I recently did a quick forty minute debate on the historicity of Jesus with Canadian Catholic (with James Is Tired capably moderating), which was remarkably content-rich for such a truncated timeline. The overall takeaway was that Canadian Catholic wasn’t correctly applying Bayesian probability assignments to the evidence of the New Testament (we ended up focusing almost entirely on the Gospels and Epistles, rather than extra-biblical mentions, which is IMO a more efficient approach for either side: see Historicity Big and Small: How Historians Try to Rescue Jesus). This is a common risk with someone who doesn’t actually understand how Bayesian reasoning works, but knows at least that it has something to do with probability equations, and then tries to “do history” with the resulting broken tool they just saddled themselves with.
It’s important instead to be able to model ordinary, everyday historical reasoning with a Bayesian framework first, so you get right what it is that Bayesian models are actually doing mathematically: they are simply describing correct reasoning. So if you can figure out how to translate such reasoning into Bayesian terms (for example: A Test of Bayesian History: Efraim Wallach on Old Testament Studies), you’ll actually get it, and end up with a functional rather than a broken tool. Indeed, at that point you might have that “Eureka!” experience most of us Bayesians once had (see We Are All Bayesians Now: Some Bayes for Beginners). This will then equip you to actually effectively evaluate any empirical arguments (Bayesian or not—for example: Bayesian Counter-Apologetics: Ten Arguments for God Destroyed).
And that’s literally true. All empirical arguments are Bayesian (see: Everyone Is a Bayesian). Because all empirical arguments are, really, arguments over probabilities; and any attempt to arrive at a coherent calculation of any total probability ends up at Bayes’ Theorem (seriously—there is literally no way to avoid it; I offer my own deductive proof of this point in Proving History, pp. 106-14; and I show this specifically for any principled hypothetico-deductive method, pp. 104-06, or argument to the best explanation, pp. 100-03, even generic historical arguments from evidence, pp. 98-100). Apart from actually getting it, there are two common responses to this discovery that I run into a lot: to not understand it (Hypothesis: Only Those Who Don’t Really Understand Bayesianism Are Against It), or to misunderstand it—or worse, attempt to run a con with it (Crank Bayesians: Swinburne & Unwin). I think Canadian Catholic is sincere. I don’t believe this of Swinburne or Unwin; I think they are grifters (and in popular company as such: see Crank Bayesianism: William Lane Craig Edition).
The sincere are educable. They can also be inoculated against error before they even make it. So a primer on where Canadian Catholic went wrong can benefit countless others, by steering them from the start toward a correct understanding of how to assign Bayesian probabilities to historical facts when testing any hypothesis.
“But Couldn’t Jesus’s Drowning Two Thousand Pigs Be Historical?”
This wasn’t an argument Canadian Catholic made. But it brings into focus what’s wrong with an argument he kept making: the “Couldn’t it be historical?” rebuttal to any and every claim of literary mimesis and mythic construction in the Gospels. It seemed as if his mistaken tack was that we have to “include” the probability that a pericope in the Gospels is historical (at least in its core, behind the Oz-curtain of any legendary embellishment), and “I didn’t do that.” He seemed to be arguing that if you “re-include” that probability, it should increase the total probability Jesus existed. But I actually did include that probability; it’s already in the math (you can see it in all the tables in On the Historicity of Jesus, and in the explicit descriptions of how I estimate the numbers for them, e.g. Ch. 10.13, pp. 506-09; and 12.1, pp. 596-601; and see my explanation, which he clearly didn’t read, in 12.2, pp. 601-06).
The total probability I assigned to the Gospels containing actual historical data about a real Jesus in OHJ is effectively 50%. That’s quite high. That’s saying it is just as likely that such exists, and Jesus really existed, as that it does not—which I also point out (in Ch. 10 of OHJ, cf. pp. ) still does not of itself entail Jesus didn’t exist, since he could still exist, and the Gospels contain no authentic data about him; but such a condition still removes the Gospels as evidence for the existence of Jesus. And yet the problem is that when that probability is 50%, that means the “possibility” of there being real historical data about a person is not likely enough to increase the total probability of that person being real. So Canadian Catholic tripped himself up into a possibiliter fallacy, “possibly, therefore probably.” That’s a non sequitur. What historicists need is evidence that it is probable that real data about Jesus exists in the Gospels—which means, more likely than not; which means, more probable than 50%.
How this probability ends up 50% and why that has no effect on the total probability is easiest to see in the Odds Form of Bayes’ Theorem, which is why I used that for OHJ (though the same effect is in the Long Form, as can be seen by anyone adept at reading equations). Stated as Odds, the Likelihood Ratio puts the probability of the evidence in question (in this case, the content of the Gospels) on the assumption that a claim is true (in this case, “Jesus really existed”) in the numerator, and then the probability of that same evidence on the assumption that that claim is false in the denominator. I show throughout Ch. 10 that every pericope in every Gospel is effectively 100% expected on either theory. That is, we can just as easily predict its content if Jesus didn’t exist, than as if he did. So we get 100% over 100%; which is identical to what we mean by 50/50: 100% divided by 100% is simply 1, and any result multiplied by 1 remains unchanged. Thus, such a Likelihood Ratio never changes the final probability.
