In the previous post I spoke of the historian’s absolute confidence — of their certainty, of no room for doubt — in the basic events of the past. I don’t know how anyone can seriously think there might be even the slightest room for doubt that Japan bombed Pearl Harbor in December 1941 and overran Singapore shortly afterwards, for example. But historians are not interested in simply documenting past events. That’s not chiefly what they do. Not most of the time. Or certainly not all that they do.
There are many ways to write history but I will be speaking about the approach well known to us all — the narrative or story approach to describing past events.
It is not the facticity of the events that is in question
It is at this narrative level where problems and disagreements, doubts and uncertainties, among historians arise.
Consider the difference in the following statements about event of Japan bombing Pearl Harbor:
- In a “Day of Infamy” Japan launched a surprise attack on Pearl Harbor on December 7, 1941, dragging the U.S. into war.
- The US suddenly cut off nearly all of Japan’s oil supply, thus compelling Japan to take the oil fields in Borneo; but first it needed to make a preventive strike on the fleet at Pearl Harbor.
In both statements the raw events are the same. There is no dispute about any of the specific events mentioned. The “facts” themselves are certain. No probability analysis is required to determine how “probably true” any of the details are. But we have two very different stories, two very different histories, all because of the way the events have been selected from the masses of other details that could have been added to both stories, and because of the way those selected events are woven together and the innuendo in which they are embedded.
That’s where the debates of historians centre. That’s where historians, for all the pains they endure sifting through the masses of data, selecting particular items they think to be most relevant, and trying to weave them into a story that they hope will be worthwhile for others to read — that’s where the real historical work happens. It is not about assessing the probabilities that this or that event actually happened. The groundwork of getting the clearly established “facts” or events is a given. (I am speaking generally, not about those special occasions where new documents are discovered and in need of verification or where some isolated point is in dispute.)
Witness the History Wars
Last time I referred to History Wars. When one side launches a salvo about, say, the good intentions of most of the pioneers and cites specific incidents to argue that very few indigenous people were mistreated directly or deliberately by white settlers, the historians on the other side might grant some of those points but pull out more newspaper stories and police records and archival material to supply the factual evidence that they expect to win the other side of the debate. The debates are about the meaning and interpretations of those recorded events. They may supply evidence to demonstrate that some of the details are exaggerated or down-played. But very rarely, as far as I am aware, do historians spend time trying to assess the probability that X or Y happened in the first place.
Undeniability of certain events
Richard Carrier does acknowledge that some of our knowledge is undeniable and not subject to any shadow of room for any doubt:
The only exception would be immediate experiences that at their most basic level are undeniable (e.g., that you see words in front of you at this very moment, or that “Caesar was immortal and Brutus killed him” is logically impossible). . . . Therefore, because we only have finite knowledge and are not infallible, apart from obviously undeniable things, some probability always remains that we are mistaken or misinformed or misled.
and
. . . . apart from the undeniables of immediate experience, all facts are theoretical . . .
Of course “historical facts” do include direct uninterpreted experience . . . .
(Carrier 25, 298, 302. My comment: Few historians would say that even direct experience is ever “uninterpreted” or that all knowledge that does not come to us from direct experience is necessarily “theoretical”.)
The war memorials, the war cemeteries, the plaques with honour rolls of the dead in countless school and club halls around the nation, the photographs, the memorabilia passed down through generations, — all of these and more tell us that the twentieth century world wars were not at any level (not even at an infinitesimally low level) “theoretical”. Those wars are not known by “immediate experience” to most people today. But those wars are “facts of history” that are undeniable. (Again, I am speaking generally. Of course we may discuss historical events as theoretical events for other reasons and in other contexts, but I am addressing more fundamental bread and butter issues here.)
When we examine why those wars are undeniable, we find certain kinds of evidence that gives us certainty. The same applies to other events in other times. The difference will be that the further back we travel, generally speaking, the more scarce various types of evidence become. But historians still look for the same kinds of evidence about the remote past as they find for more recent events. Naturally questions of authenticity arise for different types of sources. But that even applies in modern times. One famous historian who specialized in the study of Hitler, Hugh Trevor-Roper, was initially deceived by the discovery of the Hitler Diaries that turned out to be forgeries. Historians are well aware of the possibility of fraud and the difference between fact and fiction when examining different kinds of evidence. But that doesn’t reduce all their knowledge of “what events happened” to a “theoretical” status along with some sense that they think they could be “possibly, even if only very very slightly, mistaken”).
So when Richard Carrier writes . . .
Most of what we can say, especially about ancient history, is “maybe” or “probably”—not “definitely.” There is obviously more than one degree of certainty. Some things we are more sure of than others, and some things we are only barely sure of at all. Hence, especially in history, and even more so in ancient history, confidence must often be measured in relative degrees of certainty, and not in black-and-white terms of only “true” and “false.”
(Carrier 23)
. . . I might be wrong, but I suspect that not even Richard Carrier entertains for a moment even the slightest theoretical possibility that there was no Roman empire in existence two thousand years ago.
Not even postmodernists view historical events as “theoretically probable”
One might expect theoretical doubts about “facts” of specific events in the past among postmodernist historians. But no, not even postmodernist historians go as far as Carrier does in the above quotations. In the words of a Professor of Religious and Cultural History at the University of Dundee, Callum Brown,
Any postmodernist historian is not being a postmodernist all of the time. Like every historian, the postmodernist must conduct empirical research, establishing that events occurred and the order of them, checking sources that verify the facts of the case, and making decisions of judgement (balance of probabilities may be the best term) where absolute certainty is not possible. . . .
Historians are probably the least likely academics to preface their books with theoretical explanation.
(Brown, 10f. My comment: note the role of probability applies to exceptional cases.)
Here is how Brown presents “a good historian”:
To be a good historian, it is thought you have to be good in empiricist method, and be seen to have a full grasp of facts. This involves the application of scholarship skills to a series of questions. These occur on different levels. On the upper level are the big questions of: What happened, when did it happen, and why did it happen? At the second level of scholarship, the historian answers these questions by asking: What is the existing state of historical knowledge? And what hypotheses best fit the known facts? At the third level, the historian tests the existing state of knowledge by locating new documents and other sources, or re -evaluating already known ones, checking their date and place of origin, their authorship, their destiny and circulation, and how these discoveries alter the existing state of understanding. Next and last, the historian writes a report or a narrative of the issue, replete with edited evidence and how to interpret it, properly sourced with footnotes, and publishes this in book or article form to be checked by peer review by other historians. If after being read by other historians the published account alters in some degree the existing state of knowledge, it acquires a degree of acceptance that other scholars then come along to challenge and re-assess, in turn to repeat the process of investigation in an endless cycle of moving knowledge forward.
This method of doing History is broadly what all academic and professional historians aspire to the world over.
(Brown 21f)
A time to be certain, a time to doubt
There is a difference between establishing facts beyond doubt on the one hand and interpreting those facts and weaving them into a bigger narrative on the other. So there is a place for doubt and debate among historians but it is rarely over whether or not a particular event at some level actually happened. Again, keeping with postmodernists (persons many would assume to “doubt everything, even facts”), we see that even they hew to “getting the basic facts right”, leaving no room for doubt in that area:
The postmodernist critic distinguishes three different aspects of empiricism. These are empiricism as an event, empiricism as a method, and empiricism as a philosophy of knowledge. To each of these, the postmodernist has different attitudes.
Empiricism as an event is the Enlightenment. The Enlightenment is an event in the History of ideas within which empiricist method and empiricism as a philosophy of knowledge originated. . . .
Empiricism as a method is the second aspect distinguished by the post modernist. This is the method by which empiricism defines knowledge. Empiricism argues that knowledge is acquired through an apparatus of human observation, experience, testing of authenticity, verification, corroboration and presentation for judgement (or peer review) by others in a value-free form. Even if the consequences of empiricism are challenged, postmodernists most certainly do not reject empiricist methods. Like all historians, the postmodernist needs empiricist method for the essential skills, and any student of History must learn and deploy them.
The postmodernist distinguishes a third aspect of empiricism, however – empiricism as a philosophy of knowledge. And this is seen as being full of problems. In the work of many academics across science and non science disciplines, there is an implicit notion that empiricism constitutes all that is necessary to knowledge – that it is a complete system of knowledge with no other connections. This notion is that human knowledge acquisition is nothing more than empiricism, and needs nothing more than this for the advancement of each discipline. In the case of History, the writing of the past has been seen by some empiricists as being satisfactorily embraced by empiricist method.
(Brown 21-25)
Historians who go beyond “getting the facts right beyond doubt” and view all their historical work as “getting even the narratives right” are not in fashion today:
One such empiricist historian was Geoffrey Elton, a leading right-wing historian, who regarded empiricism as the only worthwhile basis of professional training in the History discipline. . . . His purist empiricist position brought him to dispute with other historians over decades – including non-postmodernists.
(Brown 25)
Callum Brown discusses the difference between undoubted (we can say “undoubtable”) events and the way historians put them together to tell a story. What we are calling an “historical event” (I prefer the term “event” to “fact”) in this post (e.g. the Japanese bombed Pearl Harbor) is “something that happened . . . the event occurred . . .” Where historians differ is how they put those events together to each tell their own distinctive account.
Once an event of the past is described, it becomes something else — it becomes a narrative. . . . There is never any neutrality in a story.
(Brown 28f)
Brown concludes his discussion by leaving the reader in no doubt about the absolute certainty of the past events investigated by the historian:
Empiricism is the basic method in all scholarship. It bears endless repetition that the empiricist skills of verification, close textual attention, proper and rational sourcing, referencing and so on, remain absolutely central to all that historical scholarship does, whether postmodernist or not. In this regard, the Enlightenment created the method of the modern historian.
But empirical method is one thing. The other is the empiricist philosophy of knowledge, or modernism, and that most certainly is challenged . Empiricism gives the illusion of delivering fact, truth and reality, by slipping from the event to a human narrative that describes the event.
(Brown 30)
But does not certainty breed arrogance?
But doesn’t certainty breed arrogance? Is not there an admirable and necessary humility in doubt? Yes, but no one can be arrogant by claiming to know the world is round. The kind of certainty that engenders arrogance is the certainty of opinion and moral perspective — of conviction of holding “The Truth”. Brown calls upon the words of Friedrich Nietzsche to make the point that doubt belongs in the way we tell stories, in the ways we interpret and understand the events of the past. It is not about the two world wars of the twentieth century or the ancient Roman empire to some theoretical status of which we can only be 99.99% certain.
What, then, is truth? A mobile army of metaphors, metonymies, anthropomorphisms, in short a sum of human relations which have been subjected to poetic and rhetorical intensification, translation, and decoration, and which, after they have been in use for a long time, strike a people as firmly established, canonical, and binding: truths are illusions of which we have forgotten that they are illusions, metaphors which have become worn by frequent use and have lost all sensuous vigour, coins which, having lost their stamp, are now regarded as metal and no longer as coins. (Nietzsche, 1873)
Certainty emerges as a linguistic construction which, Nietzsche went on, ‘prompts a moral impulse’ to doubt the denier of certainty, to see in doubt the very basis of immorality. This is a massive irony. Nietzsche says doubt itself becomes superior to fact and moral certainty. This seems like craziness. It seems to be completely absurd and inverted in logic. It over turns everything we are trained to believe as students at school and in college. It may be difficult to grasp that doubt is superior to certainty. This is why postmodernism is truly revolutionary as a philosophical system of thought.
(Brown 30f)
Was the gingerbread vendor really kicked to death?
There are occasions, however, when historians do actually disagree over the factual status of an asserted event of the past. The historian Edward Carr once argued that a past event known to have happened only becomes a “historical event” when it is used by historians in their narratives.
At Stalybridge Wakes in 1850, a vendor of gingerbread, as the result of some petty dispute, was deliberately kicked to death by an angry mob. Is this a fact of history? A year ago I should unhesitatingly have said ‘no’. It was recorded by an eye-witness in some little known memoirs; but I had never seen it judged worthy of mention by any historian. A year ago Dr Kitson Clark cited it in his Ford lectures in Oxford. Does this make it into a historical fact? Not, I think, yet. Its present status, I suggest, is that it has been proposed for membership of the select club of historical facts. It now awaits a seconder and sponsors. It may be that in the course of the next few years we shall see this fact appearing first in footnotes, then in the text, of articles and books about nineteenth-century England, and that in twenty or thirty years’ time it may be a well-established historical fact. Alternatively, nobody may take it up, in which case it will relapse into the limbo of unhistorical facts about the past from which Dr Kitson Clark has gallantly attempted to rescue it. What will decide which of these two things will happen? It will depend, I think, on whether the thesis or interpretation in support of which Dr Kitson Clark cited this incident is accepted by other historians as valid and significant. Its status as a historical fact will turn on a question of interpretation. This element of interpretation enters into every fact of history.
