime with which he is charged (although they can sometimes be admitted on some other issue). This isn’t because the detective is being foolish. Rather, the risk that the jury will convict based on the prior acts is deemed too great. Although I don’t know whether any court has ever expressed the rule in Bayesian terms, the essence of it is the risk that the jury will overestimate the prior probability.

            My evidence professor in law school said he was always annoyed when a lawyer on a TV show would rise and shout “Objection! That’s prejudicial.” My professor would say “Of course it’s prejudicial. Why would any lawyer ever introduce any piece of evidence that didn’t prejudice the jury in favor of his client?” The issue is whether the evidence is unfairly prejudicial given its probative value. In doing history, that shouldn’t be a consideration.

          • arcseconds

            Ah, yes, good point about the detective. We do think background probabilities are relevant, and I was trying to think of some obvious way where it comes into play, and that’s a good one.

            I was going to mention prior convictions as evidence my example. My understanding is that they can be tabled if they are similar enough. E.g. shoplifting can’t be tabled for a burglary crime, but a burglary with the same modus operandi could be. Is that not correct? Or does it vary by jurisdiction?

          • http://youcallthisculture.blogspot.com/ VinnyJH

            Under the federal rules of evidence, prior bad acts can be used to “show motive, opportunity, intent, preparation, plan, knowledge, identity, absence of mistake, or lack of accident” and I’m sure that all jurisdictions have something similar. I never practiced criminal law, so I’m not sure how much similarity is necessary in order to get them in under the exceptions or how much the application of the rule varies in different jurisdictions.

          • arcseconds

            Responding to your first sentence, again, you’re normally the first to note that history is an inherently probabilistic enterprise. Certainty or even nigh-certainty are not often achievable, and in fact on many questions ‘beyond reasonable doubt’ is hard to meet, also. So rather than saying ‘definitely X’, all we can normally do is say ‘X is more likely than not’ or frequently the weaker ‘X is more likely than any single alternative’.

            So by ‘deductive’ you surely can’t mean in the sense of the deductive methods of formal logic and mathematics, where true premises always lead to true conclusions, but rather in a sense that encompasses probabilistic reasoning (I’m not denying that portions of the argument can have a deductive character in the aforementioned sense, of course). So you’re going to come up with things like ‘most likely’, ‘very likely’, ‘not very likely’ and so forth.

            And Bayesian confirmation and the probability calculus more generally are the formalization of this form of reasoning, and it’s exactly these kinds of assessments that it’s supposed to be able to represent. I don’t think anyone seriously thinks we’re in a position to fully detail in a totally formal, numerical way the sophisticated reasoning we do, often implicitly, in history and science, but its adherents, of which there are many, would claim that in principle it should in fact be able to do just that.

            It’s still an open question in epistemology to what extent Bayesian-style epistemology really captures our scientific reasoning (say), but it’s certainly the single most popular approach. There are at least reasonably persuasive arguments that it represents a constraint on inductive reasoning, so if you reason in a way that goes against the probability calculus, you’re likely to get yourself into trouble. And if you could demonstrate that something about historical reasoning that’s pretty established as sound historical method which directly contradicts a Bayesian treatment, you should really team up with a philosopher of science (or a philosopher of history, if any of them do this kind of stuff) and write a paper on it, because it could be ground-breaking, and at the very least would cause quite a stir.

            It also seems worth noting that probability calculations, including the use of Bayes’s theorem, are par for the course in areas like epidemiology. Your opening statement almost sounds like you’re asserting that history has a more rigorous methodology than areas that employ sophisticated statistical techniques to establish, say, the causal factors of disease.

            Finally, people have proposed that juries assess evidence in a Bayesian way, especially when some of the evidence is explicitly expressed in probabilistic way, e.g. DNA evidence. One of the arguments for doing this is that it gives juries some kind of a tool to assess the impressive-sounding numerical probabilities alongside more traditional pieces of evidence like fingerprints, albis, etc.

          • Paul E.

            I haven’t read Carrier and I know nothing about Rank-Raglan, so I can’t comment on the broader aspect of this thread, but as a lawyer I am uncomfortable with analogies between what historians do and what the criminal justice system attempts to accomplish, especially when the analogy focuses on one aspect of one part of an interdependent system and fails to account for the nuances of even that aspect.

            Both the criminal and civil systems of justice in common law countries have particularized systems of gathering, presenting and assessing what qualifies as evidence that can change depending on the stage of the process, the issue involved, the burden of production/proof, individual rights, etc., and other considerations. Much of that process can indeed deal with might be called prior probabilities. One example that may be relevant is in the types of allowable presumptions and the effects of other evidence on those presumptions. In reading history, I can see some of the same types of methodologies being used by historians but it isn’t always clearly expressed or understood/argued against from a competing perspective.