Of course there is a “coefficient of contingency” being subtracted out here (see Proving History, index). The probability that a particular story would be chosen to be told, and told in a particular way, is unpredictable on either theory (we can’t start with simply “Jesus existed” and then deduce any pericope in the Gospels); but the concomitant improbability is the same either way, and so cancels out. Let’s say that shared improbability is a million to one. Then we’d have (100% x 1/1,000,000)/(100% x 1/1,000,000), or (100% x 0.0001%)/(100% x 0.0001%), which is the same as (100%/100%) x (0.0001%/0.0001%), which by sixth-grade arithmetic reduces to 1/1 x 1/1. And 1/1 is simply 1, and anything multiplied by 1 remains itself. So we can ignore all those shared contingencies (the 0.0001%/0.0001%), so we can focus on whether any difference in probability remains after that—in other words, is there any greater expectancy of any content of the Gospels if Jesus existed, than if he didn’t. And what we find is, well, no, there isn’t.
This does not mean I have concluded everything in the Gospels “is” fiction. And Canadian Catholic made the mistake of thinking that’s what this means. This is a common confusion I notice, particularly among people trying to use or criticize Bayesian reasoning without understanding it: proving that we don’t know something is true is not the same thing as proving it’s false. Yet Christian apologists especially have a hard time understanding the difference. When I conclude it’s 50/50 whether the Gospels contain authentic data about Jesus, I am not thereby claiming “the Gospels do not contain authentic data about Jesus.” To the contrary, I am claiming we do not know if they do. And consequently, we can’t use any assumption that they do as a premise. What this translates to in the Bayesian framework is simply a 50/50 Likelihood Ratio: the contents of the Gospels are just as likely either way. Therefore, we cannot use the Gospels as evidence either way. Note that that includes mythicism: the fact that we cannot prove the Gospels do or don’t contain data about Jesus means the Gospels also don’t argue for mythicism either. There is one other sense in which they do (how they establish Jesus in a familiar and countable reference class), but we’ll get to that later. Apart from that, our being unable to prove anything probably is true about Jesus in the Gospels, leaves us simply not knowing—from only the Gospels—whether Jesus existed or not.
One might ask (indeed, to explore and understand Bayesian reasoning, one should ask) how the probabilities would change if, for instance, we could prove conclusively that everything in the Gospels was indeed made up. In other words, what if instead of it being 50/50 whether the Gospels include historical data, it was more like 1/50, i.e. barely a 2% chance they did (which I think is what Canadian Catholic mistook me for arguing in OHJ, when in fact I take pains to make very clear there that it is not). In that latter case we’d then be saying we are fairly certain the Gospels are 100% bogus. Would that have changed my conclusion? This actually corresponds to a real case, one Canadian Catholic might even agree with: the non-canonical Gospels. Consider the Gospel of Jesus’s Wife, now even proved a modern forgery: I think even Canadian Catholic will agree that the probability that that Gospel contains any authentic data about Jesus is effectively zero (so, certainly no better than that hypothetical 2%). So what effect does that have on the historicity of Jesus? Well, none. Obviously. It makes no difference to whether Jesus really existed or not that someone forged a tale of him thousands of years later. Lo, the same holds for someone forging a tale of him forty years later: no effect, either way.
This might also be where the mistake occurs. By not realizing that there is no difference in effect between concluding the Gospels have a 50/50 chance of containing historical data and the Gospels having a mere 1 in 50 chance of containing historical data (even though, on the a fortiori side of my error margin, I never argue the latter, only really, more or less, the former), Canadian Catholic assumed the huge difference in those probabilities should show up in any final calculation of the probability that Jesus existed. But it won’t. Because the only probabilities that matter are the probability of either theory causing the same evidence. So it doesn’t matter how independently likely it is that the Gospels contain historical data, if their content remains nevertheless equally likely whether Jesus existed or not. The only probability that actually matters here is the differential between those two probabilities. Everything else is irrelevant. Just as we see for the Gospel of Jesus’s Wife.
In order to change this outcome, what a historicist needs is something in the Gospels they can show to be more likely historical than mythical. Not “just as likely” (hence not just “possibly”), but more likely (hence “probably”). That’s the only thing that can bump the relative odds in favor. If you could show, for example, that some pericope in a Gospel is 60% likely if Jesus existed but only 50% likely if he didn’t, then you’d have a ratio of 60/50, or 6/5, which would increase the odds Jesus existed by about a tenth (as an output of 6/5 odds would be a 6/11 probability, so if the probability started at 50%, then that evidence would increase it to about 55%; or if it started at 33%, then to about 37%; etc.). This would mathematically model the effect of the simple observation that “this content is slightly more likely to be here if Jesus existed, than if he didn’t.” That’s what you need. Or anything of the kind (and really, ideally, something much better than that). Which is why I take pains to show no such passage exists in the Gospels: every piece of them is either just as expected as a mythical device than as a signal-distorted transmitted memory. (I had previously already showed historians’ attempts to argue the reverse don’t logically work in Ch. 5 of Proving History.)
In fact, most pieces of the Gospels turn out more likely mythical than historical (and the logical effect of that observation I’ll elucidate next). But this still does not mean the same thing as “more likely if Jesus was mythical.” Because, as Canadian Catholic correctly intuits (and as I explain myself in OHJ), all sorts of myth can be built up around a historical person. In order for the Gospels to increase the probability of Jesus being a myth (and not just fail to increase the probability of his historicity, which is not the same thing), we would need something in the Gospels that is actually not as likely if Jesus existed than if he didn’t (and note, accordingly, this is what historians attempt to produce in reverse, so they at least understand the concept: see Historicity Big and Small: How Historians Try to Rescue Jesus; except when they give up on that and attempt something illogical instead: see How to Successfully Argue Jesus Existed (or Anything Else in the World)). This means not just something we don’t know is mythical or historical (that won’t have any such effect), nor even something we know is mythical (likewise); in fact, not even knowing everything in the Gospels is mythical would suffice. Because a historical man can just as easily inspire a completely mythical account of him. Unless there is something about said account that isn’t as commonly inspired by historical men as it is by mythical ones instead. In other words, is there anything about the Gospels that simply isn’t likely unless Jesus never existed?