(Carr 12)
Notice that Carr understood the event to have been factual because it was found in an eye-witness’s memoirs. He was discussing what he understood to be a real event, not a theoretical one. But another historian checked the source and raised doubts:
The likelihood of the gingerbread salesman’s unfortunate death being a historical fact in this sense is moderately but not overwhelmingly high because the reference Kitson Clark used for it was not a contemporary one, but a set of memoirs written long after the event, and memoirs are sometimes unreliable even where they are giving eyewitness accounts of happenings in the past. If I had been Kitson Clark, I should have looked for a contemporary document to verify my claim. It is for this reason, I think, not because it has not been widely quoted elsewhere (except in discussions of Carr’s What Is History?) that the status as a historical fact of the gingerbread sales man’s murder in 1850 must be regarded as still provisional, to say the least.
(Evans 66f)
Richard Evans is far from being a postmodernist historian. Note the kind of source he prefers to use to help establish the historicity of a person or event. It is a contemporary one. Not even a late memoir of recording personal reminiscences is considered as secure as a contemporary reference to the event. Let those who rely upon Josephus as a slam-dunk verification of the existence of Jesus take note of how historical research is undertaken in “non-biblical departments”.
Where probability lurks in history
Notice also that Evans did acknowledge a role probability (and its attendant humility) in historical studies:
No historians really believe in the absolute truth of what they are writing, simply in its probable truth, which they have done their utmost to establish by following the usual rules of evidence.
But that sentence should not be ripped from its context. Evans was specifically addressing the narratives historians write, the way they interpret the established events. Here is the context:
In similar vein to David Harlan, Ellen Somekawa and Elizabeth Smith argue that because “within whatever rules historians can articulate, all interpretations are equally valid,” it is necessary for historians to “shift the grounds for the assessment of integrity from the absolute or objective truth to the moral or political. That is,” they continue, “rather than believe in the absolute truth of what we are writing, we must believe in the moral or political position we are taking with it.” They add that they “reject the assumption that if we abandon our claim to objective truth we must be writing in bad faith (writing propaganda in the most pejorative sense of the word),” but they offer no reason to suppose why this should not be the case. In fact, of course, in classic postmodernist fashion they are caricaturing the position they are attacking by pushing it out to an extreme. No historians really believe in the absolute truth of what they are writing, simply in its probable truth, which they have done their utmost to establish by following the usual rules of evidence. In the end it simply isn’t true that two historical arguments which contradict each other are equally valid, that there is no means of deciding between them as history because they are necessarily based on different political and historical philosophies.
(Evans 188f)
Evans is disagreeing with claims that interpretations or understandings of historical events are all equally valid and the accusation that historians necessarily believe that those interpretations are “absolute truth”. Note further that Evans began his book with the following quotation, even comparing the fact-finding methods of historians to the work of astronomers, a comparison used by Carrier though with a quite different perspective (Proving History, p. 105). In what follows I cannot detect any sense that the “raw events” of the past are to be understood as theoretical knowledge with a minimum of some room for doubt about their historicity. Quite the opposite, in fact.
However much they might have agreed on the need for accuracy and truthfulness, historians down the ages have held widely differing views on the purposes to which these things were to be put and the way in which the facts they presented were to be explained. . . .
. . . Ranke introduced into the study of modern history the methods that had recently been developed by philologists in the study of ancient and medieval literature to determine whether a text, say, of a Shakespeare play or of a medieval legend like the Nibelungenlied was true or corrupted by later interpolations, whether it was written by the author it was supposed to have been written by, and which of the available versions was the most reliable. Historians, argued Ranke, had to root out forgeries and falsifications from the record.They had to test documents on the basis of their internal consistency and their consistency with other documents originating at the same period. They had to stick to “primary sources,” eyewitness reports and what Ranke called the “purest, most immediate documents” which could be shown to have originated at the time under investigation, and avoid reliance on “secondary sources,” such as memoirs or later histories generated after the event. . . .
Ranke’s principles still form the basis for much historical research and teaching today. . . .
Whatever the means they use, historians still have to engage in the basic Rankean spadework of investigating the provenance of documents, of inquiring about the motives of those who wrote them, the circumstances in which they were written, and the ways in which they relate to other documents on the same subject. The perils which await them should they fail to do this are only too obvious. All these things have belonged to the basic training of historians since the nineteenth century, and rightly so. . . .
Skeptics who point to the fact that all sources are “biased” and conclude from this that historians are bound to be misled by them are as wide of the mark as politicians who imagine that future historians will take their memoirs on trust. Nor is there anything unusual in the fact that a modern discipline places such heavy reliance on principles developed more than a century and a half before: Chemistry, for example, still uses the periodic table of elements, while medical research continues to employ the mid-nineteenth century device of “Koch’s postulates” to prove that a microorganism is the carrier of a particular disease. These analogies with scientific method point up the fact that when source criticism was introduced into historical study, it, too, was regarded as a “scientific” technique. Its use legitimated history as an independent profession. . . .
The understanding of science which these claims implied was basically inductive. Out there, in the documents, lay the facts, waiting to be discovered by historians, just as the stars shone out there in the heavens, waiting to be discovered by astronomers; all historians had to do was apply the proper scientific method, eliminate their own personality from the investigation, and the facts would come to light. The object of research was thus to “fill in the gaps” in knowledge—a rationale that is still given as the basis for the vast majority of Ph.D. theses in history today.
(Evans 13-17)
Evans goes on to point out that even the pioneer of modern history himself, Leopold von Ranke, failed to produce a genuinely “objective history” despite his claims to be attempting to do so. Ranke failed to understand the subjectivity that enters when we seek to understand and use data in a narrative. For that reason his approach to historical knowledge (not his methods), known as historical positivism, has long since been discarded by most historians today. But as Evans reminds us, the spadework required to establish facts as certain remains with historians today. (Here we are entering another misconception I very often find among biblical historians: they all too frequently tend to equate discarded positivism with the methods of positivism!)
History is not the same as science
Contrast Carrier’s comparison of historical research with the historical method:
Geology and paleontology, for instance, are largely occupied with determining the past history of life on earth and of the earth itself, just as cosmology is mainly concerned with the past history of the universe as a whole. . . .
For example, we can document our testimony to seeing highly compressed rock on a mountaintop with extinct seashells embedded within it. But this information is only useful to us if we can infer from such observations (and others like it) that that rock used to be under the sea and thus has moved from where it once was, and that this rock has been under vast pressures over a great duration after those shells were deposited in it. . . . A particular pattern and sequence of layers in a rock formation can even confirm to us specific historical facts, such as exactly when a volcano erupted, a valley flooded, or a meteorite struck the earth thousands of miles away. . . .
History is the same. The historian looks at all the evidence that exists now and asks what could have brought that evidence into existence.
(Carrier, 46f)
I have to disagree with Carrier here. No, history is not the same. The geologist is interpreting the rocks through a knowledge of physical laws and seeing how those laws have acted out in the past to shape our earth today. History is not the same.
Historians study human actions and few historians would agree today that the persons or events they study follow predictable laws. There was once a time when a good number of historians hoped or believed they could find laws at work in historical processes but generally speaking those days are gone. The one example of historical processes following laws that most of us have at least heard about is Marxism: the view that historical events were all manifestations of class struggle. (Another “history follows laws” approach that I was introduced to at high school was Toynbee’s “challenge and response” model of historical events.)
The reason history cannot predict the future is because “it”, or human societies are all different, events are never repeated, they are never the same. They are not governed by the (theoretical) laws of science as is the physical matter of the cosmos. We can predict outcomes only at the most general level and at that level it tells us nothing more than we already know about human behaviour.
In principle the methods of the historian may be the same as the bulk of those of a scientist and for that reason history is sometimes called a science or scientific, but few historians are trying to understand theoretical laws to explain events.
Continuing in the next post…..
Brown, Callum G. Postmodernism for Historians. Routledge, 2005.
Carr, Edward Hallet. What Is History? Vintage, 1967.
Carrier, Richard. Proving History: Bayes’s Theorem and the Quest for the Historical Jesus. Prometheus Books, 2012.
Evans, Richard J. In Defence of History. Norton, 1997.
Neil Godfrey
Latest posts by Neil Godfrey (see all)
- Is Everything a Question of Probability? - 2024-12-15 03:04:03 GMT+0000
- The Folly of Bayesian Probability in “Doing History” - 2024-12-13 05:51:46 GMT+0000
- Jesus Mythicism and Historical Knowledge, Part 4: Did Jesus Exist? - 2024-11-27 08:20:47 GMT+0000
If you enjoyed this post, please consider donating to Vridar. Thanks!
Give the words “Jesus Mythicism” appear in the title of these posts.
I am wondering if I need a definition of what you see as “Jesus Mythicism”.
I see two possible meanings of this.
1./ An intellectual pursuit for arguments striving to prove that Jesus narratives have no historical basis.
2./ An approach to studying Christian origins which down-grades the role of a hypothetical historical founder (Jesus), focusing instead on how the texts and the communities may have come into being through creative myth-development.
If we go with 1./ then mythicism is the goal, to determine one way or the other whether Jesus once existed and mostly win the debate! I sort of imagine that sums up a lot of R Carrier’s approach. It sells books and gets clicks but leaves us a bit empty.
So, with all this ground work on how to do history, which form of mythicism is about to be subjected to the hard interrogation of modern historical reasoning? (or is that asking for spoilers?)
My posts relating Jesus mythicism to historical knowledge and methods were prompted by Richard Carrier’s publications and posts arguing that historical methods, properly done, make the existence of Jesus an unlikely prospect. I have held off from doing these posts for a long time, partly because I do not want to be seen as adding any fuel to an already and often overly heated polemic against Carrier personally. I have long wanted to “set the record straight” re the place of history in the question.
My position is, ironically, the same as that of many biblical scholars. I seek to do no more than ever address the Jesus we have, which is the Jesus of the literary and theological texts. All the efforts of scholars to arrive at a historical Jesus through form criticism, criteria or authenticity, memory theory are all — ALL (no exaggeration) — based on circular reasoning. And that circular reasoning is erroneously justified by many of them by reference to the writings of Hans-Georg Gadamer. I have seen biblical historians misread and/or misapply so many authorities in other fields — Vansina, Halbwachs, Ricoeur, Hobsbawm ….. They can’t even get positivist history right when they claim to reject it. They are in a club of their own but they have a huge audience and I try to do my little bit to leave a record of correction where I can. (I have addressed this question many times, but for the benefit of anyone not familiar with the history of blog posts here, even an ancient figure like Alexander the Great is known in large part from contemporary sources since we know those sources were covered and addressed by centuries later authors whose writings we do have.)
For me, the question of the historicity of Jesus does not even arise. There is simply no contemporary evidence for the figure. The sources we have are all late and often of questionable authenticity. The canonical texts are all found in the independent record no earlier than the second century. They may be earlier, but it hardly matters since they meet none of the characteristics of sources historians deem appropriate for determining the historicity of any ancient person.
The only Jesus we have is the Jesus of theology and literature. He is the only Jesus that matters. It is that Jesus that “changed history”.
To respond directly to your questions:
1. I think it is a waste of time trying to prove Jesus has no historical existence. But I am speaking from the perspective of having in one way or another making contact with this discussion for twenty years now. I suppose it does not hurt to show why the gospels and Pauline epistles and early Roman and Jewish references do not prove the existence of Jesus. Many people assume they do so it is fair to point out why they don’t.
2. I don’t even “downgrade” Jesus — I simply do not have any historical Jesus in my sights when I study Christian origins because there is simply no evidence for such a figure. Christian origins has to grapple with the evidence we have — not with imagined and presumed evidence “behind” the sources we have. James McGrath wrote a little booklet (self-published, I should add, since he loves to note when others self-publish) in which he wrote pages explaining how “biblical historians” get behind or beneath the texts to find Jesus. No historians in other fields that I am aware of apply such a method unless there are independent grounds for justifying the process. Again, biblical studies lives in a world of its own.
The question of Christian origins is probably unanswerable in any detail simply because of the state of the evidence. But to the extent we have some evidence, we at best need to ask about the origins of the gospels and the epistles and other early para-Christian and non-canonical writings like the Odes of Solomon and Ascension of Isaiah and Didache — and especially investigate what we can of the likes of the earliest “church fathers”. That’s where we are going to get closest to trying to understand Christian origins (and even simply “what Christianity was” at that time). I am sure it must have something to do with the wars and destructions from 66 through to 135 CE and the place of gentile “god-fearers”.
