A Teeny, Tiny Possibility You’re Wrong

A Teeny, Tiny Possibility You’re Wrong May 12, 2019

Everything can be connected to a Friends episode, in my experience, and most things can be usefully illustrated with some scene from Friends. A case in point is the exchange about evolution that takes place between Ross and Phoebe. While I enjoyed this episode, the more I’ve thought about it, the more disappointed I’ve become with Ross’s final reaction when Phoebe accuses him of having caved on his beliefs. The confidence that we have in scientific, historical, medical, and other conclusions of those with expertise who study evidence should not be dogmatism with no room for uncertainty. As I emphasized in recent interaction with mythicists, the mere existence of alternatives that are not entirely impossible is not at issue and not a problem for mainstream scholarship. Lots of things are in theory possible. The question is which is most compatible with the evidence. Sometimes one stands out as clearly the best fit to the evidence. Sometimes that isn’t the case. But either way, there is always a “teensy, tiny possibility that you’re wrong.”

And so now the question I’m wrestling with is how to get people to understand this element of nuance and probability in academic study and its conclusions. It is neither the case that there is no room for uncertainty, nor is it the case in most instances that all possibilities are equal. Sometimes several could fit the evidence equally well – and getting people to understand that as a source of scholarly agreement and lack of consensus is also important.

Everything can be connected with a Friends episode. At least, I think that’s true. There’s a teensy weensy possibility that I’m wrong about this. I’m confident, yet at the same time open to changing my mind.

But in this case, there is a clear connection with Friends, and I kind of wish that Friends had done a better job of depicting academia. It wouldn’t have been as entertaining as what was actually in the episode. But it would have been more beneficial to the public understanding of these important matters.

Browse Our Archives

Follow Us!

TRENDING AT PATHEOS Progressive Christian
What Are Your Thoughts?leave a comment
  • John MacDonald

    One thing I found as an elementary school teacher is that people (children) assign vastly different probabilities to the same event, such as the children predicting the likelihood that they will watch television tonight after dinner. Maybe one of the differences that fuels the mythicism/historicism debate is the different background/frame of reference for PhD New Testament scholars when compared with mythicists when assigning probabilities to the same evidence?

    • John MacDonald

      One activity I did to lead up a probability unit in math was, half an hour before recess, have the children write on a post-it note how likely they thought it was that they would be going outside for recess, considering whether it might rain, whether they had been naughty, etc. Even though in the past the students would generally be “sure” they were going out for recess, in the context of this activity there would be a wide range from “possible” to “certain.” Having the students post their post-its on the board really gave the students pause because they saw how profoundly other children could differ from them when assigning probabilities regarding the same event.

      • Gary

        “how likely they thought it was that they would be going outside for recess, considering whether it might rain, whether they had been naughty…”
        I think children respond to the most recent “in the moment” thought. If you said, “whether they had been naughty”, then asked them to write the probability, your results probably (no pun intended) reflect their current thoughts on their having been being naughty recently.
        Conclusion: pre-existing bias in the participants of your study. Much like PhD New Testament scholars and mythicists!

        • John MacDonald

          I looked on line and it said

          The Objective Probability of an event is based on the likelihood of that event occurring. In most forms of probability, quantitative information is gathered and interpreted to help determine this likelihood through a mathematical mechanism, normally relating to the mathematical field of statistics. The percentage chance of a flipped coin landing on heads or tails can be interpreted as a probability, expressed as a 50% chance that it will land heads up, and a 50% chance it will land tails up.

          Subjective Probability can be affected by a variety of personal beliefs held by an individual. These could relate back to upbringing as well as other events the person has witnessed throughout his life. Even if the individual’s belief can be rationally explained, it does not make the prediction an actual fact. It is often based on how each individual interprets the information presented to him.

          If we plot “analysis of Jesus evidence leading to inferring historicity/non historicity” on a continuum between ‘highly objective probability’ to ‘highly subjective probability’ based on an evaluation of the evidence for the existence of Jesus, I wonder where the analysis would be plotted. From what I’ve seen of the debate, the problem is mainly about the fact that what one side sees as evidence suggesting historicity, the other side doesn’t interpret the same evidence as warranting the same inference.

          • Gary

            I think the problem lies in, “The Objective Probability of an event …quantitative information is gathered”.
            Whether we like it or not, no quantitative data exists. I guess Carrier’s Baysian stuff is an attempt to generate quantitative data, but it is not. It is generated by him, and as they say, garbage in, garbage out. Maybe it’s not garbage in, but no one knows for sure, since there are no actual quantitative events to count, or quantitative data to analyze. It’s his best guess, that’s no more valid than anyone else’s guess. So lacking anything else, might as well go for the “overwhelming consensus of scholars”, like Ehrman, than anything else.