“But Shouldn’t We Treat the Gospels Like All Other Biographies?”
Canadian Catholic’s second mistake was to resort to the Formalized Gullibility of arguing that because the Gospels share genre features with ancient biography, therefore they should be treated the same as all other ancient biographies. Notably, the actual scholars Christian apologists will usually cite in defense of this move, actually specifically denounced it (proving apologists don’t actually read these scholars; they just “heard tale” and made up their own story in their minds of what they said). In actual fact, the literature establishing the genre features of the Gospels find that they do not match ancient rational-historical biography, but rather, ancient mythical biography. In other words, the Gospels don’t look like any researched biographies of the ancient world (e.g. Suetonius’s Life of Caligula, Plutarch’s Life of Alexander, Xenophon’s Life of Agesilaus II, Tacitus’s Life of Julius Agricola). They look like the fictional ones, of either non-existent people (e.g. Plutarch’s Life of Romulus; the anonymous Lives of Aesop or the Life of Adam and Eve; the various Lives of Moses then written; the biographies of Osiris, Hercules, Homer, and so on), or of real people yet whose Lives are all but entirely made up (like the various lives of philosophers and poets analyzed in the works of Ava Chitwood and Mary Lefkowitz: see their works as listed in the bibliography of OHJ, and my discussion of their contributions in Why You Should Not Believe the Apostle John Wrote the Last Gospel).
This illustrates the crucial importance, to any valid Bayesian reasoning, of always locating and faithfully employing the correct reference class (see, for example, Jesus and the Problem of the Fraudulent Reference Class). It is epistemically fraudulent to leap from “the Gospels have biographical genre features” to “the Gospels have rational-historical biographical genre features” and then, that fallacy secured, argue “the Gospels are as historically reliable as other rational-historical biographies of the time” (which is not as hot a recommendation as Christian apologists think, but would at least carry their minimally required point). In actual fact, the Gospels belong to a different genre subcategory: mythobiography. They conspicuously lack “rational-historical biographical genre features” and instead abundantly exhibit “mythographical genre features.” Thus, if you want to estimate the prior probability that the Gospels contain real historical data about their subject, you have to look at how commonly that is the case in the actual reference class they belong to. And the result of that does not go well for the historicist.
For the Gospels I’ve already thoroughly covered what happens when we do this correctly in Jesus and the Problem of the Fraudulent Reference Class. So I won’t repeat all that here. You can delve into it there. But the upshot is simple: when we look at the correct reference class (all ancient books actually most like the Gospels, and not books least like them in the “biography” category), we find that it is actually quite rare for any genuinely historical person to receive that treatment. It’s not that it never happens; what matters to our estimation of what’s likely is that it does not typically happen. When we narrow the reference class to the most comparable, without narrowing it so far we can’t derive a usable frequency, no matter which way we do this, we always end up in the same place: it’s simply not a priori likely anyone in that category of literature really existed. Which means, the prior probability Jesus existed is likewise low—as low as it is for anyone else in that category. Because we cannot engage a “special pleading” fallacy and insist (on a basis of no evidence at this point) that Jesus is “special,” and somehow unlike everyone else he is most like. You can do this on a basis of evidence that makes him different, if you have some—which is why the prior probability is not the posterior probability (I explain this very clearly in both OHJ, Ch. 6.8, pp. 252-53, and PH, pp. 70-76). But when it comes to the priors, you have to treat everyone in the same reference class as the same. If they are most alike, then it’s just as initially likely Jesus existed as Osiris or Hercules or Moses did. Absent any evidence to the contrary, that’s how you have to conclude. And that’s why you need evidence to conclude otherwise. Hence it’s then that we notice we don’t really have any. If, of course, we’re being honest (I lay this out most clearly across Chs. 4-6 of my new book Jesus from Outer Space).
The main takeaway of Jesus and the Problem of the Fraudulent Reference Class is that Jesus is among the most mythologized figures in antiquity (even right out of the gate, in the Gospel of Mark, the first we ever hear of him being definitely placed in history); and among the most mythologized figures in antiquity, real people are rare (to a straightforward observed frequency of less than 1 in 16, or a most speculative maximum of about 1 in 3: OHJ, Ch. 6). I employ a number of metrics to determine that, none of which are honestly escapable. But one of the metrics I develop in my analysis of whether the contents of the Gospels contain any evidence for historicity as I just defined (in Ch. 10 of OHJ) is to ascertain the iterative prior probability of any pericope in the Gospels going back to a historical core. I already explained above that finding this to be below 50% (even as low as 2%, should that be where we get to, though I didn’t claim to) does not of itself increase the probability that Jesus is a myth (and therefore does not decrease the probability of historicity). But what it does do is show Jesus is heavily mythologized. Which supports the larger point that the prior probability he exists must come from the corresponding reference class of heavily mythologized people, and not the set of just any person written about then (on which point see my discussion in So What About Hannibal, Then?, which is reproduced and gathered with numerous other examples in JFOS).
