Richard Carrier’s “Proving History: Bayes’s Theorem and the Quest for the Historical Jesus” Chapter 1 (A Review)

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

by Neil Godfrey

Till now I’ve always been more curious than persuaded about Carrier’s application of Bayes’s Theorem to what he calls historical questions, so curiosity led me to purchase his book in which he discusses it all in depth, Proving History: Bayes’s Theorem and the Quest for the Historical Jesus.

Before I discuss here his preface and opening chapter I should be up front with my reasons for having some reservations about Carrier’s promotion of Bayes’ theorem. (Allow me my preference for Bayes’ over Bayes’s.) I should also say that I’d like to think I am quite prepared to be persuaded that my resistance is a symptom of being too narrow-minded.

My first problem with Carrier’s use of the theorem arises the moment he speaks of it being used to “prove history” or resolve “historical problems”. For me, history is not something to be “proved”. History is a quest for explanations of what we know has happened in the past. Historical problems, to my thinking, are problems having to do with how to interpret and understand what we know has happened in the past. The milestone philosophers of the nature of history — von Ranke, Collingwood, Carr, Elton, White — have certainly spoken about history this way.

I have always understood that where there is insufficient data available then history cannot be done at all. Ancient history, therefore, does not allow for the same sorts of in-depth historical studies as are available to the historian of more recent times. Historical questions are necessarily shaped (or stymied altogether) by the nature and limitations of the available sources.

Criteriology (I take the term from Scot McKnight‘s discussion of the historical methods of biblical scholars in Jesus and His Death) has always looked to me like a fallacious attempt to get around the problem of having insufficient data to yield any substantive answers to questions we would like to ask. We don’t know what happened? Okay, let’s apply various criteria to our texts to see if we can find out what “very probably really did happen”.

Carrier’s introduction of Bayes’ theorem has always appeared to me to be an attempt to salvage some value from a fundamentally flawed approach to “history” — the striving to find enough facts or data with which to begin to do history.

I should add that I do like Carrier’s offering of hope that Bayes’ theorem can promote more rigorous and valid thinking and applications of criteria. But I can’t help but wonder if in the end the exercise is an attempt to patch holes in the Titanic with admittedly very good quality adhesive tape.

What is really accomplished if we find only a 1% probability for the historicity of Jesus? Improbable things really do happen in the world. Otherwise we would never know chance and always be living with certainty. Or maybe I’m overlooking something about Carrier’s argument here.

Not that I’m a nihilist. I do believe we have lots of useful evidence to assist us with the study of Christian origins. I think scholars are agreed that pretty much all of that evidence speaks about a Christ of faith (a literary figure) and not an historical figure. That’s where our historical enquiry must begin — with the evidence we do have. After we analyse it all and frame such questions as this sort of evidence will allow us to ask then we can begin to seek explanations for Christian origins. This will probably mean that we will find answers that do not address the life and personality of someone who is hidden from view. Our understanding will address religious developments, ideas, culture, literature, social developments. We will probably be forced to conclude — as indeed some historians do — that if there is an historical Jesus in there somewhere he is irrelevant to our enquiry.

So that is where I am coming from.

Let’s see if I am being too narrow-minded. Here is my reading of Carrier’s preface and opening chapter.


Carrier informs us from the start that this volume is preparation for his next, On the Historicity of Jesus Christ, in which he will use Bayes’ theorem to test the arguments for the historicity of Jesus. Till then, this book is an exploration of Bayes’ theorem as a method to be applied to historical studies albeit with specific application to the question of the historicity of Jesus.

He makes bold claims. This method is of general applicability, he suggests, for testing “all historical theories of any sort whatever”. I wonder. Is history a series of progressions of movements and forces of various kinds? Or is it after all simply one damn thing after another and without any meaning or even existence apart from the biases and personal meanings through which an historian imposes order on those events?

The aim here is to develop a formal historical method for approaching [historical] debate, which will produce as objectively credible a conclusion as any honest historian can reach. One need merely plug all the evidence into that method to get a result. That’s a bold claim, I know; but the purpose of this book is to convince you . . . All I ask is that you give my argument a fair hearing. (p. 7)

I cannot deny I approach this claim with scepticism. I don’t even consider “Was there an historical Jesus” to be an historical question. No evidence (only evidence for a Christ of faith, as I mentioned above) so no scope for asking the question. The data we have only allows us to seek historical explanations for the origins of Christianity as a social or religious movement. But Carrier has asked for a fair hearing so I’m willing to give him a go.