            “If we plot “analysis of Jesus evidence leading to inferring historicity/non historicity”…
            Plotting something implies putting error bars on the data. When the error bars are wider than the plot itself, the plot is pretty useless.

          • John MacDonald

            One of Carrier’s critics, Christoph Heilig, says

            Carrier’s general approach is fine but his assessment of the evidence is just horrible at so many points that it doesn’t help his conclusion that he updates his prior constantly

            I have no problem applying a mathematical model to historical inquiry, since history is about establishing probabilities, and probability is a mathematical concept.

            But the point with Carrier is his questionable interpretation of the evidence that is resulting in the numbers that he is plugging into the equation.

            I certainly don’t think probabilistic historical inquiry needs math, and using math can sometimes produce the illusion that one has proven the resurrection, like William Lane Craig tries to do, or establishing that it is more likely than not that Jesus didn’t exist, as per Carrier’s attempt.

          • And as long as one understands that both Carrier and Craig are plugging their own fallible and clearly biases evaluations of probability into the Bayesian formula, then one should be able to understand that what is being calculated is not the probability of Jesus having been a historical figure of having risen from the dead, but only Carrier’s and Craig’s evaluation of the evidence, turned into a specific number.

          • John MacDonald


          • It’s even worse. The way Carrier derives his numbers are even explicitly _forbidden_ by Bayes’s theorem. It has been a long time that I’ve watched WLC, but as I remember it he was at least clear what kind of prior he presupposed and where he was getting his estimations from. One can debate whether there is much value in mapping out why one believes a certain thing (and why one thinks that people who share certain assumptions also should do so), but it’s not formally wrong. Carrier is just wrong. Apparently, he hasn’t understood how Bayesian confirmation works. So this is not an example of the limited use of Bayes’s theorem for historical research, it’s just an example of an outstanding misuse of the theorem. If I say a calculator helps me doing calculations but I then type in a number and claim it’s the result of a calculation process, that’s not the fault of the calculator! Ok, but I said elsewhere I wouldn’t comment on this again… so I won’t. Really. Now I am serious. 😀

          • Thanks for sharing your thoughts on this – and since you did so publicly, there is now a reasonable chance that this will be shared, and that mythicists will impose on your time in ways that hinder you doing more useful scholarly work! 🙂

          • Haha, James, I hope not. I’ve tried one more time to explain in a comment here https://vridar.org/2019/05/12/the-questions-we-permit-ourselves-to-ask/#comment-93064 how a likelihood-ratio would have to be determined properly for Gal 1:19 as evidence – and that Carrier, not showing any indication for doing so, obviously hasn’t understood the basics of Bayesian confirmation theory. I don’t think I can do it better than there. It’s really quite simply: in order to determine how “expected” something is on the assumption of a hypothesis, you need to know the other possibilities and how frequently they occur in similar situations. The fact that something _can_ occur in such a context doesn’t tell you anything about the likelihood-factor (except that it’s not zero, which it is never anyway in such matters). So if you talk about Gal 1:19 but don’t count other ways of referring to physical siblings on the one hand and fellow Christians on the other hand, you simply haven’t understood how likelihoods work. Period. Such arguments do not deserve to be taken seriously. (And, just to say that very clearly, this has nothing to do with the applicability of Bayes in general on such questions – which I’d welcome.)

          • arcseconds

            I have read your comment on Vridar, but it is rather long and involved, and to understand it properly it seems I need to both understand Carrier’s argument at this point, and follow Godfrey’s analysis too, and… well, I’m never convinced the game is worth the candle…

            But it’s not clear to me that what you are saying really means Carrier has failed to understand Bayesian confirmation, as opposed to something else. He could just being extremely intellectually lazy or having bizarre beliefs about how to tell what someone means or something else (maybe because he hasn’t thought it through has done something that logically entails a bizarre take on meaning but doesn’t realize it…)

            (Given his conviction that Mark is a complex allegory for something-or-other, perhaps he does have some kind of odd idea about how to determine what something means.)

            I mean, if you’re sticking just to strict Bayesian confirmation, and nothing else, then all that matters is that you update your subjective probabilities according to Bayes’s theorem. You can literally set your priors and likelihoods to whatever you like (well, they have to obey the probability axioms, but that’s all). To take an extreme example, one could simply have a prior that the chances of Paul referring to a believer with the phrase “brother of the Lord” set so high that no matter how many times Paul clearly uses it to refer to a brother you’re going to be practically certain that nexttime, he will use it to refer to a believer, just like he’s supposed to (even if you’ve never seen him do so at all). This is completely unjustified of course, but it’s not a misunderstanding of Bayesian confirmation.