            So my thinking is that the use of prior probabilities in certain circumstances can be quite useful and the proper expression of how such probabilities arise and are expressed could be a valuable contribution to historical methodology. I think in that context, the avoidance of blunt analogies to the criminal justice system is probably best.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            I think the problem is not the Bayesian reasoning per se, but the attempt to use Bayesian reasoning while adopting a dubious category for one’s assessment of prior probability. A category like “jilted husband” might be a useful one in Bayesian terms when a wife is murdered. Skin color, on the other hand, is not at all an appropriate category for assessing prior probability of someone having committed a crime. And so the point of the analogy is not to suggest that Bayesian reasoning is inherently problematic, but to challenge the way it is being used. If the category used to determine prior probability is in fact irrelevant to whether a crime has been committed, or to historicity, then it is either a misguided or a deliberately malicious attempt to force a desired outcome from the calculations.

          • Paul E.

            Fair enough, and you may be absolutely right that Carrier is adopting an irrelevant category or doing something otherwise untoward – as I said, I haven’t read him and I don’t know thing one about Rank-Raglan. Nevertheless, I continue to have reservations about your original analogy (which you are changing somewhat in this post) because of its failure to to account for the nuances, some of which I mentioned above, especially in the context of the first sentence in the post to which I originally responded. But debating the propriety of a particular analogy has its limits, and we’re probably there. I think you have made your overall point, and I appreciate the clarification about the use of prior probabilities.

            Incidentally, I think a historian-legal scholar dialogue or symposium about evidence would be interesting! Has anything like that happened to your knowledge? Particular types of evidentiary applications dependent on the stage of litigation could be fruitfully applied in historical argument as a way to increase clarity, I think. And I think legal theory has a lot more to learn from historical methodology than we would care to admit, particularly in the area of common law rule development.

          • http://www.patheos.com/blogs/exploringourmatrix/ James F. McGrath

            That is a great idea! I wonder if we could organize a conference, or at least some kind of joint session with guet speakers at a national conference. The conversations would surely be amazing! I am not aware of anything like this ever having been done. Shall we try to be the first?

          • Paul E.

            I like it!

          • Anon-a-Mouse

            For that critique to be relevant to RR, at all, then linguistic meanings would have to have changed radically from the ancient world to other mythical eras; for instance if the meaning of “virgin” meant something different in Rome than it did in say Medieval England, since RR considered all myths together universally. Clearly that’s not the case.

            Second, it’s even less relevant to the specific point about the change of a few decades between Paul and Synoptics. Obviously no such drastic changes in meaning occurred in that short a time.

          • arcseconds

            What on earth are you talking about?

            I’m claiming that a story can change over time, that’s all, so that a high RR ranking based on later versions doesn’t indicate that earlier versions giving a low RR ranking are really talking of a mythic individual.

            Why is this so hard for you to understand? Why are you so insistent on lumping everything together in some kind of textual stew, without considering the relationship of early versions with later ones?

            And what has linguistic change got to do with any of this?

          • Anon-a-Mouse

            My apologies, I should have just asked what you meant, I thought you used diachronic and synchronic to mean linguistic meaning.

            To your real point, to the extent that it follows a hero/royal motif it should be based on RR, if not, then it shouldn’t.

            The point in which we disagree is whether it’s really the same story, I think mythicists would argue that the original is different from what followed later and thus is not applicable to RR whereas the later one is.

          • arcseconds

            So, it seems to me that you’re telling me the following:

            *) The earliest extant material — the most likely to reflect the original version of the story whether it was a myth or an account of real events — is not something amenable to a Rank-Raglan analysis

            *) The later version is.

            *) Mythicists prefer to focus on the later version of the story, because that’s the one where Rank-Raglan can be applied.

            I agree with all of that, but what I fail to see is how you can say all of that and have any sympathy whatsoever with the mythicists here. They’re choosing to focus on a version of the story which is least likely to reflect historical reality and most likely to look mythical, and using that to argue that Jesus is mythical. How is this anything other than cherry-picking evidence in quite a flagrant and non-historical way?

            Once again, mythicists appear to be following a similar approach to Christian apologists, who also want to treat the material in a kind of a timeless manner and ignore the historical development of the material.

          • Anon-a-Mouse

            I think there’s been a misinterpretation of the the theory then, the point is that historically post-Paul Christianity became a new sect (but which still tries to reconcile with Paul).

            In this sense it doesn’t matter what the RR of Paul’s Jesus was since what followed was totally different and, ostensibly, our interest is in the post-Paul sect. Although I agree it doesn’t help the case that Paul’s Jesus wasn’t ranked high.

            That’s why I’m surprised no one has really criticized the historical argument of the evolution of the church when it is pretty heterodox, and it’s also why Paul’s letters ends up becoming the deciding factor for everything.

          • arcseconds

            What an utterly bizarre statement. How can being interested in a later version of Christianity determine whether Jesus existed or not? It almost sounds as if you would start to think Abraham Lincoln didn’t exist so long as enough people told extravagant tales about him for long enough, no matter how good the pre-extravagance evidence was.

          • Anon-a-Mouse

            It doesn’t, the content of Paul does, that’s the proof for historicism. The point of the Rank-Raglan debate is just about the viability of the method to get a good priors in-itself, and whether certain scholars understand what the probabilistic debate is to begin with.

            I think Carrier’s intention was to measure how likely a historicist vs mythicist figure BECAME an RR hero myth which results in the same outcome but is dubious since we know very little about pre-RR stories.