I don’t think (?) Canadian Catholic had made this mistake, but one could: if one were to say there is a 50% chance of historically authentic data per pericope in the Gospels, that would entail a high probability of some authentic data in them, somewhere. Hence my conclusion is not that, but rather, that the total probability of any authentic data in them is 50%. And this derives from the method of iterated priors (PH, p. 337; OHJ, pp. 240, 509):
[F]rom [my] survey [of many of the pericopes in the Gospels] it’s clear that if we went from pericope to pericope assessing the likelihood of it being true (rather than invented to communicate a desired point or to fit a pre-planned narrative structure), each time updating our prior probability that anything in the Gospels can be considered reliable evidence for a historical Jesus, then that probability would consistently go down (or level off somewhere low), but never rise. In fact I have not found a single pericope in these Gospels that is more likely true than false. These Gospels are therefore no different than the dozens of other Gospels that weren’t selected for the canon (…). They are all just made-up stories. To change this conclusion, historicists need to find a way to prove that something about the historical Jesus in the Gospels is probably true (not possibly true, but probably true).OHJ, p. 509
In other words, if we started at one pericope, and calculated how likely it was to be historical, and got “50%” as the result, that would then become the prior probability that the next pericope was historical. So we’d go and look at that, and if no evidence changed that expectation, we’d end up with an output, again, of 50%. So that’s the posterior probability of any historical data in either of those two pericopes, and thence the prior probability of such data in the next pericope we examine. And so on. If you continue this process until you’ve processed every pericope, and still your output is “50%,” then you have determined there is at best a 50/50 chance any real data exists anywhere in the whole of the Gospels. This does not entail asserting there is none. Rather, it only entails asserting that we don’t know whether there is.
Yet you might observe from the above quote that I said I personally think we’d end up with a final output lower than 50%. Hence I only settled on 50% a fortiori. Because anything less would make no further difference (for the reasons I explained here earlier). But to understand why it’s lower, consider again simple arithmetic: if in some cases (“magically drowned two thousand pigs”; “irrationally withered a fig tree”; “walked on water”; “glowed on a mountain”; “flew with the Devil”) the probability of historical fact is actually, obviously, quite low, then on account of that we’d have valid reason to start increasing our doubt that anything else in such a collection of lore is true (see Craig vs. Law on the Argument from Contamination and For the Existence of Jesus, Is the Principle of Contamination Invalid? Cavin & Colombetti vs. Law). After all, the inclusion of these obviously mythical absurdities entails the gullibility (or dishonesty)—and thence unreliability—of anyone who’d include them without blush, as we see happened. Any author retaining such material without skepticism clearly does not know how to determine if anything is true (because, remember, Naturalism Is Not an Axiom of the Sciences but a Conclusion of Them); and therefore, their ability to do so with mundane material is not going to be much better.
This effect has to be represented in our probabilities. We cannot honestly treat such authors as “the same” as reliable, objective, rational historians. Let’s say we were so generous as to give each of these absurd tales a 1/3 relative odds of being added onto a collection of oral lore that otherwise still somehow retained real (albeit more mundane) data, then if everything else is 50/50, your net outcome (the final probability that “anything in the Gospels about Jesus is historical”) will be far below 50/50. For instance, assume just the five examples I listed. The iterative prior then will be: 50/50 x 50/50 x 1/3 x 1/3 x 1/3 x 1/3 x 1/3 x (50/50)^n = 1/243, or less than half a percent chance anything in the Gospels (that is not externally corroborated) is real. That’s how mythical they look to be. This outcome could have been different. Take the works of Josephus for example: we doubt many things in his opus, even within the sections that deal with non-mythical history (from his dubious tales of several famous mass suicides, to outright myths like his presenting the morality tale of Cannibal Mary as historical fact), but we also trust a lot, because a lot of it has been corroborated (unlike anything about Jesus in the Gospels), and most of the rest sits on higher priors due to its framing and context (he cites credible sources, and discusses material there is less basis for anyone inventing, and even good reason to avoid inventing; unlike anything about Jesus in the Gospels).
Assessed “story by story,” the fabrications in Josephus are very few relative to the remainder. Much of that remainder may sit at 50/50 odds of being true, but much of it sits higher—and there is a lot of that (we can be sure at least a hundred higher-credibility stories, over on top of countless stories whose credibility we might not know either way). So if we did the same analysis on the relevant material in Josephus (like, say, all his tales about the historical periods he isn’t using mythical sources for), and here just reasoning a fortiori rather than more meticulously than the point being made here requires, we might get something more like: (50/50)^n x 1/3 x 1/3 x 1/3 x 1/3 x 1/3 x (60/50)^100 … = 1 x 1/243 x 82,817,974/1 = 82,817,974/243, which comes to about 340,815. Which means it is over three hundred thousand times more likely “something” in Josephus is historically true than that nothing is (and that, you might be able to tell, is an outrageous undercount). This is a result of there being so many corroborated and reliably presented “stories” (the functional analog to Gospel “pericopes”) across the whole of Josephus compared to so (relatively) few fabrications (while all the material in Josephus we can’t tell either way has no effect, just as with the Gospels). Thus it matters that the Gospels contain nothing peculiarly more credible than not, and yet at the same time are full of the incredible. This puts them on exactly the opposite side of any measure of credibility from ancient rational historians and biographers—even given the fact that (for example) Josephus also makes a lot of stuff up or reports it unreliably.