            So one could just e.g. not assign priors and likelihoods in a way that means Paul’s use on other occassions ends up being very much of a guide to what he’ll do on this occassion, so there would be not much point in looking at what he does on those other occassions.

            And in fact I do wonder how far these sorts of frequency analysis really do get you in determining probabilities when it comes to meaning. Working out what someone means is an interpretative act, and while frequency can sometimes help, it’s not determinative. “I need to clean the crap off my computer” probably means deleting files or removing software, not cleaning literal faeces off it, and I would not think otherwise even if the writer every single time in the past used “crap” to mean literal faeces and “junk” to refer to unwanted computer artefacts. So I probably do have my priors and liklihoods set in a way that mean looking at elsewhere a writer has used an expression may not actually shift my beliefs that much.

            (No doubt this does somehow involve frequencies and could be expressed using Bayes’s theorem in some complex way, e.g. the frequency of ” computer” used by the linguistic community)

            I appreciate that Carrier often does seem to want to reduce things to simple frequencies, so he may well be guilty of inconsistent treatment…

          • arcseconds

            Forgot to add: sorry to be bugging you about something you’ve hinted that you’re getting tired of — feel free to ignore me!

          • No reason to apologise! You’ve made some very good points. On meaning and frequencies: perhaps my comments in “Paul’s Triumph” might be of interest to you – though I plan on writing sometime in the future more on the subject. (In a nutshell: yes, I think you are right about this problem and others associated with basing likelihoods – and priors even more so – on statistical frequencies in semantic studies. I still think Bayes is of some use, at least in cases where contextual priors and likelihoods based on very consistent patterns of lexical realisations of concepts point both in the same direction, i.e. where in comparing two hypotheses we don’t have to deal with the question of whether the likelihood-advantage is “enough” to tip the scale.)

            You are right that there can be many different reasons for why an application of Bayes’s theorem ultimately fails. In the case of Godfrey’s blogpost on Gal 1:19, for example, I’d say that he probably understands the theorem (and actually pretty well summarises it in everyday language) – he’s just inconsistent in the application of the individual elements to the real world. I.e. he first says the wording of Gal 1:19 is the new evidence, then suddenly everything that should already be incorporated into the prior is lumped together with it, etc. Also, I should clarify my comment on Carrier: I guess he also “understands” Bayes. For actually there isn’t much to understand. It’s not complicated. I mean: I even get it. So I assume that he also understands how updating priors works and what role likelihood-ratios play. (As I said, I haven’t read his first book, where he probably summarises this in more detail and demonstrates his knowledge of these basic facts.) However, I would maintain that the problem with his analysis is not only the “quality” of the numbers that he uses. For example, I make a Bayesian argument for a specific meaning of 2 Cor 2:14 in my book “Paul’s Triumph.” I think I apply the theorem correctly but even if I do so, perhaps my analysis of the TLG data is not good, perhaps I’ve missed attestations in the papyri of relevant occurrences, my interpretation of some attestations is wrong, my understanding of the context rests on assumptions that can’t be defended, etc. But I think in Carrier’s case I’d go further: anyone who says that for a complex series of historical series of events being E he’s got a hypothesis for which the likelihood P (H I E) = 1, seems not to have understood the concept of the likelihood-factor. To move from something being “explainable” in a framework to it being “predictable” to it having the probability of “100%” on the assumption of that hypothesis in my view shows that this person can’t have a good grasp of what P (H I E) actually means. Similarly, if a scholar analyses the occurrences of an expression in a corpus and how often different meanings are attested and he or she then claims that P (word choice X I meaning Y) can be deduced from this (without looking at other lexical realizations of meaning Y, then this person again does not seem to have a good understanding of a fundamental part of the theorem – even though he or she might be able to describe it correctly and perhaps even apply it correctly under other circumstances. So yes, Carrier understands Bayes in the sense that he “update[s his] subjective probabilities according to Bayes’s theorem.” But the way he determines individual elements goes beyond bad research but involves serious misunderstandings what the individual components express – or, if it doesn’t, it’s a rather extreme example of the possibility of the intellectual laziness you mention. I call it “extreme” because I really can’t imagine that someone thinks so little through his argument that he might actually think that the phrase “brother of X” might favour the semantic hypothesis ‘fellow Christian’ over ‘physical relative.’ Even without looking at the data in greater detail, you need to have some awareness that the phrase is probably the default option for expressing the latter thought and that there are by contrast many and frequently used options for the former, some of which you even mention. So what would the “laziness” consist in here? In not thinking about what H and E actually represent? In just beginning some kind of statistical analysis involving E and then assuming that this must be the likelihood-element? Perhaps this kinder interpretation of what Carrier is doing is possible. I am not sure. And in the end it’s not very relevant for me personally: for scholarly interaction you need a certain level of scholarship. Of course, we often write books just to show that other peoples’ books are full of mistakes. That’s our basic motivation, right? 😉 But in these cases, there are at least some other readers who can appreciate the fact that these mistakes are pointed out because they “speak the same language.” In Carrier’s case, one would have to write a huge book explaining Bayes and why one can’t do with it what Carrier uses it for if one is a responsible scholar – just to reassure the peers who in the first place never thought that Carrier’s argument might be legitimate (even though they perhaps don’t understand Bayes and their reasons for rejecting his proposals might be rather less justified) and to provoke those who in the end don’t want to be convinced? That’s also why for years I haven’t commented on the whole issue. Perhaps I should just have stuck with that. 🙂