Those are the areas where anyone interested in Christian origins should begin to look — in my opinion.
I don’t see myself as a “Jesus mythicist” because the label implies I am trying to prove Jesus is a myth. I am not doing that. Others do that and that’s fine. They don’t need me to pitch in. As I said, I take the position of many biblical scholars and work with the Jesus we have, the literary and theological figure — the only figure of any historical relevance in my view. My interest is in Christian origins and my methods are those of mainstream historians.
An interesting point…
But how would you see Acts 1 as flying in the face of a historical Jesus if it’s the allegorical construct God/Angel model?
Jesus flits off to heaven. Where then are his bones and DNA? This is integral in a way to the Talpiot tomb debate elsewhere…
But that’s just it. The writer of Acts has Jesus literally take the EVIDENCE of Jesus’ existence WITH Him. There is, therefore, NONE on the planet.
Would this in real terms be the “intangible” paradigm ANYWAY?
Tell me what you think of this…
Kitos as the BACKLASH to forty-five years of alternative narratives to Torah being ascendant and “the Good”/ChrEstos/Chrestianoi”…and the ROMAN imposition of a “new law” directly after 70 c.e.?
Kitos never quite made sense for the INTENSITY until I worked that out.
Is it possible to EXPLORE what happened to such alternative narratives as the second century progressed…? From their high point 70-115, though to DISTRUST from Hadrian afterwards (Letter to Servanius AND Passing of Peregrinus Proteus presenting the Peregrinus AS a Frumentarii) down to Alexandrian revolts in the Marcus Aurelius Period…through to being HERETICS in the Commodan era?
Would that be more realistic than believing Church narratives about those times?
Your Acts question has no interest for me — besides not really being germane to the specific points of the post or my comments. Acts is a late propaganda fiction and that’s all that needs to be said. As for the Jewish rebellions in the early second century, again, that is a discussion for another place. I hinted at them in my earlier comment but only as a subject to be discussed another time and another place.
(I had to delete your other comment because it was simply too far away from the rules for comments I have tried to make clear in the past … see the side bar and look in the list under About Vridar for the link to “comments and moderation”.)
Hi Neil,
I have to break my silence here with a (probably polemical) question. You write:
I don’t even “downgrade” Jesus — I simply do not have any historical Jesus in my sights when I study Christian origins because there is simply no evidence for such a figure.
Are you saying that a historian has not “even the right” to raise the question: is this figure the historical Jesus? I ask since I have seen Muhammad’s scholars to raise the question if the figure described in this document is the historical Muhammad (and accordingly to answer with a “yes” or a “not”, or a “I don’t know”). While it seems from the your post that you are going to deny to yourself a level of detail in the inquiry that prevent you from “even the right” of raising similar questions.
That’s ridiculous, Giuseppe. What do “rights” have to do with anything? Of course the question can be asked. That’s what research is about. How you can read into anything I have written here such a sinister notion is simply beyond me. Yes, your polemic against me appears not to have waned after all this time.
The historian studies the sources and what is in them. I see no hint in the sources that the Jesus in our sources is anything other than a literary and theological figure. If another researcher sees differently of course they can make their case. But I have not seen one yet.
The link you point to, about the Samaritan prophet being the Jesus of history, is simply one more instance of circular reasoning. It begins with the same broad methodological assumptions and approach of mainstream biblical scholarship and tries to go a little further. It begins with the assumption that there was a historical Jesus and that somewhere in the records he might be found. But that is entirely supposition without any basis in the evidence to begin with. It comes from the general belief in our culture that there was a historical Jesus. This assumption is read into the sources.
Note — the person reads their assumptions of the historical Jesus into the sources. This is called begging the question. It is the same method I have tried to explain is at the root of some of your arguments, too. It is a fallacious method. It is circular reasoning. It is not valid reasoning. The result is, one might say, a kind of invalid parallelomania (of the real kind spoken of by Sandmel).
Of course you and others have the “right” to go that path. People have the right to follow their ignorance and fallacious reasoning as much as they like.
Hence your answer is that the Samaritan false prophet is not the historical Jesus. If so, then you betray a knowledge of the concept of a ‘historical Jesus’. That is the my point.
In addition, I can well imagine what are the “laws” for Richard Carrier: the immutability of the human psychology. That is decisive in his comparison between Pliny the Younger and Paul: both are weeping a deceased figure but only the former is doing what he was expected to do in a such circumstance (to remember details of the his concrete existence on earth). Paul didn’t. Hence Carrier has a point.
Therefore I agree with Carrier on that decisive point, against you. But a better and more honest criticism by you would be to point out that Paul is himself a Marcionite construct therefore the Argument from Psychlogy is not applicable here, contra Carrier. (At the end, all has already been said by Turmel against Dujardin).
You are being absurd, Giuseppe. Of course I have knowledge of “the concept of a historical Jesus”. You have refused point blank with every effort of mine in the past to point out simple basic logical flaws in your methods but you have deflected every one of those efforts of mine without comment and continue to dig further into your pit of confirmation bias. This will be the last comment from you here. I will see no more of your comments because they will be directed straight to spam. The main reason? Your evident hostility towards me personally reeks through your comments. The other reason? You refuse to engage with any understanding with anything I write and seek to twist it to your own ends.
“For me, the question of the historicity of Jesus does not even arise. There is simply no contemporary evidence for the figure.”
I would imagine that many people antagonistic towards mythicism would see that statement as ‘mythicism’ in that you are denying what to them is the ‘evidence’ that Jesus existed. (Needless to say, this is not my contention)
Certainly I agree wholeheartedly with you about the evidence, but I am not sure about the whether that means the question of historicity won’t naturally arise. In studying the literary ‘Jesus’, we would naturally ask, does this ‘literary’ Jesus arise from mythologising of a historical person. We might come to a conclusion that these myths can easily arise without having a historical core, or we might make a case the other way, However, we would be limited by the vague nature of the evidence we are dealing with.
We might look at those debating these issues and think this is not a fray worth getting involved in, as we can make plenty of deductions without engaging with this question. This might cover some of what I mean by ‘downgrading’, as the existence or non-existence of this figure can be bypassed. Important conclusions can be reached without taking sides here.
Then we could also take a minimalist mythicism position that states: If there was a historical Jesus, he is so obscured by mythology that for all intensive purposes his existence is not particularly important to understanding the beginnings of Christianity any more than trying to identify the very butterfly that flapped its wings that led to a cyclone forming out at sea. (With apologies to mathematicians and meteorologists who study chaos theory).
I think we are in for a fun ride, if we can avoid the usual pitfalls of this loaded issue 🙂
As for R Carrier, I think I may have presented a slightly negative assessment above, and possibly over-simplified his motives. I certainly find his articles very interesting and have a high regard for his analytical skills.
I think the question only “naturally arises” if we start from the presumption that Christianity and the social movements at its core began with Jesus. Which is partly why I find the entire focus on his historicity (and historical figure) to be non-historical. It is basically just great man theory on both sides. They both regard the origin of Christianity as coming down to either a mythical figure, or a historical one, and present his (non)existence as being the focal point of origins.
I take a much broader, and more sociological approach to the issue. Movements come about because of the social and material conditions of their day which they are situated within. There are for sure charismatic organizers of the movements, but they are not the most important members, nor are they the ones which give the movement its shape and continuation. Martin Luther King Jr. did not start the Civil Rights movement, nor was he the one who built its infrastructure. His speeches and his charisma were propped up on the movement as a whole. He was, as Albert Kalthoff would describe him, a “coworker” among many.
Similarly, this is how I would treat Jesus, if historical. Nothing special, nothing unique, not worth us dissecting the gospels and other sources to reconstruct. And if he was nonhistorical, same thing. It doesn’t really make a difference to how we study Christian origins either way. The only thing a historical/mythical Jesus explains to historians, is Jesus. It is a tautological exercise of basically no value, to me. Which makes debates on his existence of even less value.
If they claim there is evidence for a historical Jesus then they are working entirely within the assumptions and methods of biblical scholars and not those of mainstream historical research. I know some “nonbiblical” scholars have argued for historical Jesus but they, too, simply jump ship and fall into line with the assumptions and methods of biblical scholars to do so. Hence my focus on historical methods and knowledge.
If anyone says my position is “mythicism” then they must also bracket a good many mainstream biblical scholars who approach the study of Jesus in the same way I do. Many are interested only in the Jesus we find in the letters and gospels (and not one supposedly found “behind” those sources).
As someone very partial to Marxist research, I do want to preface that Marx and Engels did eschew the idea that economics and class struggle were the only drivers of conflict and change in society, e.g., they didn’t hold “the view that historical events were all manifestations of class struggle.” Modern class reductionists do so, but that’s a different issue. For instance, here is Engels to J. Bloch (1890):
“According to the materialist conception of history, the ultimately determining element in history is the production and reproduction of real life. Other than this neither Marx nor I have ever asserted. Hence if somebody twists this into saying that the economic element is the only determining one, he transforms that proposition into a meaningless, abstract, senseless phrase. The economic situation is the basis, but the various elements of the superstructure — political forms of the class struggle and its results, to wit: constitutions established by the victorious class after a successful battle, etc., juridical forms, and even the reflexes of all these actual struggles in the brains of the participants, political, juristic, philosophical theories, religious views and their further development into systems of dogmas — also exercise their influence upon the course of the historical struggles and in many cases preponderate in determining their form.”
Engels and Marx’s comments in the Communist Manifesto (e.g., everything is class struggle, at the beginning) I don’t really take as emblematic of their actual theoretical process, because in terms of genre it isn’t a book of theory, it is a work of rhetoric (a manifesto), and sometimes reductionist comments like that are simply more effective for the political cause.
Anyways, just a brief side-note as someone who is a bit of a Marx reader.
Thanks for the clarification and yes indeed. I am also generally quite partial to Marxist interpretations (though I should not presume you are also since you say no more than that you are “a bit of a reader”) as well as to Gramsci’s fine tuning of the ideological cum political manifestations of the class struggle. Certainly, as you point out, economics is not the only driver in very many cases. The means of production generate the institutions that can take on a life of their own. I posted a little on Gramsci’s analysis in two posts:
https://vridar.org/2019/06/12/understanding-the-rise-of-trump-1/
and
https://vridar.org/2019/06/17/understanding-trumps-rise-presidency-and-beyond-4/
I’ve become actually rather partial particularly to Soviet approaches to history. Zhukov and Chesnokov I’ve become a bit of a “fan” of in recent time.
Gramsci is also great, and I also take a bit from Adorno. Trying to get into some Veitnamese and Maoist theory as of late. My most recent take on the historical Jesus is actually greatly informed by how Soviet Marxists treated the issue in the aftermath of their mythicist debates.
“According to the materialist conception of history, the ultimately determining element in history is the production and reproduction of real life. Other than this neither Marx nor I have ever asserted. Hence if somebody twists this into saying that the economic element is the only determining one, he transforms that proposition into a meaningless, abstract, senseless phrase. The economic situation is the basis, but the various elements of the superstructure — political forms of the class struggle and its results, to wit: constitutions established by the victorious class after a successful battle, etc., juridical forms, and even the reflexes of all these actual struggles in the brains of the participants, political, juristic, philosophical theories, religious views and their further development into systems of dogmas — also exercise their influence upon the course of the historical struggles and in many cases preponderate in determining their form.”
In other words, Marxism is vague enough to say anything, or nothing at all. Brilliant.
Oh dear, I always find it amusing when Americans (they are mostly Americans) jump into false vacuous presumptions of DISinterpretation of anything a Marxist happens to say. Have you ever actually read a Marxist history of anything? Why not pick up a history by E. P. Thompson or by Eric Hobsbawm or Christopher Hill …. you might be surprised.
What useful predictions has Marxism actually made? It seems that all of Marx’s predictions vis a vis the supposedly revolutionary nature of the working class and the inevitability of Communism have not come to fruition, and in the case of “actually existing socialism” (Russia, China) have been completely discredited.
Atta boy — fight the good ol’ fight against them nasty Reds! If you knew the first thing about Marx — actually read his work — you would know very well that Russian and Chinese totalitarianism was NOT Marxism except in the eyes of American propaganda. (Even Russia and China acknowledged they had changed Marxist ideas for their systems.) It is American Cold War propaganda that still has not died to this day that equates totalitarian regimes with actual Marxism.
One of the first things Lenin did after seizing power was crush the Marxist communes that had sprung up in places in the wake of the revolution.
What do you know of Marxism that is not taken straight from Western/US propaganda?