This is why it is so important for a historian to find pericopes in the Gospels that are more likely to derive from a historical Jesus than to have been contrived to represent him and his message. But there just aren’t any (hence OHJ, Ch. 10). For what I mean, consider my analysis of the pericope of Simon of Cyrene (and other like peculiar tales in Mark: OHJ, 10.4, pp. 444-56). I explicitly note there that I don’t claim to have proved that was a mythological contrivance; rather, only that a full analysis of all its features (including the features it lacks), and the contextual principles of ancient authorship (e.g. no author ever added material for no reason, and their reasons are always made evident by what they do or don’t say or how they employ the material, its overall and immediate context, and that author’s overall discourse style), leaves us with equal odds it was contrived to a purpose or preserves some sort of historical memory (however distorted).
Until we apply the iterated prior, of course. By itself (if, for example, the Simon of Cyrene tale was the first pericope we started with), we’d honestly get nothing better than a 50% chance it’s history rather than myth. But when we then notice how routinely Mark fills his book with symbolic and allegorical myth and deliberately crafted irony, including many convenient symbolical meanings of the characters’ names and roles that he populates his narrative with, the prior probability that he is doing the same with Simon of Cyrene is substantially above 50%. Whereas the complete absence, for example, of Mark ever citing sources prevents our making the converse case for his doing that with Simon of Cyrene and his sons; likewise any other historicist “interpretation” (none have any support in Markan discourse practice and style or any individual pericope’s framing or context). Thus the “possibility” Mark is “doing history” suddenly only here, when throughout his book he’s regularly been mythologizing (see, e.g., Mark’s Use of Paul’s Epistles and Why Did Mark Invent an Empty Tomb?), is not credibly at all high (see, e.g., Mark Goodacre on the Historicity of Jesus’s Execution). It’s not even 50%. But since it’s being much lower than that changes no result, we don’t need to evaluate what that probability is. We know it’s no more than 50/50. And that’s all we need to dismiss the Gospels as completely unusable for reconstructing anything about a historical Jesus.
Again, the probability that there is any fact in the Gospels is not the probability that the Gospels would look that way if Jesus existed or not; because “historicity” could produce a “totally” made-up biography almost as easily as a “mostly” made-up one (OHJ, Ch. 6.7). So even the complete, guaranteed absence of any history in Mark (as with the Gospel of Jesus’s Wife) would not, by itself, be more unexpected if Jesus existed than if he didn’t. That’s more or less an expected an outcome either way (hence, that’s how I scored it: as increasing neither historicity nor mythicism, cf. OHJ, pp. 506-09)—until we consider what reference class this puts Jesus in. Then it starts to look bad for historicity. Real people rarely got such mythologized biographies written of them; that usually happened to mythical people. And all of this is just logic. It’s not some mysterious sorcery with numbers. There is no logically valid way to avoid these conclusions. All that the Bayesian framework does is reveal why.
“But Doesn’t That Require an Elaborate Conspiracy Theory?”
A third argument Canadian Catholic muddled up even more. The gist of it was: all extant late second century Christian authors were historicists (he mentioned for no discernible reason Pantaenus and Heracleon, from whom we have no texts, only references to what they wrote in other authors; he could more easily have named Justin and Ignatius, as even earlier authors from whom we do have relevant texts; he instead went in the opposite direction chronologically and mentioned third century authors Clement and Origen), therefore to deny historicity we are committed to believing an implausible global conspiracy to suppress all record of original Christianity. This was a strange argument because I extensively address this in OHJ, and Canadian Catholic claims to have read that, yet he seemed utterly clueless as to what my responses to this argument would be (see OHJ, index, “conspiracy theories,” and most especially Ch. 8.12, 7.7, and 6.7; I expand on all this in Ch. 7 of Jesus from Outer Space).
The Bayesian lesson here is that all probabilities—literally every single probability in any Bayesian equation—must be conditioned on total background knowledge (usually b or k in most fully expressed equations). That means you can’t just “come up” with probabilities that ignore everything else we know in the universe (about everything, from physics and human psychology, to what actual evidence survives from any given century of the Roman Empire). Those probabilities are dependent probabilities, not independent. That means they are the probability given everything else being the case. And in this case, that means the fact that almost all evidence pertaining to the entire first century of Christianity’s history is completely lost. In fact, of what few things survive for us to examine now, they consist solely of what’s in the New Testament, and texts that never reference a historical Jesus (e.g. 1 Clement, Pliny the Younger, the earliest reconstructable redaction of the Ascension of Isaiah). Moreover, all subsequent Christian authors demonstrate to us that they also had no knowledge of any other texts than these. They had no other sources regarding their religion’s first hundred years other than what we have (or else, none they chose to ever mention, much less quote or describe). Thus, for example, when affirming the historicity of Jesus, from the first hundred years of Christianity, they could only cite the Gospels. They evidently had nothing to corroborate them by (OHJ, Ch. 4, Element 22). This complete hole in their own religion’s history was so severe and complete they eventually had to resort to fabricating some—faking both documents (OHJ, Ch. 5, Element 44; with Ch. 7) and entire histories (e.g. How To Fabricate History: The Example of Eusebius on Alexandrian Christianity).