          • P.s.: But yes, your reference to the gospel of Mark is a good example for your point. I think the same could be said about Carrier’s treatment of the evidence from Acts: Perhaps his likelihoods are fine (which I don’t think they are) and he incorporates them correctly. The problem is that it’s kind of strange to talk about the “best” possible likelihood (with regard to historicity) – after having just incorporated the worst possible version of background knowledge imaginable, i.e. a set of assumptions no-one would be willing to agree to. Such an argument might be formally correct but is of course not very persuasive – plus, it also seems to be dishonest. For why pretend that you are gracious when it comes to the likelihoods if you do so only against the backdrop of not allowing any of the usually held beliefs concerning the work in question? (Well, I guess he’d claim that his characterisation of Acts is indeed faithful to the scholarly opinion – consensus? – on the matter? That’s at least what some of his footnotes seem to imply. Then, again we are of course “only” in the territory of – horribly – bad interaction with the secondary literature…). Ok, enough! 🙂

          • arcseconds

            To move from something being “explainable” in a framework to it being “predictable” to it having the probability of “100%” on the assumption of that hypothesis in my view shows that this person can’t have a good grasp of what P (H I E) actually means.

            Wait, what? Where does he do this?(*) That’s bad even for Carrier, and I agree, someone doing this is playing so fast and loose with likelihoods that they certainly haven’t internalized what it means and possibly haven’t understood even in a moment when they are focused on it — or they just don’t care. It doesn’t seem like probability theory is providing any real constraint on the reasoning if that’s what you’re doing.

            The only likelihood that should get a literal 1 is the case where the hypothesis is logically entailed by the evidence. Occasionally things might be so certain that we ’round up’ to 1 for most purposes, but we must keep in mind these are really 1 minus a tiny bit, because otherwise no evidence could ever count against them (assuming Bayesian conditionalization). But even here we would be saying there is no other possible explanation (aside from, maybe, truly bizarre and massively unlikely explanations, like time travelling aliens), which seems out of place in ancient history (assuming we’re in more contentious and less evidenced areas than ‘Julius Caesar existed, given all the evidence to this effect’).

            At any rate, you’d want an argument to this effect, not just an increase in confidence as you continue to talk about it, which is what you make it sound like Carrier is doing.

            (I have kind of wondered whether Carrier’s project is really just him talking himself into mythicism and other left-field ideas. He didn’t start off a mythicist, and unlike common-or-garden mythicists he doesn’t think mythicism is completely obvious, you need his insightful analysis to get there. )

            I should point out that I’m coming from the position of Bayesian epistemology, which I’m a (not particularly fervent) believer in, in the sense that as far as I know it’s the best available theory as an account of ideal empirical reasoning. I studied this stuff once upon a time, although I’m pretty rusty now, and I perhaps never ‘grokked’ it as soundly as might have liked. The model of reasoning there is an ideal agent equipped only with Bayesian conditionalization (and other applications of probability theory. And deductive logic). What priors (and likelihoods) they have is a matter of debate, but some think “any” is OK. As I said earlier to John, as far as I know Carrier doesn’t engage at all with the ‘objective’ Bayesian stuff about how to select priors prior to any consideration of the evidence, so he seems at least in practice a more ‘subjective’ Bayesian. He does of course try to motivate his priors a bit.

            (I’m a bit ambivalent about how useful explicit Bayesian treatments are in actual empirical work. Discussing Carrier has actually made me more pessimistic about this, not because of Carrier himself particularly but because it’s made me more aware of the difficulties of casting a decent historical argument in terms of probability theory.)