Carrier lays out his biases — as I have laid out mine here. Quite rightly, as an atheist, he does not need Jesus to be a myth but Christian believers do need Jesus to be historical. I would add some few Christians, for example Albert Schweitzer, have argued for a rebuilding of Christianity upon “a new metaphysic” and not on an historical Jesus at all. I would also suggest that Jesus has an iconic place in our culture that extends beyond religious institutions, and some people (including non-Christians) have staked careers and reputations upon the study of this Jesus. The debate is not entirely between believers on one side and unbelievers on the other.

He points out that both sides of the debate have advanced dubious arguments at times. That’s a reason for his book. To help lay readers and experts evaluate these arguments. Carrier also promises to avoid technical or academic jargon and to attempt to reach wider audiences by means of “a style more attractive and intelligible to ordinary people.”

He gives special thanks to Atheists United for supporting his research.

Chapter 1: The Problem

Carrier begins with the failure of historical Jesus studies to make any advances in reaching agreement upon anything about this Jesus. He cites well known biblical scholars who have conceded the unreliability or invalidity of the criteria that have been used in the most recent “quest for the historical Jesus”: Gerd Theissen, Stanley Porter, Dale Allison.

For Carrier, there are three reasons these criteria have failed to yield progress in the historical quest:

  1. Specific criteria are invalidly applied
  2. Specific criteria are invalid
  3. The “Threshold Problem”: that is, how do we decide if and when a question that passes criteria tests tells us anything more certain than simply that something plausibly may have happened?

It is that third point that Carrier describes as the “fatal flaw” in the “entire methodology” of the use of criteria to investigate the historical Jesus. I agree. But Carrier will argue for a fix to this problem.

Carrier surveys the evidence for the spectacular failure of all attempts of biblical scholars to find “the historical Jesus”. He points to the many scholarly variations of Jesus and says all are very plausible but obviously they cannot all be true. I don’t know if very many are plausible at all when it comes to thinking through how each one ever produced a religion that within a handful of years exalted that person to pre-existent and divine status. But let’s not quibble.

I like setting our handy lists for rainy days, and here is a list Carrier takes in part from Mark Strauss of the various Jesuses that have been argued for recently:

  • Jesus the Jewish Cynic Sage
  • Jesus the Rabbinical Holy Man
    • or Devoted Pharisee
    • or Heretical Essene, etc etc etc
  • Jesus the Political Revolutionary
    • or Zealot Activist
  • Jesus the Apocalyptic Prophet
  • Jesus the Messianic Pretender
    • or Actual Messiah
  • Jesus the Folk Wizard
  • Jesus the Mystic
  • Jesus the Nonviolent Social Reformer
  • Jesus the Actual Davidic Heir and Founder of a Royal Dynasty
  • And more . . . .

Obviously something is fundamentally wrong with the methods of the entire community. . . Indeed, some critics argue the methods now employed in the field succeed no better than divination by Tarot Card reading . . . . (p. 14)

Richard Carrier calls for an end to methods that lead to such chaotic results.

Historians must work together to develop a method that, when applied to the same facts, always gives the same result; a result all historians can agree must be correct (which is to say, the most probable result, as no one imagines certainty is possible, especially in ancient history). (p. 14)

I am not entirely sure I fully understand this. What “facts” are we referring to here? I have not thought of criteria as being “applied to facts” but only to “data” (or “testimonies” or “accounts” or “narratives”) in order to find or conclude what might be a probable fact. My understanding is that historians do have much data and many “facts” (e.g. persons and events well established by means of primary and secondary evidence, independent confirmations of sources that are analysed as being reliable, etc) — even in ancient history — with which to work and seek to understand.