That is not conjecture. It is not some ad hoc “commitment.” It is an indisputable empirical fact. And all probabilities must therefore be conditional on that being the case. So, we don’t need any conspiracy theory to explain why the historicist church, the only one to politically survive, preserved only historicist texts (and texts they could spin as such), none of which evinced any knowledge of any evidence for historicism other than the Gospels, and texts and interpretations derivative of them (and thus incapable of corroborating them), including forgeries and fake histories. By contrast, when we look at the earliest verifiable example of an extant Christian author outside the New Testament attempting to take a historicist stance, we find a deeply suspicious desperation to push historicity against fellow Christians who evidently were denying it (see How Did Christianity Switch to a Historical Jesus? and the corresponding chapter building on that in Jesus from Outer Space). But we don’t get to hear from them; their texts were all destroyed, which we know for a fact, because we don’t have them.
We see this again in the forgery of 2 Peter (see OHJ, Ch. 8.12), which may date from around the same time (the Ignatian letters were traditionally dated to the 110s AD, but the latest trend in scholarship is to place them decades later, which still makes them earlier than any author Canadian Catholic named, and probably earlier than any known: see OHJ, Ch. 8.6, with Ch. 7.6, p. 274, n. 41). Even by the 160s (before anyone Canadian Catholic named), that the Christians might have invented Jesus is imagined by Justin a common enough accusation that he has his fictional Jewish opponent Trypho suggest it offhand (“you invent a Christ for yourselves from a baseless report,” Dial. 8.4). Justin’s rebuttal consists solely of gainsaying (“we have not believed empty myths or unproven stories, but [stories] filled with the Spirit of God, bursting with power, and flourishing with grace,” Dial. 9.1) and a posture of indignation (“at that I rose up and was about to leave,” Dial. 9.2), threatening to depart if they dare even suggest it again. All of which, those of acute mind will notice, amounts to admitting the total lack of any evidence by which to rebut the charge. There was no evidence known for him to cite. Thus, they had to invent it: 2 Peter was forged to fabricate an eyewitness testimony to the actual historicity of Jesus, in a passage that in fact sounds an awful lot like Justin in its worry, and exhibits the same denunciative desperation as Ignatius: “we did not follow cleverly devised myths when we told you about the coming of our Lord Jesus Christ in power, but we were eyewitnesses of his majesty.” Note this says “we did not follow myths,” rather than “we” did not “create” myths: so this author sees the opposition as claiming no one even met a historical Jesus so as to make up stories about him, thus they had to invent an eyewitness testimony to doing so. Evidently no real ones were available. And like Justin and Ignatius, the forger of 2 Peter’s solution to these pesky “mythicist” Christians was to declare them evil and to be shunned.
Possibly earlier than Ignatius is Papias, but we don’t have what Papias wrote, only scattered and incomplete quotations, and they only verify the point: Papias says he had no sources (no books, no texts, no letters) to consult to corroborate any Gospel, but only scattered, unverifiable rumors going about in his own day, which we know were false (e.g. he was persuaded by these “rumors” to believe Matthew was written in Hebrew and independently of Mark, and that Mark wrote down the gospel of Peter when in fact his Gospel supports the opposing faction of Paul); and his gullibility regarding those rumors is so infamous as to well establish Christians clearly were the easiest of dupes (see my discussions of Papias in Did Polycarp Meet John the Apostle? and Why You Should Not Believe the Apostle John Wrote the Last Gospel and Is Jesus Wholly or Only Partly a Myth? The Carrier-MacDonald Exchange; and these don’t even cover the most absurdly gullible things Papias believed, such as that Jesus taught there would be talking grape clusters the size of houses in paradise and that Judas became so filled with worms he bloated to the size of a wagon trail: OHJ, pp. 324-25, and “Papias and the Miraculous Vines” by Philip Jenkins).
With this as the background, we need no conspiracy theory to explain how historicists could end up immovably convinced of their own position and take no bother to preserve any contrary view. Hence I discuss several pathways in OHJ, Ch. 8, that would explain this evidence; none require any “conspiracy.” More to the point, we don’t need a theory, because we have the fact. It is idle to bicker over what theory explains a broken window, when the fact that the window is broken is indisputably in evidence. It would be silly to insist that window can’t be broken “because” believing it was would require an elaborate global conspiracy. It doesn’t matter how the window came to be broken. We can plainly observe something broke it. Ditto the state of Christian evidence: that all record of what happened in the church between its original cosmic gospel and its historicization is gone, and wasn’t known even to the earliest authors we have discussing their adamant historicist beliefs, is a fact. Not a conjecture. Again, as I note in How Did Christianity Switch to a Historical Jesus? and Jesus from Outer Space, we have no record, either, of how or when belief switched from Jesus appearing in inner mystical visions (Galatians 1:11-12 & 1:15-16) to hanging out in an animated cadaver for a month of dinner parties (Acts 1:3-11). Nevertheless, it happened.
It need not have been this way. Imagine the counterfactual condition of having a good solid record of the first one hundred years of Christianity, at least as good as we have for the origin and evolution of the modern Cargo Cults or the turning of the historical Haile Selassie into the mythical version of Ras Tafari (for both see OHJ, index), or even Mormonism and Scientology. If we had that, we would indeed be able to make significant probabilistic assertions based on what is and isn’t in that record. Consider:
- In the case of Ras Tafari, enough evidence survives for us to be able to firmly establish he existed and that the Rastafarian gospels are just myths about the man (in which almost nothing is historically true, equivalent to the currently dominant position of mainstream Jesus scholarship). One could indeed rebut any claim that Ras Tafari never existed by appealing to that continuous record attesting him, and dismiss any attempt to deny his historicity as requiring some elaborate conspiracy theory (to “plant” and “alter” all manner of records and reports). But we aren’t in this position with Christianity. Everything we have to prove Ras Tafari’s historicity, does not exist to prove Jesus’s. It’s gone. And even the earliest Christian commentators we have didn’t know of any either.