            You seem to be coming at this from the perspective of a historian wanting to use Bayes’s theorem as a tool — which is fine, I’m sure I could learn a lot from your treatment. So you’re expecting a certain amount of prep work to marshall the data before getting out Bayes’s theorem, and you’re treating this prep work as part of the activity of a Bayesian historian, and failure to do this prep work is failure to ‘get’ Bayes (perhaps ‘failure to act as a responsible Bayesian historian’), is that roughly correct?

            Obviously I’m no fan of Carrier’s, I just think criticisms need to be accurate. Playing fast and loose with likelihoods so they can drift from ‘possible’ to ‘certain’ is far enough away from anything to do with Bayesian reasoning that that strikes me as entirely fair to criticize his understanding of likelihoods in that case, but failure to look or think very hard about how Paul and others use certain phrases, while a totally fair criticism, isn’t a failure of his Bayesianism, per se. He can insist that “brother of the Lord” is unlikely to mean a physical brother in his head and claim that’s his likelihood pre-conditionalization if he likes.

            Of course we don’t need to be impressed by this, we could indeed insist that “OK, that’s your unjustified pre-theoretic likelihood, now start conditioning that according to how people actually use phrases.”, as you say.

            I have also toyed (really toyed, the chances of me actually putting a decent amount of time into it was always slim to none) with the idea of a more thorough engagement with Carrier’s stuff. I’m sure that a lot could be learned by re-doing his analysis, although maybe the learning would be “it’s way too hard to apply Bayes’s theorem directly to these kinds of historical arguments and get anywhere with it”. But I think you’re right: the project is a big one, and the effort could be better spent somewhere else, I suspect. I still toy with the idea though 🙂

            Speaking of which, my footnote:

            (*) It’s not that important to me really where he does this, I’m not actually going to buy his book and look it up or anything. So don’t go to any effort to cite him!

          • arcseconds

            BTW, I think I have made a similar complaint in the past about Carrier not doing his homework with his Rank-Raglan stuff. If I recall correctly, he sets his prior for Jesus existing on the basis of this, and uses a handful of figures in order to do this. I think they’re even the figures Raglan uses?

            If someone really were to take this seriously as a potential guide to historicity (I’m really not sure why one would), they would need to do significantly more than just use a arbitrary and small selection of figures assembled to establish a different point (they’re certainly not a random sample in the statistical sense), but rather scour the literature comprehensively (or at least systematically, using random sampling methods or something to avoid cherry-picking) and show over a much larger sample that increasing Rank-Raglan counts are correlated with increasing probability of being entirely mythical.

          • John MacDonald

            It’s an interesting question. Carrier says if you were to put the names of all individuals as heavily mythologized as Jesus into a hat, the likelihood of picking the name of a historical person out of the hat is at best 1/3. One point I might raise is that just because we view those figures as mythical now, that doesn’t mean the people with whom the stories originated thought the people they were talking about never existed. Did Homer (or whoever came up with the story), for instance, think Odysseus never existed? Another question is whether we are just arbitrarily picking reference classes? For instance, instead of the Rank Raglan mythotype, why couldn’t we use “Messsianic claimants around Jesus’s time,” in which case the members of the reference class would all be historical.

          • arcseconds

            Reference classes are a big problem, yes.

            I’d go as far to say that because we really have no way of telling whether any particular mythic figure was originally historic, this particular reference class is kind of hopeless. It seems pretty clear that the Iliad is based in some way on historic events and captures norms etc. from times long before the point they were cast in the form we have them today, so it’s far from impossible, and in fact kind of likely, that some of the figures actually did participate in a war between Greece and Troy.

            With other myths we simply don’t know. Was there actually a Theseus who united the Attic peninsula under Athens? This is not impossible either, of course, so for all we know Carrier’s list of ‘mythic heroes’ could mostly be based on historical figures.

            Carrier’s defenders have been known to say “it’s just a starting point, you need to get a prior from somewhere, it’ll all just come out in the wash as he considers more evidence”.

            There are actually convergence proofs to this effect, so it’s not necessarily wrong to insist that this is the case, but they all require at least some sanity with respect to priors and likelihoods. The prior can’t be 0 or 1, otherwise Bayesian conditioning just reproduces those values as the posterior. Some proofs require ‘objective’ likelihoods, but as far as I know Carrier doesn’t attempt to engage with those theories (he’s a ‘subjective’ Bayesian, I suppose).

            (These terms are a bit confusing as a Bayesian understanding of probability is sometimes called ‘subjective probability, even in the case of ‘objective Bayesians…).)

            At any rate, if all you’re committed to is conditionalizing on Bayes’s theorem, you could always set your likelihoods such that the evidence is unrelated to the hypothesis, and then the prior is unaffected by evidence (i.e. set P(E | H) = P(E)).