“If you knew the first thing about Marx — actually read his work — you would know very well that Russian and Chinese totalitarianism was NOT Marxism except in the eyes of American propaganda. ”
Do you say that those regime are/were not Marxist insofar as they were totalitarian? This is funny since even in Marx’s own day he was derided by his numerous critics within the socialist movement itself for his deeply authoritarian and statist tendencies. Of course Leninism is a “development” of Marxism (as it must have been since the events leading to and following the Russian Revolution disproved Marx’s historical prognostications), but it carried on Marx’s inherently totalitarian project.
You clearly have not read Marx. American propaganda has shaped your views.
Do you have anything besides anti-American ad hominems? I assure you I have read Marx, probably far more than you.
I’m sorry, but you don’t realize how silly you sound to non-Americans who have found their experiences with a little leftiness much to commend them. It is obvious you are informed entirely by anti-marxist, anti-socialist propaganda. Good bye to you, sir.
For the record, it is worth posting here a quotation I used in Part 3 of this series of posts where Marx’s view of history was genuinely aligned with the concept of true scientific laws:
The Marxist theoretical conception of history can be traced back to Hegel. It is sometimes crudely oversimplified to the notion of “Thesis –> Antithesis; Thesis + Antithesis –> Synthesis”.
The Marxist conception of history is rooted in wishful thinking, working backward from the desired result (a worker’s revolution will inevitably usher in a global egalitarian utopia free of exploitation or societal conflict). That his view of history is based on ‘relations of concepts’ via dialectics is no compliment, since Marx simply assumes that concepts relate “dialectically” in the way he needs for his desired result to be “proven”. Nothing is actually proven this way, though, since he employs concepts in ways that so often contradicts reality.
Neil, this discussion is excellent and draws attention to the issues that historians have been wrestling with for decades. Your argument resembles what the historian Keith Jenkins wrote about in 1991 is his Routledge Press book: Re-thinking History (reprinted in 2003). In a section called “On Facts and Interpretation”, he wrote:
“Are there ‘past things’ that seem to be factually correct? In one sense one can say yes. Thus, we know that the so-called Great War/First World War happened between 1914 and 1918. We know that Margaret Thatcher came into power in 1979. If these are facts then we know facts. However, such facts, though important, are ‘true’ but trite within the larger issues historians consider. For historians are not too concerned about discrete facts (facts as individual facts), for such a concern only touches that part of historical discourse called its chronicle. No, historians have ambitions, wishing to discover not only what happened but how and why and what these things meant and mean. This is the task historians have set for themselves (I mean they did not have to raise the stakes so high). So it is never really a matter of the facts per se but the weight, position, combination, and significance they carry vis-a-vis each other in the construction of explanations that is at issue. This is the inevitable interpretive dimension, the problematic, as historians transform the events of the past into patterns of meaning that any literal representation of them as facts could never produce. For although there may be methods of finding out ‘what happened’ there is no method whatsoever whereby one can definitely say what the ‘facts’ mean. (p. 40)
I cite this lengthy section just to show to other readers of this excellent post that you are working within the broader historiographic discipline that appears to elude biblical ‘historians’. Brown and Evans are very worthy reads and raise important questions for the study of history.
I am looking forward to your further work on this important topic.
Yes, Keith Jenkins does a good job of discussing the methods and philosophies of historians. (I suppose for the sake of the record I should add that I myself am not a “postmodernist” though I acknowledge that “postmodernism” has made at least a somewhat worthwhile contribution to self-awareness of what historians do, and may have contributed much to expanding the scope of areas of history covered by historians — and I don’t share all of Jenkins’ pessimism about history per se.)
You have added a quotation that echoes my quotes from Brown, so good to have your input here. Both Brown and Jenkins fall into the postmodernist camp so one should note that even the postmodernists — and Brown is not an exception — stress the importance of the raw event that is not questioned, but established as “factual”.
And the section you quote is one I wish I had included in my post. As it says, just listing events of themselves is meaningless. They don’t tell a story by themselves. It is the historian who breathes meaning into them by selecting certain events with which to weave a story. That’s when the events become interesting and full of meaning.
When we in the West say, for example, “Japan bombed Pearl Harbor”, we often say that in the sense that it will be understood that Japan was responsible for the US entering the war and we recall the “day of infamy” narrative about the surprise, meaning treacherous, attack. So just repeating the statement “Japan bombed Pearl Harbor” in our current cultural climate can often bring with it all of that meaning. In other words, we are stating more than what Jenkins would call a simple “fact” and are making a moral condemnation of Japan. And I tried to point out that the same event can be spoken about to blame the US for dragging or forcing Japan into a wider war.
So “facts” in that sense are never neutral because they often imply a narrative from which they have been plucked.
The narrative, the morality, the meaning and interpretations of those events can be contested. But beneath all of that contest lies the simple, raw, meaningless, neutral matter of physics, oil supplies, bombs, ships being sunk and damaged…. But even by adding “oil supplies” in that context automatically in the minds of interpreters conveys a meaning somewhat different about the bombing than if I had left it out. Even stated like that, selecting those items and placing them together creates a kind of narrative suggestive of meaning.
Re “. . . I might be wrong, but I suspect that not even Richard Carrier entertains for a moment even the slightest theoretical possibility that there was no Roman empire in existence two thousand years ago.”
I think this is the point. The “fact” that the Roman Empire existed is at a probability of 100% +/-1%. But the “fact,” say, that Christians started the fire in Rome during Nero’s reign is much, much lower.
The evidence for the existence of the Roman Empire is mountainous, about as much as for any historical “construct” And probabilities are to be based upon evidence, hence the lingering debate over Jesus the Christ. The evidence is not mountainous one way or the other.
I don’t believe we can couch our knowledge of the Roman empire in terms of probability (pace Carrier). Probability simply doesn’t enter into it. It is a fact. We have the evidence before our eyes. Carrier did acknowledge that certain facts are indeed undeniable — even at the theoretical (or probability) level, like our experience that we are reading these words right now. We can see the remains of the Roman empire before our eyes so we know it existed. It is as certain as knowing that Hitler came to power in Germany in 1933. There is no “probability” in that statement nor is there with respect to the Roman empire. The sources historians use are “before their eyes” so they can “see” what has happened in the past, if you like.
The probability enters to justify certain hypotheses about the raw events. What were their causes, etc?
Is there really even an margin for doubt of 0.00000000000000001% about the existence of the Roman empire? I simply don’t see how probability arises with such knowledge. It is there. We see the evidence before our eyes. It exists, just like the computer screen before my eyes exists. It does not exist with a probability of anything. It just is.
Dear Neil, I regularly read your very interesting texts (for which I thank you very much), while I react only very, very seldom. But now I feel that I have to say that I do not understand your remarks towards Carrier.
It seems like you criticize something in his approach, but I cannot really formulate what is this what you criticize and what your arguments are. Reading just your texts, I would think that Carrier claims, e.g., that history is the same as science, or that we can reasonably doubt that there was a world war in 1914-1918, or so … But it is obvious that Carrier does not claim anything like this, so your point is really incomprehensible for me …
E.g., when I contemplate your sentence
“A fundamental point on which I disagree with Carrier is the claim that the most a historian can say about any historical event is that it is “probably” true.”
I can not see anything fundamental here. It seems to me that you just refer to the colloquial use of the word “probably” which entails that we are not quite sure, while in Carrier’s context the meaning of the word “probably” also comprises the cases where we are 99,99% sure. So I think that this is just a misunderstanding in the semantics of words, nothing fundamental …
Hi Petr. Yes, Carrier does say history is the same as science in principle and method and that the only difference is one of degrees, not kinds of processes. I believe I quoted enough of Carrier’s words in my first post to demonstrate that that is indeed what he claims.
Carrier does not say that we can “reasonably doubt” that there was a WW1. He does say, however, that there has to be some room for doubt in all historical claims — even if that doubt is of an infinitesimally low level. I tried to make this point in my first post. That is what I object to — that the nature of all historical knowledge comes in principle and “in fact” with some level of probability and never with absolute 100% certainty.
I spoke of “infinitesimally low level” of doubt in my previous post to try to point out that that is not what most historians (or social scientists) who address such questions believe about events that they study or that they know happened in the past.
Is there even room for 0.000000000000000000001% doubt that there was a World War 1? Or add any more zeros as you like and I still ask the same question. No, I do not believe that there is any room for one gizillionth of a doubt that there was a WW1.
The reason for this is that events like that are not susceptible to probability analysis. They happened. There is no probability about them. They are as clear as day before us as Carrier’s “undeniable” facts such as knowing one is reading words on this page/screen. The evidence we have of those events is as secure as the evidence we have that someone wrote the words you are reading now.
The point is more than semantics. It is critical to the entire approach to how historical analysis is done. Whether Jesus existed or not is not a question of probability. He either existed or he did not. See my earlier post where I cited the important book Systematic Empiricism: Critique of a Pseudoscience.
(I do not say that every conceivable event in the past is certain, but most of the ones historians deal with are.)
I have no intention to somehow argue here, I just try to give my feedback in case it could have some value for you. In my viewpoint, you interpret the quotes of Carrier like, e.g.,
“History is the same. The historian looks at all the evidence that exists now and asks what could have brought that evidence into existence.”
differently than what they express in their contexts: by your texts I would get an impression that Carrier claims that the historical events are governed by the laws of science as is the physical matter of the cosmos. But, in fact, I have not found anything like this in Carrier’s texts, and I fully agree that the historian looks at all the evidence that exists now and asks what could have brought that evidence into existence. (Of course, this quote does not describe the work of historians fully.)
I myself cannot observe the World War 1, so I am relying on the evidence that exists now, and I indeed accept that there is an infinitesimally low (but nonzero) level of possible doubt. If you say that you are 100% sure that WW1 happened, then I cannot see any fundamental difference between us; for me you are just quite reasonably neglecting the infinitesimally low value. But if you indeed can see this difference between us as fundamental, then let it be so.
Another example where I find your text a bit misleading: When you say
“Whether Jesus existed or not is not a question of probability. He either existed or he did not.”,
you give an impression that you are exposing a fundamental problem in Carrier’s approach (which you also aim to illustrate by some quotes from Systematic Empiricism). But I do not doubt that Carrier agrees that Jesus existed or he did not (after we make clear what we mean by “Jesus existed”). Carrier just tries to demonstrate that we have really good reasons to doubt his existence (which contradicts the widespread claims that his existence is certain, being simply a historical fact). I find Carrier’s endeavor to quantify the level of our (un)certainty as valuable; we might surely have reasons to be not convinced by various concrete steps in his analysis but I do not find anything fundamentally flawed there.
Neil, I hope you take my critical remarks calmly (unlike Richard Carrier in my email communication with him :-)). If they have no value for you, then let it be so …
I certainly do take your feedback “calmly” and without any thought that you are being in the slightest polemical or hostile. I do appreciate your feedback. I am trying my hardest not to be adding to the hostile polemic that is “out there on the web” against Carrier. I trust others and Carrier in particular will see nothing hostile in any of my criticisms.
Have you read Carrier’s Proving History? The next post I am preparing makes it absolutely clear that Carrier does indeed equate the historical methods with those of science. He makes it perfectly clear that historical explanations are equivalent to scientific theories and that it is only semantics that leads us to think differently.
If you think I have misled anyone with my statements than I would very much appreciate it if you could give me the page references where my comments are demonstrated to be misleading or misrepresenting Carrier in any way.
For all practical purposes there is no difference between someone who believes WW1 happened as a fact without a shadow of a doubt and the one who believes it happened with the infinitesimal sliver of a doubt.
But Carrier goes to some length to justify his claim that historical knowledge of events is the same as scientific knowledge and that both are always “probable”.
I consider it simply impossible that there should be any realistic — even infinitesimally small — room for any doubt whatsoever that WW1 happened. I was not there but that doesn’t make any difference to the very clear and undeniable evidence I can see that leaves no doubt that it occurred. The gravestones, the war memorials, the publications with pictures and copies of documents and diaries, and so on. WW1 has no probability status to it at all. It does indeed come to us with a probability of 1, if you like.
It makes a difference because Carrier’s point about all knowledge being probable is the foundation on which he rests his claim to “predict” that Jesus is unlikely to have existed.
You referred to my quote: “History is the same. The historian looks at all the evidence that exists now and asks what could have brought that evidence into existence.” —
What I was trying to point out is that Carrier is saying here that history is “the same as science” — he is making a direct comparison between historical methods and those of geologists etc. But it is not the same. Geologists don’t ask “what could have brought this mountain into existence?” They know before they start what brought it into existence — they understand and know that the physical laws and how they brought it into existence. Scientific explanations are not just different in degrees from those of historians, as Carrier writes, but they are different in kind.