- Conversely, we were lucky to have scientists on the island where and when the Cargo Cult movement began and thus have a reliable eyewitness account of it, and of subsequent events, by which we can confirm John Frum and Tom Navy never existed, and that the attribution of an entire religion to one or the other’s ministry is mythical. We don’t in that case have a complete record (like a continuous year-by-year collection of correspondence or scientific or journalistic reports) so as to reconstruct when that subsequent mythical construct was proposed and how it spread and came to be believed; but we have enough evidence to prove that that is indeed what happened. So questions about “how” it did cannot be posed to challenge that it did. All the more so in the case of Christianity, where the silence of the record extends not just a matter of years as in the case of the Cargo Cult movement, but an entire century.
An even better example is Ned Ludd (mythical founder of the Luddite movement; OHJ, index), where the newspaper and documentary record in England is solid all the way through—which is how we can now tell Ned Ludd never existed. Yet within forty years, he was unquestioningly believed real, and entire biographies written about him. No conspiracy theory required. We also have the example of the Roswell “flying saucer and alien autopsies” (OHJ, index), where excellent documentation survival allows us to prove not only that that never happened, but exactly when and by whom the idea originated, and how belief in it then spread even despite stalwart refutations and a documentary record fully refuting it, conditions hostile to such legendary development not even present in antiquity (where myths were even more likely to spread and eclipse the truth: see Chs. 7, 13, and 17 of Not the Impossible Faith and No, Mr. Christian, A.N. Sherwin-White Didn’t Say That. And Even What He Did Say Was Wrong. and The Rain Miracle of Marcus Aurelius: A Case Study in Christian Lies). No conspiracy theory was required for this to happen. Or for any of these historicizations of myth. And since all probabilities must be conditioned on background evidence b, we need no “conspiracy theories” in b. We already have all we need without it.
“But What If the Forged Epistles Are Authentic?”
Extending his first mistake from the Gospels to the Epistles, Canadian Catholic then tried arguing we have to “include” the probability that a forgery or interpolation among the Epistles is instead authentic, and “I didn’t do that.” He seemed (again) to be arguing that if you “re-include” that probability, it should increase the total probability Jesus existed. But I actually did include that probability; it’s already in the math (this time, regarding the forgeries, covered in OHJ, Ch. 7.1, 7.3, and 7.7; and regarding 1 Thessalonians 2, Ibid., Ch. 11.8, pp. 566-69). Here there is at least the substance of a different point, which I already discussed in Kamil Gregor on the Historicity of Jesus. It differs from the case of the Gospels because any significant probability of the authenticity of strong references in Paul would affect the final probability of historicity because we aren’t here dealing with the alternative that he is making them up (that would be too improbable to credit, and thus mythicism cannot survive on such a supposition, as I explained to Gregor).
Regarding interpolation in 1 Thessalonians, the probability of that being instead authentic I show in OHJ to be less than even a single percentile; it therefore falls far below my margin of error (of over thirty percentiles), and thus has no observable effect on the mathematical result (as I also explained to Gregor). This is why I don’t have to account for it. Indeed, at the top end of my margin of error I round to the nearest whole percentile, so a variance of a fraction of a percent will be invisible there (and 1 Thessalonians being authentic won’t be included on the lower margin). This is why we don’t have to “take into account” in our calculations and estimates every harebrained “possibility” anyone can think of (of which there will be billions). That, and the fact that there are as many harebrained possibilities to conjecture for either hypothesis (mythicism or historicity) that their probabilities therefore cancel out (ibid.).
All the same goes for “forged epistles,” like (as Canadian Catholic cited) 1 Timothy—the pertinent passage in which (1 Tim. 6:13) I actually do discuss in OHJ, as you can tell by simply consulting its scripture index (you’ll find there that it’s a more problematic passage than he thinks, but even apart from that, there’s no point in addressing it, as there’s no appreciable likelihood that Paul wrote it). I didn’t run the math in that case, because the pastorals being forged is the mainstream consensus, and for efficiency of word count I didn’t need to argue points overwhelming consensus already grants (OHJ, p. 65). But if I were to do, we’d definitely get well below a one percent chance 1 Timothy is authentic. Its stylistic and theological differences from Paul (and similarity to second century Christian dialect and concerns) are far too severe to imagine it at all likely to have still been written by him (see Katarina Laken, An Authorship Study on the Letters of Saint Paul and Bart Ehrman, Forgery and Counter-Forgery, pp. 192-217). So that probability falls below even my rounding error. It therefore need not be considered.
Indeed, from a “harebrained possibility” perspective, it’s essentially as likely that Galatians was wholly forged as that 1 Timothy was not; so those two “harebrained probabilities” cancel out anyway. This is the problem with possibiliter fallacies: as soon as you start acting like you get to pull that on someone, you’ve thereby given them epistemic permission to pull the same right back on you. And then maybe you’ll realize why it’s a fallacy.
“But Doesn’t Doubting Historicity Require Too Many Commitments?”