            Heilig is raising some doubts as to how reasonable Carrier is being with his likelihoods. If he’s not being reasonable (e.g. Heilig has at least a prima facie case when Carrier toys with the idea that Galatians 1:19 actually supports mythicism over historicity) there’s absolutely no guarantee his probabilities will converge with the truth, or for that matter with any other Bayesian agent.

            Putting aside doubts about to what extent Carrier really is a responsible Bayesian agent, the use of the Rank-Raglan thing has always struck me as kind of bizarre. Why on earth is he basing this on work done by a disciple of a now-discredited psychologist (Freud) , and an amateur folklorist, done in the first half on the 20th century? They weren’t historians, they weren’t even really attempting to do history, and the ‘monomyth’ research programme that this is a part of has little currency any more, so far as I’m aware.

            Plus the ‘reference class’ seems highly unlikely to be robustly tracking non-historicity in any sense, as even if we want to take this mythic archetype seriously, it seems just as likely that it could ‘pull’ myths surrounding a historical figure into its orbit as it could myths where the figure has no basis in history.

            This is only going to cause a lot of eyebrow-raising from serious historians, even amateur ones who make some attempt to pay attention to modern scholarship and avoid crank treatments.

            If it’s just for an initial prior that doesn’t matter much and is going to be ‘washed out’ (as they say) by later evidence, why open yourself to this (ex hypothesi irrelevant) attack?

            One thing to say ‘for’ it is that it may have some rhetorical force among Carrier’s audience, ‘street’ mythicists still seem to be open to parallelmaniacal arguments like ‘Horus was Jesus’, so this may be familiar territory to them.

            The other thing I wonder is that mythicists often seem unable, in the background, to let go of the fantastic stories surrounding Jesus. This is very evident with ‘street’ mythicists, but both Price (in his debate with Ehrman) and Fitzgerald (in a podcast) have suddenly expressed frustration that anyone can take seriously a figure that e.g. walks on water could possibly be real. I wonder whether that’s operating somewhere in the background with Carrier, too…

          • John MacDonald

            You said

            “Heilig has at least a prima facie case when Carrier toys with the idea that Galatians 1:19 actually supports mythicism over historicity.”

            If that is Heilig’s interpretation, he is wrong on that point. Carrier says the James passage in Paul is 2:1 in favor of historicity on its face, but ends up a 50/50 split between historicism and mythicism once we analyze it.

          • arcseconds

            Heilig doesn’t say that that’s Carrier’s official pronouncement, but that he ‘entertains’ the possibility (something I expressed with ‘toys with’, which again doesn’t mean ‘believes):

            Now perhaps you might understand why I found it so outrageous that
            Carrier at least entertains the possibility that Gal 1:19 might favour
            mysticism (or is at least not far worse than the physical-brother
            alternative). The fact that he can manage to entertain such a thought
            implies that he hasn’t understood how likelihoods work. For if he had
            understood the conceptof likelihood-rations he would have had to think
            at least for a second: “The expression ‘brother of the Lord’ is a less
            obvious choice for a physical relative than it is for the concept of a
            fellow believer.” And he can’t be serious about that, can he? Of course,
            saying “brother of X” is quite a default solution for referring to a
            brother. But on the other hand, even if it’s possible to refer like that
            to a fellow Christian, it’s certainly not the first choice and he
            himself never claims so.

            (from his comment on vridar)

            This does ring vague bells with me, so at the moment I don’t think he’s making this up, and he has something which I presume is a direct quote there.

            It was inaccurate of me to say “toys with the idea” that the passage supports mythicism, so I hearby retract that. Rather the idea toyed with is that “brother of the lord” fits better with referring to a fellow believer.

            But this would still making the likelihood lower than it ‘should’ be, if not as far as actually proposing the evidence supports mythicism. f this or anything similar affects his assessment of the likelihood (P( Jesus exists | reference to “brother of the lord”), then it’s not going too far to suggest he’s at minimum subtly discounting likelihoods.

          • John MacDonald

            Thanks so much for explaining these issues with Carrier’s method so clearly and concisely!

  • Matthew

    How does God factor into all of this?

  • Gary

    “I kind of wish that Friends had done a better job of depicting academia…”
    I guess I’m an old prude. “Friends” was at first glance, entertaining. But it seemed to devolve in all the characters trying to see who they should sleep with, with no real permanent relationships except sex. And coffee.

  • Is there a particular scientific conclusion that you’re concerned about, that you want us to see has a teensy chance of being wrong?