The scientist is viewing the world through abstract laws.
Carrier writes: “a historian can make valid (but still general) predictions about the future course of history, if the same relevant conditions are repeated (such prediction will be statistical, of course, and thus more akin to prediction in the sciences of meteorology and seismology, but such inexact predictions are still much better than random guessing). Hence, historical explanations of evidence and events are directly equivalent to scientific theories, and as such are testable against the evidence, precisely because they make predictions about that evidence.”
This again is wrong. It is also wrong about the role of statistics in the kinds of knowledge they lead us to infer. Again, all of this was addressed long ago in the social sciences in the J. and D. Willer book that I quoted from.
I appreciate your friendly approach, which encourages me to write still an addition, trying to illustrate two basic points that I do not understand in your texts.
1/ When geologists ask the questions like “What has brought the Alps to existence?”, the answer is not “Oh yes, the physical laws that we know”, but they try to describe concrete “historical” geological processes that (might) have lead to creating the Alps. I am not a geologist but I believe that there are usually several possible variants of such processes, and the geologists try to find the most probable ones, using the evidence they have and the relevant “laws of nature” that are known.
I cannot see any flaw in Carrier’s analogy with historians; the historians should proceed similarly in the sense that they search for the most probable descriptions of historical processes, using the evidence they have and the -relevant- “laws of nature” that are known.
I am not aware that Carrier would say that the “laws of nature” that are relevant for geology are the same as those that are relevant for history, and I thus cannot see that you have exposed any flaw by Carrier in this matter.
2/ You seem to find as (methodically) faulty to accept that there is a quite negligible “infinitesimally small” level of possible doubt regarding the existence of World War 1, though you have not demonstrated which faulty consequences such acceptance can have, or, more concretely, which faulty consequences have appeared in Carrier’s work.
Related to this, let me recall the case of Septuagint. You know that we have “historical reports” that 72 translators translated Hebrew Bible into Greek identically while working independently. This event obviously either happened or did not happen. Can we say something on the “probability”, i.e. on the level of (un)certainty we have in this case? I think we can. Knowing the relevant “laws of nature”, the prior probability of such an event is non-zero but extremely small. Since the evidence (the “reports”) we have is clearly weak, the final (posterior) probability we attach to the event is (slightly bigger but) still negligibly small, being zero in practice. (If I wanted to provoke you :-), I would say that I am more sure that the 72-event did not happen than that WW1 happened.)
Another example: I am from Czechoslovakia (now Czechia). If somebody told me in the spring 1989, when Vaclav Havel was again arrested, that the Czechoslovak communist parliament will unanimously elect Vaclav Havel president still in the year 1989, I would find this somebody crazy. The probability of such an event was -practically- zero for me (in spring 1989). But it did happen. (It was like a fairy-tale, a miracle, not only for me.) Nevertheless, I cannot say that I was mistaken in the spring 1989. I believe that, given the evidence I had at that time, my probability estimation was in principle ok.
Finally I can say that I find your formulation “Carrier “predicts” that Jesus is unlikely to have existed” as a bit misleading. In my view, Carrier analyses the evidence for Jesus’ (non)existence that we have, taking the relevant “laws of nature” into account, and comes with an estimation of our (un)certainty level in this case. He surely knows that Jesus either existed or did not exist, and he clearly does not exclude that a new evidence might appear that would influence his estimations. So, Neil, also here I can not understand your point against Carrier.
In my next post I might address more specifically some of the points you raise in your comment here.
Carrier uses the word “expect” — such and such is what we would “expect” — and he means by this that the evidence was predictable on the basis of what we know beforehand (this meaning is clear in his Proving History). But all he is doing is describing the state of the evidence and not predicting anything. He argues that history and science work by the same methods of hypothesis + prediction. They simply don’t.
This is important, as I said before I think, because it is the foundation on which Carrier justifies his argument for the (probable) non-existence of Jesus. He is attempting to equate the conclusion with the result of a “scientific” method — with mathematics to enhance the confidence level.
Geologists and historians are thinking entirely differently in assessing causes. Historians have no laws with which to think about the past and apply to events.
Just one more thing while I prepare my next post….. In response to your point:
I agree with you that you are describing your personal level of “certainty” — or your expectation that the event would have happened. But one’s confidence or feeling of certainty is irrelevant to whether the event happened or not.
But Russell Gmirkin sees in the account an apt explanation that is compatible with his hypothesis that the OT was translated in the Hellenistic era. The event without the miracle in his view is very likely to have been historical because of its explanatory power.
Carrier attempts to argue that Bayesian probability and “frequentism” are the same. But they are only the same if one has, in Carrier’s words, a very large number of frequent events to work with. I think he is understating it. How many dice rolls would one need to make before one had all possible sides finally turning up each 17% of the time?
Bayesian probability would work for historical events if we had an almost infinite of events to work with as options for comparison but we don’t. In the Rank-Raglan case, for example, we have only a score or so of events. Throws of the dice can result in a string of the same number turning up. That’s how it is with historical events. They are not “probable” or subject to probable interpretations of their chance of occurring — they are each contingent.
So with the example of the 72 translators, we have no reason to believe that it definitely was a tale based on a real event. But we do have one historical analysis of other evidence that leads to a strong likelihood that there is some historicity to it. But if we rely on mathematics and probability then we can say it most likely didn’t happen. The former is the historical analysis and conclusion. The latter is a theoretical conclusion that really can only be justified by reference to an essentially infinite number of comparable events, which do not exist in the real world.
I am glad that you might plan to address some points I am raising in your next post. Thank you.
I will thus wait for it, hoping that I will understand you more after that. Now I just feel like adding a few additional points to be potentially considered by you (in this next post); they have been triggered by your last responses (which contain further things that I do not understand at all).
– I do not understand why you say
“But one’s confidence or feeling of certainty is irrelevant to whether the event happened or not.”
I find this statement trivial, and I cannot see any relation to what we are discussing. E.g., I am sure that Carrier would agree that all his analysis is irrelevant to whether the respective event (historicity of Jesus of Nazareth) happened or not; i.e., it is trivial that Jesus either existed or did not exist independently of Carrier’s analysis. (Carrier just says: since we do not know whether Jesus existed or not, we can only analyze the evidence we have and come with a “probability” estimation.)
– You say “Carrier attempts to argue that Bayesian probability and “frequentism” are the same.”
I have just randomly look at a text by Carrier on Bayes; e.g. in
https://www.richardcarrier.info/archives/9581
I can read
“Flam summarizes how frequentism fails to accurately describe reality, and that major scientific errors are being corrected by bringing problems back to a Bayesian framework where they should have started.”
Neil, I hope you understand that I do not understand your claim at all.
– Your claim “Historians have no laws with which to think about the past and apply to events.” sounds to me like “Historians can do whatever they want in explaining the historical processes”. So maybe that you will explain this in more detail in your next post. Thank you.
On your first point, I was attempting to address the difference between our confidence that something happened on the one hand and the fact of it having happened on the other. What Carrier is doing, I believe, is attempting to measure the extent to which we would expect certain evidence to be found. But our expectations and confidence have nothing to do with the reality of what might be actually found. Indeed, Carrier speaks throughout of “expectations” (and explains these are a form of prediction) but a moment’s reflection indicates that all Carrier is doing is describing what we have and not “predicting” anything. What he calls “expected” evidence is nothing more than what we have.
As for frequentism, yes, a sample is not the reality of the whole. But Bayesian probability is even less so where the data are limited. Bayesianism says the odds of a die roll turning up a 1 are 1 in 6 or (as Carrier says) 17%. But we know that you can throw a die 20 times and still not get a 1, but you might get more than half a dozen 5’s. Historical data is like that die roll — we cannot predict from limited numbers of examples for comparison. I will be addressing this in more detail in the next post.
As for historians not having laws, no, that does not mean they do not have methods. Recall my first post where I noted the difference between empiricism as a method and empiricism as a philosophy of knowledge. Historians follow empirical methods. But they have no abstract laws through which to explain events like the geologist does — who interprets all the data in the context of abstract physical laws of pressure, mass, etc. that enable real predictions to be made.
One thing I will say is that I actually do think that things change depending on the era and context. For instance, I do find that when we study ancient history, there is a larger proliferation of arguments about whether X or Y event did actually take place. For instance, until rather recently (when scholars essentially threw up their hands), debating the tidbits and life of Socrates and the specifics of it, as well as how much the dialogues pertained to any historical events of his life, was fairly commonplace. Likewise, poorly documented events are often debated. I am part of a pretty long tradition among critics to enter into the debate on whether or not the Neronian Persecution was a historical event (a debate stretching back into the 1800s). The historicity of the “Jesus” who predicts the end of the Temple in Josephus’ Jewish War, as well as all the events around him, are hotly debated in some circles now too. These are all cases where entire events, and whether they happened at all are debated, and that list is pretty exponential in terms of ancient history.
And part of that is because of recent increased awareness at just how willing ancient Greco-Roman historians and biographers were to just make things up, or report folklore as a fact, which makes the lives of Nero and Caligula, in particular, quite difficult to assess at points. Lots of these kinds of debates, of course, also persist in Ancient Near Eastern studies as well, because of a mixture of poor documentation, and then the same tendency to just report things that never happened (or likely never happened).
So, while it is rare, particularly when discussing more modern history, these debates get more common the further back you go just out of necessity.
I mentioned one such question arising in the post — the question whether the gingerbread vendor was kicked to death by a mob.
Certainly there are many such questions, but they are embedded within the world of events of which we do know. Even in modern history we can raise questions about the role of the CIA in this or that political event in some country, etc etc etc etc. There are many unknowns and debated questions in history of all periods. Historians worth their salt acknowledge the status of them when they use them — and usually provisionally if they are controversial.
That those questions arise demonstrates my point: that we are very well capable of sifting the known from the unknown or questionable in the past.
(Some have pointed out that contemporary history is the hardest to write simply because so much is not “lost” as in the remote past but inaccessible, often deliberately hidden, and often won’t be available until the present becomes a distant past.)
“It is not the facticity of the events that is in question
It is at this narrative level where problems and disagreements, doubts and uncertainties, among historians arise.”
In my mind, Neil, I keep coming back to your oft-stated position that your interest lies in Xtian origins; also, in the above responses, you repeat the idea that the historicity of Jesus is not the issue for you. The quotation above, creating a dichotomy between facticity and the “narrative level” raises an interesting issue, which is that determining the facts of narrativity sometimes intersects with determining the facts, qua facts. So my question for you is, what is your view of Robert M. Price’s idea (probably the idea of many others as well, but I recently encountered it in Price) that the Gnostic gospels, or certain of them, tell the same story as the church gospels, and that the church gospels are simply literalizations of the Gnostic gospels?
Can you give me the source for Price’s discussion? I don’t recall having read that one.
I don’t see how any narrative alone and of itself could ever enable any determination of the historicity/real-world extra-narrative-factiveness of anything.
This was from “The Amazing Colossal Apostle: The Search for the Historical Paul.” Signature Books 2012. Don’t have the page number, but I think it’s in Chapters 4-5. If the Gnostics were writing the same story as the church evangelists, then the question arises as to whose story came first. If the Gnostics were first (in contradiction to the accepted wisdom), then the canonical works would be fictions based on a previous fiction. You almost don’t need that, because of the nature of the canonical story (fantastic), but it would be another determiner of facticity or fictionality.
Other writers have said much the same thing, i.e., that the Gnostics considered the story of Jesus true, but taking place in the purely spiritual realm.
What is your view on the question? I am sure you have given it much more thought than I have. I can see little reason to date the gospels to the first century and several reasons to date them to the mid second, while some form of gnostic Christianity surely antedated them. That perspective doesn’t require any mental gymnastics or speculative interpretations relating to the contents of the canonical gospels. Rather, it seems to me to be the simplest and least complicated scenario on the cold hard basis of the evidence. I’m not suggesting Christianity itself began then, of course — surely the violence involving the Judeans from 70 to 135 had something to do with it taking the shape it did as per the canonical gospels. As for historical events behind the narratives, I am sure you know my view that we cannot presume anything but need independent evidence to go down that route.
I’d like to hear your thoughts on these questions.
“As for historical events behind the narratives…”
Robyn Faith Walsh’s post announcing her chapter “City and Country” @ https://bsky.app/profile/zafulotus.bsky.social/post/3l72sukdzqg2f
https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:djgdb3lcldda2k3ek5u2yskt/bafkreiahl5oj6ngabm5mqaagmmptmaj63mbhqlekmupd5yxfvx6gi33oqu@jpeg
Perhaps future scholarship might also explore the socio-political and cultural landscape of the time, per the relationship between urban and rural spaces. Providing insights into how early Christian communities formed and interacted with their surroundings, potentially influenced by the events of the Judean-Roman conflict.