I think Carrier is saying that historians must find an agreed upon method that will lead them to agree on the “facts” they “find” (by means of criteria) about Jesus himself. Indeed, he appears to confirm this in his conclusion to this chapter when he writes

In this book I will present a new method that solves the problems attending the ‘method of criteria’ so progress can finally be made in the field of Jesus studies. (p. 15)

Fair enough. I guess. My reservations about the possibility of doing any history at all “about Jesus” are yet to be laid aside. Carrier says his method will have applicability beyond Jesus studies. I don’t see that yet. But I promise to give his argument the fair hearing he has asked for.

I personally think historians already have and use valid methods, and we have seen these applied to Old Testament studies by historians such as Thompson and Lemche and others. Why not apply the same methods to the question of Christian origins? Analyse documents for what they are (that is, don’t assume the historical origins of the narratives a priori), seek to explain them by means of external or independent attestations. If such a method leads us to explain Christian origins quite apart from any necessity for an historical Jesus then so be it.

What is missing in Carrier’s early argument is an explanation of what he means exactly by “history”, by “historical method”, by “facts”, by “data”, by “proving history” to help one like me orient my thinking and lay aside some of the uncertainties I have discussed here as I begin to read his book.

But I have seen some good points in other briefer discussions by Carrier on Bayes’ theorem. These include the need for historians to face up to the fact that a narrative about an event is not itself evidence for the event, but only evidence for the existence of a narrative. That’s good. So I’m not closing the door on the usefulness of what Carrier is offering. Let’s see where he can take me with his fuller discussion in the coming chapters.

Enhanced by Zemanta

  • Will A
    2012-04-10 05:34:39 UTC - 05:34 | Permalink

    I like Richard’s methods a lot although I think he sometimes overstates the strength of his case. He uses terms like “tour de force” to describe his own work, which sometimes comes across as a bit pompous. “Proving History” does sound a bit too strong a title, but then I think we can be sure that Richard will deal very sensitively with probabilities and possibilities here – that’s the whole point of his use of Bayes.

    He also does plenty of interpretation as well as explanation: I think what he tends to do in his recent writing is state the evidence (e.g.. what texts simply say) and then deal with the interpretations probabilistically (as he has done recently with the “James brother of the Lord” detail on his blog): i.e. what does a word most probably signify? So he is not ignoring the interpretative side, but I think integrating it into its proper place as a probabilistic theory about the meaning of the evidence.

    I don’t think he’s using Bayes as a new criteriology to make weak evidence speak volumes. If the evidence is weak, if interpretations of that evidence are shaky, that will be factored into the equation and his conclusions will be correspondingly dubious. It’s a great way I think to demonstrate honesty in interpretation: Bayes forces you to say why you think an interpretation solid.

    As for the evidence, I think you are being too pessimistic about how much relevant evidence there is. We have quite a lot of fairly early texts to go on, which bear on early Xianity and Xian origins. Richard will presumably investigate in the second volume the most expected consequences we should see in the early evidence, if HJ is valid: the paucity of HJ in the early texts is very good “silence”-type evidence that he will probably use.

    I think part of the problem “going forward” will be that a lot of the historical research that needs to be done is still waiting to be done, because it cannot be conceived until mythicism gains respect as a theoretical paradigm: e.g. work on the reception of the Gospels and Gospel-based traditions by Xians who received them for the first time but previously had little or no HJ conception.

    I think HJ will come to be seen as so unlikely, given the scriptural, allegorical and didactic sources, and its absence from so many early texts so unlikely if HJ is true, that HJ will fade out from the realm of probability. You are right it would be very difficult to “Prove” HJ decisively wrong, but it could be marginalised by the combined strength of other factors.

  • Will A
    2012-04-10 05:42:30 UTC - 05:42 | Permalink

    To continue on an interesting topic… 🙂

    I suspect by “facts” Richard might alternatively mean “evidence”. It’s whatever gets plugged in to the front of the Bayesometer: the texts that come out of pots in the ground, the inscriptions engraved on the pedestals, etc..

    Richard’s main point seems to be that good historical methods that other historians are using all actually boil down to Bayes, so yes he accepts other historians’ effective methods. You just need to understand their worth in a quantitative as well as a logical way: e.g. a criterion may favour the validity of some theory, but by how much?

    I think he also wrote there about how you can’t apply criteria anyway until you’ve got things like genre right. That is probably the key point to be made to people who apply criteria to the Gospels, and which makes them totally invalid as currently used.

    Give him a chance, I think it makes good sense.