By now we can tell that this, Canadian Catholic’s final refrain, simply isn’t true. The concept is correct: the more ad hoc commitments a theory requires in order to maintain it, the less likely it has to be on account of that (see “The Cost of Making Excuses”). But as I explained in Kamil Gregor on the Historicity of Jesus, I do not adopt any such commitments in On the Historicity of Jesus for reaching my upper margin of error, the only probability that matters to the point. And accordingly, neither critic has ever identified a single such thing in my study (and you can’t just “claim” something exists, and fail to present any actual example). “Assumptions” I condition all my a fortiori probabilities on I meticulous prove are facts, not conjectures (e.g. OHJ, Chs. 1, 4, 5, and 7). And insofar as any are mere “commitments,” they are matched and thus canceled out by the comparable commitments of historicists (e.g. OHJ, Chs. 2, 3, and 12).
By contrast, the excess commitments required of historicists are more dubious and ad hoc than any I require. Mythicism, for example, does not require assuming anything historical is or is not in the Gospels, whereas historicity does; mythicism already predicts the odd silences in Paul’s letters, whereas historicists must resort to implausible and unevidenced ad hoc suppositions to explain them away; mythicism reads the texts as they are, in context, whereas historicists have to “reinterpret” what the plain language says, often by ignoring its actual context. And so on (e.g. OHJ, Chs. 6 and 8-11). This is why the mathematical model is so important: it actually allows you to do what Canadian Catholic and Kamil Gregor attempted to do—find unevidenced assumptions whose improbability has not been accounted for. Unfortunately, both simply ignored what’s actually in my study when doing that. They thus whiffed on their ability to use Bayesian reasoning effectively.
These same points about how to correctly model our reasoning with actual probabilistic math carry over to every other point Canadian Catholic attempted to make. I have shown elsewhere how it actually follows from defensible Bayesian estimation that There Is No Logically Sound Case Against Interpolation in 1 Thessalonians 2 and that Yes, Galatians 4 Is Allegorical (he did not seem to know that; he thus prepared poorly for our debate). Hence, we must remember, any prior odds must reflect the actual frequency in comparable cases—which have to actually be comparable enough to count. Your reference class must be credible, and complete (no cherry-picking; everyone in the same hat must be treated equally). All probabilities must be conditioned on background knowledge (not contrived in ignorance or disregard of it). And when there is no evidence either way—when some fact is just as expected whether Jesus existed (“as a vestige of historical memory”) or not (“as a piece of allegorical mythologization”), it increases the probability of neither. In Bayesian terms, it’s then just as likely on either theory, and thus the likelihood ratio is effectively 50/50, hence 1/1, hence simply 1—and any probability multiplied by 1 remains unchanged. This does not require proving it’s “a piece of allegorical mythologization.” All it requires proving is that we have no more evidence it is not, than that it is. We don’t know. So it’s 50/50. And 50/50 means no change in probability. You must respect all of these principles if you want to construct a valid Bayesian model.
By contrast to mythicism, for example, it is actually historicists who have to pile up ad hoc suppositions not in evidence to get there to be no interpolation in 1 Thessalonians, for example, or to convert an argument Paul explicitly says is allegorical from top to bottom to suddenly burp out an incongruous historical assertion that (it turns out) he then must never have explained the relevance of. Why is Jesus being born “of a woman” at all historically relevant to the argument is he making there? It had no role then in defining one’s status as a Jew or under covenant law (only patrilineage and circumcision could do that; and Paul doesn’t say this woman was Jewish anyway, so that can’t have been his point in mentioning it). Nor is this something one usually has to assert about anyone. Who isn’t born of a woman? Indeed, if Jesus was a historical person, why would there be any confusion as to whether he was? By ignoring the context (which perfectly explains this as referring to the allegorical “women” Paul explains we can all be born of, the corrupt sublunar world or the celestial order of the heavens), historicists end up making this text less explicable, indeed unintelligible, requiring them to pile on a bunch of ad hoc unevidenced suppositions about what Paul is then supposed to have meant. Mythicists have no need of those commitments. They need merely take the text, in context, exactly as written. As such, it entirely explains itself. No commitments required. In Bayesian terms, therefore, it is the historicist reading of this verse that is improbable, not the other way around. I nevertheless score it the other way around in my study (On the Historicity of Jesus, pp. 592-600), to account for the possibility I missed something (and thus I counted it as evidence for historicity!)—a generosity historicists would be horrified to reciprocate.
In the end, no matter what historical argumentation you are modeling so as to evaluate the merits of, lack of knowledge either way must always come to a 50/50 likelihood and thus have no effect on the final probability of any theory. If some piece of evidence is just as expected to look that way no matter which theory turned out to be true, then it simply makes no difference to the probability of either theory. You can’t just “assume” one thing over another instead—that would be a straightforward fallacy of circular argument, which will be captured in the invalid likelihood ratio you would then be asserting. Likewise, all your probabilities must be conditioned on background knowledge. So, if all the documents you need to test a claim are lost, you cannot argue from what was or was not in them; because you don’t have them. Accordingly no “conspiracy theory” is needed to explain why something you expect is not reflected in the record—if there is no record, then you don’t know what it would have reflected had it survived. And this includes background knowledge’s effect on prior expectations. If a document looks like a piece of faked nonsense, then the probability it is must derive from the frequency of all other documents that look that way being that way. You cannot ignore the evidence of past cases; nor try to derive a preferred frequency from a false reference class (like claiming the Gospels look like rational-historical biographies, when they explicitly don’t).
As such, Bayesian methods only quantify and thus describe, and therefore expose ordinary everyday reasoning about history, revealing when it is valid and sound, and when it is not. But to use it that way, you have to use it correctly. Just like any other method or logic.