This Price book argues that much in the Pauline material is written by Marcion, and that ideas in the letters are often gnostic, with “Catholic” interpolations to clean them up. In short, they are cannibalized documents, almost totally pseudonymous. I haven’t the knowledge to question the positions taken. Dating is the most interesting issue, but it’s one on which I don’t have great expertise. Justin Martyr seems to have very little to say which seems to come from a gospel, although I think some small part of it may come from the synoptics. Irenaeus is concerned with the gnostics. So I probably agree with you (and with Price) that some form of gnosticism was around at or before the composition of the canonical gospels. In that case, a canonical gospel could have taken orally transmitted stories about the demi-urge and the archons, etc., with the outline of the descending son who is then crucified and ascends–all of it symbolic–and simply literalized it into the stories we have–stories strangely lacking in any physical description of the hero, because he originally had no physical body. This would be consistent with the supposedly earliest gospel–Paul’s–in which that little outline is the entire story. It’s all speculative, of course. But it shows how the issue of the dating/ordering and content of fictional texts leads to a conclusion about a physical fact, in this case the non-existence of Jesus as a real person.
Ah, I see your point. Indeed, those kinds of analyses of the sources and the hypotheses they generate do offer a cogent explanation of the origin of the figure of Jesus in the gospels. So we have before us a literary figure of Jesus and we have hypotheses that explain that figure as originating from a mystical idea of some kind. Though even before we compare the gnostic ideas with the gospels we have already seen that the literary Jesus in the gospels was sown together from threads in the OT. The gnostic hypothesis adds an additional (prior) source. So there is no room for a historical Jesus. And though the question of the historicity of Jesus does not arise I can agree that the gnostic source hypothesis does offer a stronger explanation for the appearance of Jesus in our sources that is an alternative to speculation that a real Jesus lay behind the gospels.
As for the question of the historicity of Jesus, would not the gnostic source for the canonical gospel Jesus be much the same as an OT “midrashic” source for the gospel Jesus — both offer explanations for the gospel Jesus that have more to support them than the assumption that a historical Jesus was known from oral traditions. Either way, the focus is on the Jesus we have, the literary/theological figure, and the historicity question is never raised. Of course if these other explanations are of sufficient scope to render a historical figure unnecessary, then so be it.
Take it a step further. If, as I fully agree with you, the figure to be concerned with is the literary/theological figure–which only makes sense because he appears in what everyone should agree is a literary/theological source, the gospels–then the proper subjects of historical research are literary and theological, not archaeological and geographical. The influence of the OT is Jewish. The evangelist who makes use of tropes from Isaiah, etc., is trying to bring the Messiah into the story, and to combine it with the effects of 70-135 C.E. The Gnostics, particularly Marcion, are going in another direction, more exclusively Greek. You have the Paul/Peter confluence. The historical Jesus is a confluence of Hellenistic influences, particularly Greco-Hebraic. He is not to be found in discussions of a Jewish peasant, a magician, influences of the Sicari, etc. We need to look at what religious enthusiasts believed more than at how people lived their daily lives.
Yes indeed, exactly so. It is the data relating to “who believed what and when” that I have been mulling over for some time now. One place I have not looked at yet is the Mishnah. It is late but is known to document some traditions about the views of earlier persons. I suspect the pre and post 70 CE gentile proselytes or god-fearers had a significant role. Were there diaspora-Palestine debates?
Russell Gmirkin and a handful of others have argued for Greek literary inspiration in the formation of the OT and done so entirely through comparative literary analysis. A comparable approach seems necessary to me — and you, I think — to understand the origins of the NT writings.
Dear Neil, believe me that it has been very difficult for me to write the following post. But I hope that you prefer an open communication even if the content might be (polite but) unpleasant for you.
You seem to think that you have exposed substantial flaws in Carrier’s approach but I am afraid that your texts give an impression that you have not understood Bayesian probability
(https://en.wikipedia.org/wiki/Bayesian_probability)
and the related analysis that Carrier is performing, and explaining in many articles (also in his blog
to which I referred).
Let me just quote from the above wiki-page
“In the Bayesian view, a probability is assigned to a hypothesis … a prior probability … updated to a posterior probability in the light of new relevant data (evidence).”
You say “Bayesianism says the odds of a die …” without giving any hypothesis etc., and then you say
“But we know that you can throw a die 20 times and still not get a 1, …” (so I phrase it: an event whose probability can be calculated to be just 0.026 sometimes happens), as if this was an argument against a method by Carrier or so.
Bayes tells us, e.g.: if you get a suspicion (a hypothesis) that the die is constructed so that you never get 1, and you estimate the prior probability of your hypothesis rather conservatively to be just 0.01, then you can start testing: if the next 10 throws show no 1, you calculate your posterior probability (by Bayes theorem) to 0.05, after next such 10 throws to 0.24, then to 0.66, 0.92, 0.98. (I hope my numerical calculations are ok, I did them just for illustration quickly on a simple calculator.)
If such a thing happens, I am rather sure that my hypothesis is correct (though I am not 100% sure, of course).
Neil, you can naturally tell me: but we cannot make such testing in history, we have the evidence that we have and we cannot produce more evidence on demand or so…
Yes, you are right, Carrier surely works just with the evidence we have. But he uses it to test hypotheses.
Let me illustrate this by my own example.
Suppose I want to test my hypothesis that Jesus of Nazareth existed (I refer to a very basic hypothesis that he was the founder of Christianity which was afterwards spread by his disciples …).
I am so confident that I put the prior probability to 0.95, say, since I heard many people (not only biblical scholars) to be so confident …
I decide to read the Christian literature we have, starting chronologically as it is presented to me. Say, I start with the epistles, and ask myself if they are relevant for influencing my hypothesis, and in particular if they fit well with it. At the same time, as a Bayesian, I ask if they do fit also with some
of the contrary hypotheses, like with a hypothesis that the Christ Jesus first appeared in the minds of some people when studying Hebrew writings, hearing the ideas like those of Philo of Alexandria, etc.
I have (and really had in my life) to admit that I did not -expect- under my hypothesis that in these texts I will not find anything about Jesus of Nazareth when some mentions would be so appropriate in so many places etc. etc. Moreover, among many discussed problems, there is nowhere an issue that a recent man is equal to God; it looks like nobody had this problem, though I would “predict” (by all my life experience etc.) that this would be one of the main issues, in particular among Jews.
On the contrary, the texts fit very well with the contrary hypothesis I mention above.
Instead of throwing a die, I read one text after another (including, e.g., Hebrews saying: If he were on earth, he would not be a priest, … and all such places).
To make it short: the calculations (setting priors etc.) are surely more rough than at the throw of a die, but the result was clear: Bayes has indeed very lowered the probability of my hypothesis …
I hope, Neil, that you can now see how I understand “predictions” and “expectations” at Carrier; I have no problem with his terminology. So maybe you take this somehow into account when composing your detailed criticism of Carrier. Best wishes.
Dear Neil, believe me that it has been very difficult for me to write the following post. But I hope that you prefer an open communication even if the content might be (polite but) unpleasant for you.
Rest assured, I found nothing at all unpleasant in what you wrote. I have always welcomed disagreement where it helps us work towards clearer explanations and understandings, both of our own and the other’s.
You seem to think that you have exposed substantial flaws in Carrier’s approach but I am afraid that your texts give an impression that you have not understood Bayesian probability
(https://en.wikipedia.org/wiki/Bayesian_probability)
and the related analysis that Carrier is performing, and explaining in many articles (also in his blog
to which I referred).
I am responding primarily and specifically to what Carrier has written in his books Proving History and On the Historicity.
Let me just quote from the above wiki-page
“In the Bayesian view, a probability is assigned to a hypothesis … a prior probability … updated to a posterior probability in the light of new relevant data (evidence).”
Yes, that is correct.
You say “Bayesianism says the odds of a die …” without giving any hypothesis etc.,
Okay, I am in conversation and what I meant by that is that when we want to predict a die roll turning up a 1 we say the odds of this hypothesis, that it will come up with a 1, are 17%. I do understand that Bayesianism is about assessing probabilities of hypotheses and have written so, both here and in other forums, exactly that.
and then you say
“But we know that you can throw a die 20 times and still not get a 1, …” (so I phrase it: an event whose probability can be calculated to be just 0.026 sometimes happens), as if this was an argument against a method by Carrier or so.
I am using an example that Carrier himself discusses in Proving History (pp 265f with an earlier implication of the same on p 86), and a response to this kind of reasoning being applied to the “real world of events” in J. and D. Willer’s book, Systematic Empiricism (pp 96ff).
Bayes tells us, e.g.: if you get a suspicion (a hypothesis) that the die is constructed so that you never get 1, and you estimate the prior probability of your hypothesis rather conservatively to be just 0.01, then you can start testing: if the next 10 throws show no 1, you calculate your posterior probability (by Bayes theorem) to 0.05, after next such 10 throws to 0.24, then to 0.66, 0.92, 0.98. (I hope my numerical calculations are ok, I did them just for illustration quickly on a simple calculator.)
This is a different example from the one Carrier uses in Proving History where he speaks only of “a fair six-sided die”, and the example I was addressing.
If such a thing happens, I am rather sure that my hypothesis is correct (though I am not 100% sure, of course).
While I agree that no matter how many times we throw a dodgy die it will always turn up a dodgy result, your calculations are erroneous within the constraints of a limited number of throws. (Endless throws is something quite different and exactly where probability theory works – with data that is theoretically infinite.) But just throwing a die 10 times and deciding to recalculate the odds that a 1 will not turn up as a result is a mistake. If it were a fair-sided die that you were throwing you could still get the same result of anything but a 1 in the next 10 throws. Each throw, each time you throw, even after 20 times, the odds of that 1 appearing are still 17%. They don’t increase because a 1 hasn’t turned up in the previous 20 throws. So if those are the odds with a fair six-sided die, then it would be a mistake to change the odds to test a hypothesis that the die is dodgy — because the same result of a non-1 could have turned up on a fair die after 20 throws.
Neil, you can naturally tell me: but we cannot make such testing in history, we have the evidence that we have and we cannot produce more evidence on demand or so…
Yes, you are right, Carrier surely works just with the evidence we have. But he uses it to test hypotheses.
Where I disagree with Carrier is that he reduces all historical knowledge to a hypothesis. Every event has a probability less than 1 of being true. It may be a very slim sliver less than 1, but it is important for Carrier that everything, all claims made by a historian, are at best “probably” true.
I come across this claim about history quite often — even by Tim O’Neill who writes History for Atheists — and it is simply flat wrong. Indeed yes, Bayesianism is useful when applied to assess hypotheses. No question. Agreed 100%. But a historians facts, on the whole, are not hypothetical at all. (I am leaving aside those specific questions where the historicity of a particular person or event is wondered about — but those are not the bread and butter of historical research. They are theoretical discussions that really demonstrate how most of the known events in the historian’s consciousness are not hypothetical and do not have any probability status about them in the slightest. What is hypothetical and what the historian is questioning are interpretations of those events.)
Let me illustrate this by my own example.
Suppose I want to test my hypothesis that Jesus of Nazareth existed (I refer to a very basic hypothesis that he was the founder of Christianity which was afterwards spread by his disciples …).
I am so confident that I put the prior probability to 0.95, say, since I heard many people (not only biblical scholars) to be so confident …
I decide to read the Christian literature we have, starting chronologically as it is presented to me. Say, I start with the epistles, and ask myself if they are relevant for influencing my hypothesis, and in particular if they fit well with it. At the same time, as a Bayesian, I ask if they do fit also with some of the contrary hypotheses, like with a hypothesis that the Christ Jesus first appeared in the minds of some people when studying Hebrew writings, hearing the ideas like those of Philo of Alexandria, etc.
I have (and really had in my life) to admit that I did not -expect- under my hypothesis that in these texts I will not find anything about Jesus of Nazareth when some mentions would be so appropriate in so many places etc. etc. Moreover, among many discussed problems, there is nowhere an issue that a recent man is equal to God; it looks like nobody had this problem, though I would “predict” (by all my life experience etc.) that this would be one of the main issues, in particular among Jews.
On the contrary, the texts fit very well with the contrary hypothesis I mention above.
Instead of throwing a die, I read one text after another (including, e.g., Hebrews saying: If he were on earth, he would not be a priest, … and all such places).