  • 2012-04-10 13:24:15 UTC - 13:24 | Permalink

    But I have seen some good points in other briefer discussions by Carrier on Bayes’ theorem. These include the need for historians to face up to the fact that a narrative about an event is not itself evidence for the event, but only evidence for the existence of a narrative.

    While I agree with this sentence, I think it’s a bit too overstated. The narrative could be evidence for an event, but as you say, it needs external corroboration. Using Bayes’ theorem would make that observation more explicit. I think Carrier’s overarching point with this book is not so much mathematical precision, but more about thinking like a Bayesian and arguing that all valid reasoning reduces to Bayes’ theorem even if conclusions aren’t accurate to the nth decimal place.

    For Bayes’, “evidence” is any fact or event that increases the probability of some hypothesis. A narrative about an event might only increase the probability of some hypothesis by only .1%. While it would be “evidence” in the strict Bayesian sense, it is very, very weak evidence. If the presence of a narrative only increases the probability of some hypothesis by .1%, then we would need more evidence — beyond just the narrative (i.e. external corroboration) — that helps to increase the probability of the hypothesis even more.

    It could also be that a narrative for an event is evidence against some hypothesis; this would be any fact or event that decreases the probability of some hypothesis (the standard example of this in probability theory is the case of Oliver’s Blood).

    This is also the reason why, contrary to popular belief, absence of evidence could be evidence of absence.

    Carrier hasn’t gone over this reasoning so far in the book (I’m up to chapter 4) so I’m not sure if he goes over it yet.

  • 2012-04-10 13:55:51 UTC - 13:55 | Permalink

    I don’t understand how the mere existence of a narrative about an event can ever increase probability for a hypothesis at all, not even by 0.1%. A mere narrative of and by itself is just that. We need other information before we can think it might be historical or not. Why not simply put some data in the “don’t know” basket? Is it valid to give everything a weight for or against?

    But I admit I am still in the early stages of my reading.

  • Steven
    2012-04-10 20:19:32 UTC - 20:19 | Permalink

    Having finished the book my basic problem with Carrier’s argument is at the end of the day it is still the individual who is providing the probabilities. Does it surprise anyone that some, for whatever reason, think it is more likely that Jesus was a Historical person. I just don’t see how this is going to make historical research into Jesus and early Christianity, or into any non-modern event more uniform in their conclusions. Rather, I think it will mean Historians will attempt to prop up their conclusion with made up probabilities. I really don’t see how this is going to help. As an alternitive to this I would suggest Historians need to remember what sources are and what they can and can’t tell us. If all we have is secondary sources without any colaboration from primary sources then we, like the Old Testament Minimalist, can’t say there definately was a “true”/”provible” historical event behind the narratives and leave it at that.

  • mP
    2012-04-10 23:03:54 UTC - 23:03 | Permalink

    From my own readings of Barts work, and what he has said in the vids i have seen on YouTube, sometimes i get the feeling that he knows the miraculous side of xianity is not true, but he still loves the warm and fuzzy Jesus like all xians do. By still believing in the man and not the god, he pays homage and perhaps is not the godless barstard most xians would think of scholars who think otherwise. From what i have read of his latest book, he seems to be really saying its not unreasonable to ebleive the man existed. I think he is muddying the waters by acknowledging that the man is enuff to justify the name. By saying the name or man exists then he in effect gives life to the god man, which is dishonest in that is a totally different matter. The mere man should not be called Jesus as it is, because anyone who quotes Bart will always say see Jesus the Son of God existed. Its a simple as that.

  • 2012-04-11 06:26:18 UTC - 06:26 | Permalink

    Looks like an interesting read. But it’s always important to remember that even one’s own process and set of criteria for “uncovering” a historical figure are always influenced at some level by one’s preconceived ideas of that historical figure, even amongst historians with the best of intentions.

  • robertb
    2012-04-11 18:26:38 UTC - 18:26 | Permalink

    That Bayes forces one to illuminate their own assumptions and preconceptions seems like a very good thing to me. Yes, you can load the theorem to reach the conclusion you wish, but spectators will clearly be able to see your ass when you do so. At least, it might help to keep the arguments honest.

  • Leave a Reply

    Your email address will not be published. Required fields are marked *