    • I take it and you have never seen the Friends episode, and so badly misunderstood my allusion to what the character of Phoebe Buffay says as being something I say – despite everything that I articulate so plainly in my blog post?

      • Yes, I must’ve completely misunderstood.

        • Gary

          I think Phoebe’s most significant contribution to mankind, was her authorship of “Smelly Cat”!

  • jekylldoc

    Well, that’s one big reason why being willing to die for your beliefs is a lot more ennobling than being willing to kill for them.

  • Eric Kurfman

    Hope I don’t get too far of subject here. I do realize the post was written in the context of academia, but I think this inabililty to admit a person might not be wrong is one of the main problems in the current political climate in our country. Many people want to “absolutize” (sorry if that isn’t a word) every issue and make it a hill to die on. It seems especially prevalent among the white evangelical church. Probably because we are taught so much about “absolute” truth.

    Admitting there is a teeny, tiny possibility your knowledge or opinion might be incorrect is really just admitting you don’t know everything…or to put it more bluntly, that you aren’t God.

    I was having an online back and forth with an atheist, and he made some off handed comment about the resurrection and when I responded with “of course, Jesus might not have resurrected” he couldn’t believe it. I told him I was not at the tomb and I didn’t see it happen with my own eyes so of course, even though I believe in the resurrection, there is a chance I am wrong. He was shocked. He said he had “argued” with dozens of Christians and no one had ever said that to him… then it was my turn to be shocked…

    This statement, from the post, is key: “It is neither the case that there is no room for uncertainty, nor is it the case in most instances that all possibilities are equal.” As I get older I am becoming more convinced that the thinker’s personality affects the way they interpret things to a great extent, but I could be wrong about that…. 😉

  • arcseconds

    A couple of points:

    Firstly, I do think court cases are an intuitive example of probabilistic reasoning that everyone can understand. If there is a good motive, documented death threats, no alibi, witness reports that they were in the area, and fingerprints on the gun, then you’re at least approaching beyond reasonable doubt, but as a good judge or prosecuting attourney will tell you, this isn’t beyond any doubt, e.g. maybe they don’t actually care that their spouse was sleeping around, the death threats are darkly humourous jokes the prosecutor has taken out of context, they actually weren’t in the area but they had an unfortunate head trauma so they actually don’t recall, they have a double in the same city that did happen to be seen in the area, and the ballsitics report was botched and it’s not actually the murder weapon. Any of these are unlikely in themselves, and they all have to be true together to explain away the evidence.

    Which has made me think I ought to familiarize myself with a few cases so I don’t just keep making up my own mini crime dramas, but so far I’ve been too lazy…

    You may need to get past the narrative assumption in TV crime dramas that the guilt or innocence is almost always made very clear to the viewer.

    Secondly, it’s worth articulating what would need to be the case in the ‘teeny, tiny possibility’. E.g. “brother of the Lord” is a special nickname for this particular James, Mark has an otherwise unattested literary dependence on Paul, has misinterpreted this as being a blood relation and indulged in a bit of fan fic, the embarassing lack of miracles in happened to some other minor prophet and Mark’s incorporated the story somehow, Peter headed up what we might call a conspiracy to make up a contemporary figure that Paul fell for, etc.

    I think the way denialist tactics normally work is that they cast doubt on each piece of evidence individually, and those doubts might be individually plausible, but when you actually put those doubts together and show what the theory would need to be, it doesn’t look so compelling to someone who isn’t invested in the conclusion (or in embarassing the so-called experts).

  • arcseconds

    I think we can take more out of this Friends snippet then just noting that Ross should be able to happily acknowlege a teeny tiny possibility that he could be wrong (I initially wrote ‘admit’, but that suggests a failing or maybe something you’re not open about).

    Ross’s behaviour is kind of obnoxious, as we’d easily be able to see if he was defending something we disagree with. He treats her like a child. No snacks for Phoebe if she’s going to utter such silliness! He also continues hounding her over several scenes, despite her being pretty clear she’s not interested in discussing it. She tries to change the subject, he keeps pushing it; she tries to leave, he blocks her path. His silly display with the toys is nothing more than childish, and finally he turns up as ‘serious scientist guy’ with a case full of stuff to berate her with. He’s condescending and oppositional and basically is treating this as a fight the only outcome of which he’s going to accept is her capitulation. He doesn’t try to engage her with the topic or genuinely try to educate her (I suppose he’s about to at the end, but all the signs are that it’s going to be a one-way street where she has to sit quiet and let the adult talk). If he was doing this in order to convince her that homosexuality is a disease, we’d see this for the dogmatic bullying that it is.