To make it short: the calculations (setting priors etc.) are surely more rough than at the throw of a die, but the result was clear: Bayes has indeed very lowered the probability of my hypothesis …
I hope, Neil, that you can now see how I understand “predictions” and “expectations” at Carrier; I have no problem with his terminology. So maybe you take this somehow into account when composing your detailed criticism of Carrier. Best wishes.
Yes, I understand that process. You are weighing up the evidence to see if it is consistent with a hypothesis. For that reason I can say that the most we can say about Jesus is that he is a hypothetical figure, that his historical existence is a hypothesis to help explain Christianity. You had a hypothesis and checked to see if the evidence was consistent with it or rather with another hypothesis.
But the heated, even hostile, arguments are not over whether Jesus’ existence is a hypothesis. They are about the black and white question, “Did Jesus exist?” Those who say “probably a 30% chance he lived” are excoriated as “mythers” who are on a par with moon-landing deniers.
I happen to think that Carrier has reduced all historical knowledge to a series of probability claims and put Jesus alongside other historical claims of probability — the only difference being that some probabilities are more likely to have been “true” than others. My contention is that that is not how historians think about events of the past (despite a specific subset of exceptions as I mentioned above). Carrier has lumped in one footnote a score or more historians who have often opposing views of the nature of historical knowledge and claimed that when all of their historical reasonings is at its best it can always be described in Bayesian terms. Such a claim is surely correct for testing interpretations and hypothesis but not as a blanket assertion applicable to whether Churchill, Cromwell, Julius Caesar or Heracles existed.
Just one more point….. You wrote:
Apart from the Bayesian question, making “predictions” as Carrier and you term the process, or looking for what is expected in the data, is actually a dangerous approach to historical research. Agreed, we always look to find evidence that supports a hypothesis that we are considering, but by focusing on what we hope to find we are always in danger of confirmation bias — until, of course, we take the next step of seeing what we can find in the data that speaks against our hypothesis. I am sure Carrier would agree, and you also, but I just wanted to make the point for the sake of the record.
As an example of the problem here, I’m reminded of your earlier example of the account of the 70 translators of the Bible in Alexandria. One can find endless reasons to believe that the story is entirely fictitious. But then it only takes one other hypothesis coming in unexpectedly from the left field, as they say, and suddenly the account is found to contain a strong explanatory power for the development of the OT as we have it. So the historicity of the event suddenly becomes highly likely despite the result that we arrived at when we focused on it alone as an event for its own sake, in isolation from anything else.
I tried to post a reaction here at 2024-11-24 around 20.00 GMT but for some reason it has not appeared here. At the moment I will not try to repeat it, but instead I try to post the following short question regarding a thought experiment for you, Neil.
Suppose there is a machine with a push-button and a monitor.
Person A tells you: whenever you push the button, number 1 appears on the monitor.
Person B tells you: whenever you push the button, a random process starts which results in 1 or 0 appearing on the monitor, each with probability 0.5.
Suppose you are a judge who should test these hypotheses against each other. You are completely impartial, so you start with attaching the confidence (prior probability ) 0.5 for the claim of A and 0.5 for the claim of B.
Suppose you push the button 10x, and always 1 appears.
You then do this several times, the second time 10x, the third time 10x, … and always 1 appears.
Would your impartiality be steady (all the time keeping the confidence 0.5 : 0.5)?
a/ If yes, then we have exposed the basic level on which we disagree, and we can finish our discussion.
b/ If no, and if you would be stepwise getting more confident in the claim of A, and you would eventually declare A as the winner (similarly as me), how would you react to the following complaints by B:
1/ the result of every button-pushing was completely consistent with my claim;
you should note that the chance of 1 is always 0.5, independently of how many times we have seen 1 before.
2/ the fact whether or not the machine has the property that I claim is independent of your confidence, and you have performed just finitely many tests anyway …
c/ If none of a/ and b/ applies, how would your reasoning go?
Please, do not try to escape by saying that this example was not in Carrier’s book or that historians are not solving such problems. You simply said “your calculations are erroneous within the constraints of a limited number of throws”, which is another of your claims that I do not understand at all; hence an honest reaction to my thought experiment will help me to clarify your reasoning. Thank you very much.
I trust you think my responses have been in good faith as I have trusted you questions also are.
If I pressed the button so many times I lost count and always a 1 appeared I would say that B’s point is unlikely to be true. Eventually I would probably give up and say A is correct.
But I don’t understand how that problem relates to what Carrier is doing in Proving History or in On the Historicity of Jesus. The “mind experiment” you present is consistent with probability theory and works only in cases where we have the ability to press the button of the machine as many times as we like.
My point in my earlier comment was that if we threw the die ten times it would tell us nothing about the truth of either A or B’s claim. After 10 throws of getting always the 1 result, yes, I would begin to suspect A was correct — that would be my confidence level, if you like, is being shaken. But certainly if we could keep on testing then we could become convinced that A is correct after all.
If you suspected I was thinking differently on that point then I suspect there is something I do not understand about your position and what it is that you do not understand about mine.
Hi Petr, You wrote,
The sentence of mine you said you do not understand is illustrated perfectly by the example (mind-experiment) you offered me. If I was allowed only a limited number of presses of the button, say 10, and no more, then I would be tempted to think that if 1’s appeared every time that the machine was built to produce only 1s. But I would also permit myself a niggling doubt that ten button presses was not enough. Since I am allowed an unlimited number of presses then I can reach a high confidence that the machine is built in a certain way.
But in the example of the Rank-Raglan hero myth type, Carrier has only a handful of instances on which he bases his probability that Jesus was also a RR type. But that is an invalid conclusion because the number of historical examples he has selected for comparison is limited.
He even assigns a 50% figure to the numbers of attributes which are relevant, and that 50% figure excludes the examples of historical persons in the RR type. But the list itself is of a “random” length — it could have been much longer according to Raglan. So who knows who would have qualified for the hero myth list if the number of attributes was much higher, as it could have been.
Now that does not have to be the end of the story. One can proceed to use other background knowledge to test the original notion that Jesus was of the hero type and therefore likely not historical. But I will show in my next post that even those processes are logically flawed – they are based on circular reasoning.
So the whole game is flawed, I believe.
Don’t get me wrong. I “believe in Bayes”! I acknowledge Bayes is a useful tool for assessing hypotheses. Unfortunately, I wonder if Carrier has confused events and hypotheses by reducing everything to a probability statement. Historical events are unique and not subject to probability calculations. We can list frequencies of certain types of events, but Carrier also equates frequentist thinking with probability. I believe, especially after having read the Willer and Willer book, that he is mistaken.
I sincerely apologize, Neil, if my expression “Please, do not try to escape by saying …” sounds inappropriate to you. I am not a native English speaker (as is surely clear from my posts), and I surely have not meant this somehow badly. But in retrospect I can see that I should have indeed formulated this differently. Sorry for that.
Hence I will try to be more cautious in this respect, but I will also try to write more directly so that the things get hopefully clear more quickly.
So when I wrote “You simply said “your calculations are erroneous within the constraints of a limited number of throws”, which is another of your claims that I do not understand at all; …”, I should not have said “I do not understand” but “you claim that my calculations are erroneous without providing any valid argument for your claim”.
In fact, I then presented a much simpler mind-experiment and asked you specifically how you would answer to the complaining B (the points 1 and 2 in b/), but you have not reacted to this. Ok, I surely trust your good will, so I conclude that you have probably thought that this was just a rhetorical question of mine. (No, it was not rhetoric, it was crafted so that you would have to react against your own arguments that you presented in the previous discussion … And you surely agree that a serious attempt to argue against our own arguments is a very recommended part of critical thinking.)
Back to the mind-experiment: When you say “If I was allowed only a limited number of presses of the button, say 10, and no more, then I would be tempted …”, you are describing just your intuitive feeling. Later you say “I believe in Bayes” but you have not really applied it here. A simple calculation would give you that after one press with 1 appearing your confidence ratio A:B should change from 1:1 (i.e., 0.5 : 0.5 in probabilities) to 2:1, after next presses to 4:1, 8:1, 16:1, and after 10 presses it should be 1024:1, hence your confidence in declaring A the winner after 10 presses should be 0.999. Still too low for you?
(Btw, I stress that we just compare two hypotheses against each other, we are not claiming to have learned how the machine is programmed. Maybe that the 11th press would give 0, and Bayes tells you A:B is 0:1 then.)
Now suppose B starts arguing, after we have these 10 presses, and we cannot do presses anymore. B says: You can see that the evidence from the first press is rather non-decisive, so let us cancel registering this first press. Similarly B would argue further, wanting to ignore some evidence we have. An impartial judge surely agrees that any out of these 10 presses is rather non-decisive but does not allow cancelling, saying that all relevant evidence must be taken into account.
We are coming to Carrier’s comparing his hypotheses A:B against each other (A is the minimal historicity hypothesis, and B the minimal myth hypothesis). We try to take all relevant evidence that we have into account. During this process we also encounter the Rank-Raglan issue. It is clearly rather non-decisive by itself (similarly as each concrete press in the previous experiment), but is it relevant or irrelevant w.r.t. A and B? Carrier argues that it is relevant and tries to estimate how, i.e., what is this small bit that RR provides for our calculations? Since now we are in the situation rather less clear that in the machine-experiment, the estimations are more arguable. Carrier decides to argue “a fortiori”, being not impartial but intentionally favourable to historicity (i.e., to the opposing hypothesis than this in which he is primarily interested).
From your point of view, Neil, Rank-Raglan is completely irrelevant here, or it is relevant but you would cancel it from the considered evidence since it is completely non-decisive by itself?
I rather stop here, waiting to see if this has somehow helped to change your understanding of the problems that I have with your approach.
I would say that though each press of the button was indeed a 0.5 chance of a 1 if the machine were configured to truly randomly generate a 1 or one other number, in sum, over so many presses of the button it is most likely, very likely to the point of certainty if there were enough button presses with the same result, that B was wrong.
Of course what the machine does is obviously not dependent on my feelings. My feelings are only an indication of what where I think the result is heading.
If I understand you correctly, I disagree with your point here. In the case of a limited number of possibilities, if I throw a 1 on a 6 sided die, the odds that the next throw will be a 1 will not change at all. It is the cumulation of many many throws that one sees the numbers predicted by “probability” at work. If I was attempting to test a machine to see whether it allowed any number other than one, it would only be by many, many presses of the button that my confidence level that it could not would increase. This is the same as testing the odds for a particular number turning up on a 6 sided die — the probabilities are only seen work out in the final results after many, many (an unlimited) number of throws.
I wonder if we are blurring different concepts here — as I think Carrier also does. Comparing hypotheses is exactly what Bayes can do. I agree with using Bayes to compare hypotheses. But whether Japan attacked Pearl Harbor or whether Julius Caesar conquered Gaul — are not and never can be hypotheses.
I disagree. The RR list of features is by Raglan’s own admission a random number and by excluding historical persons of this class by saying they receive less than 50% of the total matches is favouring mythicism, not historicism. It is not a fortiori in favour of historicity. Raglan further indicates that the RR list cannot be used to measure historicity or otherwise of persons, because he explains that mythical stories have a life of their own that is quite independent of the real historical details of historical persons.
I can see that there is a misunderstanding between us already on a very basic level. I think that this misunderstanding is somehow related to the fact that the area of probability, including the Bayes theorem, sometimes goes against our (wrong) intuition, even if we think that we have studied and understood the area sufficiently. (Here is a well-known nice instance of this fact:
https://priceonomics.com/the-time-everyone-corrected-the-worlds-smartest/ .)
To remove the misunderstanding between us step by step, I am now concentrating in detail just on one particular point in our discussion, around your expression
“If I understand you correctly, I disagree with your point here.”
So let us recall my mind experiment:
We have a machine where we can (repeatedly) press a button and see what is displayed after each press. We have no idea how the machine is programmed.
Person A tells us that it always displays the number 1, person B tells us that it is programmed so that it always displays 1 or 0 while each of these two possible results has probability 0.5.
We are NOT guaranteed that any of these claims (let us called them hypotheses A and B) is true, but it is obvious that at least one of them is false (since they contradict each other).
We are given a possibility to make several presses, and we are asked to decide in which of the hypotheses A and B we get a reasonably greater confidence, comparing them RELATIVELY – against each other.
So we can get a higher confidence in one of the hypotheses even when still having a very low confidence that any of these hypotheses is true. Expressed mathematically, we are working with conditional probabilities: we simply assume that one of A and B is true, and we try to assess our confidence ratio A:B under this assumption.