    If he thought about it, he might also realise that convincing her is not going to be easy, and he’d need different tactics if that was the goal. He knows she has flakey beliefs, isn’t scientifically informed, and doesn’t pay much attention to anything like carefully weighing the evidence but rather tends to say whatever comes into her head.

    Phoebe, on the other hand, apparently has paid so little attention to a friend she sees often that she hasn’t realised that his entire job centers around evolution. Or she’s so thoughtless it hasn’t occcurred to her that it might be contentinous and maybe personally confronting if she just denies evolution. Or she doesn’t care…

    (This lack of awareness seems frequent enough among new agey sorts. Not just them by any means, but they usually seem otherwise kindly disposed towards people and they are aware that their ideas aren’t mainstream, unlike holders of other fringe ideas I could mention…)

    She also doesn’t seem to be taking the discussion at all seriously, and may in fact be messing with him. Phoebe is established as being kind of an airhead and a bit cheerfully oblivious, which also comes out in the acting here, but if the script was interpreted by another actress with a more sarcastic bent and we weren’t familiar with the character, the stuff about gravity and the “That was fun” at the end could easily make us believe this is a wind-up.

    The other ‘friends’ do nothing to intervene, of course, they just sit there and smirk.

    I think we could lean a lot from this. I suspect most of us here are Ross, aren’t we? How could we do better than he has done with the Phoebes in our lives? Are we being Phoebe to anyone?

    • This has the makings of a great sermon, preached on the Book of Friends.

      • arcseconds

        I thoroughly dislike Friends, it’s always struck me that the ‘friends’ are shallow, self-absorbed, kind of petty people who aren’t very nice. Frequently they’re not even particularly nice to one another, as this segment shows!

        They are also very annoying individuals: Ross is kind of whiny and moany, Joey is comically thick-headed, Chandler less annoying than the rest but still smarmy, Phoebe loud and super flakey (although she seems to be the only one to actually care for anything beyond herself and her immediate circle)…

        But maybe it’s a clever study of shallow, self-absorbed pettiness, and I haven’t noticed because I avoid it?

        • At one point Phoebe dates a psychologist who assesses them much as you do…but he comes off even worse!!! 🙂

          • arcseconds

            Hopefully there’s not a lesson there for me 🙂

            I suppose comedies aren’t about well-adjusted, high-functioning people typically, and the ones I find funny are no exception, so I’m not entirely sure why I find Friends so irritating and unfunny. I suppose one thing is that as a sit-com it’s more towards the drama/soap end of the spectrum than a pure comedy, so perhaps I’m more inclined to see them as real people (realistic characters at least) than cartoons or caricatures. But I think given how it is marketed and how people describe it other people do, too, they see this group of people as their friends, as they’ve gone through several years of their life-drama with them, and they kind of love them, as people.

            I don’t think anyone really loves Sir Humphrey as a person, they love him as a fictional character (if he actually existed, we probably ought to despise him).

            Another thing that’s always struck me about the show is how insulated they all are from any genuine concerns in life. The thing about the fantasy New York apartment (the only one who could conceivably afford the lifestyle he’s presented as having is Ross (*)) is often commented on (I like to say it’s as much a fantasy you have to just buy into as warp drive is on Star Trek), but it goes beyond that it seems to me. For example even though employment isn’t a constant with them, this doesn’t seem to matter much.

            (*) Ross is also the only one who could be considered a success by normal standards. He is also perennially painted as the loser of the group.

          • Michael I

            I’d think that Chandler would actually have had a better chance of affording the portrayed lifestyle. An “IT procurement manager” for a large company (especially one who keeps getting offered raises) could possibly make a very good salary. Probably more than most professors.

            (Joey could presumably also have afforded the lifestyle during those time periods when he had a starring role in a soap opera. We’ll ignore the fact that the soap opera he supposedly starred in isn’t filmed in NYC.)

          • arcseconds

            I thought the joke was that no-one knew exactly what he did for a living?

            I was even thinking of joking “who knows how much the Russians pay their operatives”… but as I said, I avoid the show, so what do I know…

            (I also didn’t realise that it was an actual soap opera that Joey starred in… I vaguely assumed it was some decidedly third-rate affair that also wouldn’t be paying particularly well…)

          • Chandler’s job is said to be “statistical analysis and data reconfiguration.” There is a fair amount at one point about him being the only one in the group that hates his job, so he spends a day doing aptitude and other tests to find out what his ideal job would be, and it turns out to be…statistical analysis and data reconfiguration.

            Days of our Lives is one of the most famous soap operas of all time!

          • arcseconds

            I have heard of Days of our Lives, but I didn’t realise that was what Joey was acting in!