We are completely impartial, so in the beginning (i.e. after zero presses) our prior “confidence ratio” A:B is 1:1, which can be phrased so that our confidence in A is 0.5 (we denote this by P(A)=0.5), and our confidence in B is 0.5 (hence P(B)=0.5). Suppose we make one press, and let us make a case-analysis of our position after seeing the result; there are 3 options to analyze:
a/ something else than 0 or 1 appears (e.g., number 2 or whatever):
In this case our confidence in both A and B changes to 0; we are sure that both hypotheses are false. Here it makes no sense to try to evaluate the “ratio” 0:0, but our confidence in A and B is the same, and we can naturally stop by saying that none of A and B is the winner.
b/ 0 appears:
Our A:B ratio changes to 0:1 (i.e., P(A)=0 and P(B)=1); we are sure that A is false while B might still be true, and B is thus surely true under our working assumption that one of A and B is true. Hence it would be ok to declare B the winner (though our confidence in B being
really true is very low).
c/ 1 appears; we denote this event as E:
In the framework of the Bayes theorem, our expectation of E happening was
P(E) = P(A) x P(E|A) + P(B) x P(E|B) = 0.5 x 1 + 0.5 x 0.5 = 0.75 = 3/4.
(If A is true, then E is sure, i.e. P(E|A) = 1, and if B is true then E has the probability 0.5, i.e. P(E|B) = 0.5.)
So the posterior probability of (i.e. our confidence in) A is
P(A|E) = P(A) x P(E|A) / P(E) = (1/2) x 1 / (3/4) = 2/3.
For B we get
P(B|E) = P(B) x P(E|B) / P(E) = (1/2) x (1/2) / (3/4) = 1/3.
If we “believe in Bayes”, in this situation our confidence has changed favourably for A:
We will continue with the ratio A:B being 2:1, i.e. now we have P(A)=2/3 and P(B)=1/3.
In the case c/, it would be surely unjust to claim A to be winner now (though many people when hearing about the confidence ratio 2:1 unreasonably think that the case is thus practically decided).
We would surely continue with another press (if we have this possibility).
Now in the cases a/ and b/ we would finish with the same results as previously, so the most interesting is the case c/, when again the event E happens (for simplicity we use the same letter E, though we refer to a different press than previously).
Now we have “expected” E more than previously:
P(E) = P(A) x P(E|A) + P(B) x P(E|B) = (2/3) x 1 + (1/3) x (1/2) = 5/6.
We compute the posteriors P(A|E) and P(B|E) as above, which now results in setting new values P(A) = (2/3) x 1 / (5/6) = 4/5 and P(B) = (1/3) x (1/2) / (5/6) = 1/5.
Hence we proceed to the next press with the confidence ratio A:B being 4:1 …
Since you believe in Bayes, Neil, I hope that you now agree with my point, with which you previously disagreed.
I have a feeling that the crucial problem was that you subconsciously thought that we should measure our confidence in hypothesis A by itself. Am I right?
(In our experiment, measuring just our confidence in A is an absolutely hopeless task if only 1 keeps appearing after our presses.
The hopelessness can be seen when, e.g., we would try to compare A with a hypothesis C saying that the machine program is the following:
first generate, in the internal memory, 500 random bits, i.e. numbers 0 or 1, each with probability 0.5, and if all of them happen to be 0, then display 0, otherwise display 1.)
If I were asked to decide if A or B was true, and if a 1 was generated on each of the first three presses of the button and I pressed no more, I would be foolish to conclude that A was true and B false. If my suspicions began to be aroused that 1 was true after 3 presses, I would rationally discount my suspicions and not let them sway me after only 3 presses to make a pre-emptive decision in favour of A.
If I were testing that something had a 50% chance of happening with each button press, surely three presses would be insufficient to make any dent in deciding one way or the other.
If A and B were both equally credible options to begin with, then I do not reasonably begin to suspect A as being correct if only three presses produced a 1.
In a game of Snakes and Ladders if I threw three 1s in a row with a fair six sided die I would not reasonably conclude that the die was loaded against me — only that my luck had run out.
Neil, you have only indirectly confirmed my suspicion on your subconscious; though you say that you “believe in Bayes”, you are not really using it here, and you rather rely on your vague, non-quantified, intuition … 🙁
Since I find this as a crucial problem (maybe even -the- crucial problem) in this series of your vridar-posts, I try to add a few comments, hoping that they still could help to remove the misunderstanding.
But if they do not help after all, and you cannot see any sense in prolonging this discussion, then tell me this explicitly, please (or simply do not react, I will not take it as impolite).
– It is surely correct when you say
“if a 1 was generated on each of the first three presses of the button and I pressed no more, I would be foolish to conclude that A was true and B false.”.
Please,
think now for a moment on the situation when 0 is generated after the first press. Now we are sure that A is false but it would be extremely foolish to say that B is true; nevertheless, our confidence in B is now infinitely bigger than our confidence in A, and we declare B as the winner of this A:B race. (In this case we could only conclude that B is true if we were somehow guaranteed that one of A and B must be true.)
I can only stress once again that our task is not to decide which of the hypotheses is true or false but we compare them to each other by measuring our -relative- confidence in A and B.)
– You say (I corrected the quote by changing 1 to A):
“If my suspicions began to be aroused that A was true after 3 presses, I would rationally discount my suspicions and not let them sway me after only 3 presses to make a pre-emptive decision in favour of A.”
Again, we are not discussing if A is true by itself but how it can be seen relatively to B. It is surely necessary to be more favourable towards A after these 3 presses (all of which generated 1). If we stayed completely neutral, then we would rather forget these 3 presses, and the situation after the next 3 presses with generated 1 would be completely the same as previously (though it is obvious that the situation after 6 presses with only 1 appearing should be clearly somehow different than after just 3 presses).
– So I hope you agree that your mind-situation after 3 presses is indeed somehow different than it was in the beginning, and you are naturally more favourable to A than to B after these 3 presses (each of which generated 1). But how “more favourable” you are? Bayes tells us that your A:B ratio should be 8:1 in this situation. (If you have found an error in my Bayes calculation, then show me where, please.)
But it seems that you do not believe Bayes when it goes against your intuition …
It appears to me that you have misunderstood — or simply disagree with — my critique of the use of Bayes to assess probability of historical events.
Probability is valid where we have in effect unlimited numbers — such as being able to press the button many times in your experiment. It is not valid when dealing with instances of which we have, say, a dozen examples, each of which is contingent in its own way.
Where you say I am disagreeing with or not using Bayes is where I believe it is in valid — after or in relation to a limited number of instances of an event.
If you want to assess my “mind-situation” after so many throws and I reply that I am aware of the need to resist my “mind situation” after only a limited number of presses of the button or die throws, then I am quite correct to do that — to let my rational mind and awareness that probability cannot be used to predict the next die roll or result of a button press.
That is because Bayes cannot be used when the dataset is so limited. Do you have any references to any discussions of Bayes that deny that? I think Carrier would also agree with my point in theory — though in my view of his work he does not accept that historical events are really contingent, and he appears to assume that historical events can be compared and assessed in the same way as a scientist assesses data. He explicitly equates their methods — which is a serious mistake, as I have tried to explain. Historical explanations have no comparison at all with scientific theories.
You have not addressed — and nor has Carrier — the critique I have raised that probability does not apply to historical events to assess their actual historicity. If I say I have a 0.000000000000000000000000001 doubt that Japan attacked Pearl Harbor I would be simply flat silly even to entertain that measure of meaningless doubt. To claim that there is any doubt at all even in theory, however meaningless in practical terms, that Japan attacked Pearl Harbor is empiricism gone mad, in my view.
Carrier has picked up with the probabilistic and empiricist errors that began to be discarded from certain areas in the social sciences decades ago. He appears not to have been aware of the logical rebuttal of his attempt to equate science and social sciences and history as is so devastatingly made clear in the book by the Willers I quoted. His book on Proving History demonstrates no awareness of the historical debates relating to the nature of historical knowledge and reasoning among historians since the 1960s – he even lumps a score or more historians who differ wildly in philosophies and outlooks from each other into one lump to say they all can be said to have their methods reduced to Bayesian thinking.
I feel it would be a pity to finish just by your saying “It appears to me that you have misunderstood — or simply disagree with — my critique of the use of Bayes to assess probability of historical events.” after I have spent so much time on trying to explain you where I can see (what in my, and obviously also Carrier’s, opinion is) a fatal problem of your critique. (I can see this again when you say “… probability cannot be used to predict the next die roll or result of a button press.” as if this was relevant to the problem we are discussing …)
Unfortunately I cannot be much reactive this week, but I try at least to hopefully remove your recurring theme, captured by your saying “If I say I have a 0.000000000000000000000000001 doubt that Japan attacked Pearl Harbor …”, as if Carrier was doing anything like this.
But ok, if you find this so crucial for some reason, then please explain to me why the following hypothesis has an absolutely zero (not virtually zero or so, but -absolutely- zero) probability:
in near future we, the mankind, will be contacted by an extraterrestrial civilization who will explain to us how they were influencing our history and how they arranged an illusion among us that it was Japan who attacked Pearl Harbor …
So Neil, show me, please, the logical impossibility of this hypothesis.
If I understand you correctly you see my position as inconsistent: on the one hand I say I agree with using Bayes and on the other I pull up short and don’t use Bayes in certain situations. I agree with using Bayes for probability statements where probability is valid; but I do not believe that it is valid to declare that because a six sided die has rolled 1 on the first two throws that it has less of a probability of rolling a 1 on the third throw to come. That is not a valid application of probability or Bayes, for that matter. You may say that “my feeling” is more certain that the third roll will be anything but a 1 but that is irrelevant.
(How I feel — or what I anticipate — about finding this or that in the sources is not the point of history.)
That you do not see its relevance to my argument indicates to me that you do not understand my argument or critique of Carrier. We are talking past each other.
I have tried to point out — with quotes from his Proving History — that Carrier does indeed mean this. He says EVERY claim in history is a probability. No exceptions. It is important to Carrier that there is some room for doubt — even if infinitesimally small — to enable Carrier to place his notion of historical claims on the same level as his notion of scientific claims.
I do not understand the relevance of your question to my critique. Again I have to think I have not explained my point to you at all. What you are asking is a theoretical logical possibility and nothing at all of the kind of data a historian works with. Of course your imaginary scenario is logically possible. Why do you think I doubt it to be logically possible? How does that relate to my argument or critique of Carrier or the right use of Bayes?
That Japan attacked Pearl Harbor is not an imaginary mind-game of which we can have an unlimited number. It is reality. It is a once off. Never repeated or repeatable.
If aliens came to us in the future and said they engineered the attack, okay — let them do so. But that has nothing to do with the fact that the attack happened in history.
If we want to experience the situation “aha, now I understand what you are saying” (and I trust we both would want this), then I cannot see another way than clarifying the situation step by step. So I start with just two questions now:
1/ You say “I agree with using Bayes for probability statements where probability is valid”.
E.g., from your viewpoint, is the Bayes calculation here
https://tomrocksmaths.com/2021/08/31/bayes-theorem-and-disease-testing/
valid (though it processes just -one- contingent event, the result of a test)?
If yes, and you still find my Bayes calculations in my A:B mind experiment erroneous, then explain me the difference, please (in particular where I am doing errors).
2/ Your quote continues:
“but I do not believe that it is valid to declare that because a six sided die has rolled 1 on the first two throws that it has less of a probability of rolling a 1 on the third throw to come. That is not a valid application of probability or Bayes, for that matter.”
Indeed, I agree with you. But the fact that you use this correct claim in this discussion only makes sense if you think that I thought differently. What in my text has lead you to thinking this?
I imagine every valid use of Bayes is applied to a specific event. One sees this particularly in the book I referred to, The Theory That Would Not Die: how to find a lost submarine, for example.
The difference between the use of Bayes in the link you provided above and the scenario of yours that we have been discussing (a very limited number of pushing a button to test if a machine is configured to generate one of two possible results) is that the probabilities factored in to the linked test are based on thousands of other results. In the situation of your case we are asked to assign the probability of one of two possible theories being true on the basis of just a handful of results.
In my opening post of this series I included a side-box outlining the origin and basics of the Bayes’ theorem. That account was directed at determining the probability of one event, a single event — calculating the likelihood that a billiard ball was at a certain position on the table. By a “contingent” event I mean an event that can only be explained by distinctive factors that collectively apply to no other event. That the billiard ball could be in a certain position as the result of a random throw is not a “contingent” event in the sense I have meant the term; nor is the event that a person might have a particular disease a “contingent” event in the example of the application of Bayes in the link you provide.
Contingent events are not measurable in terms of probability theory. And the mind-experiment example you have been offering is not an expression of how probability theory works in the real world for Bayes — Bayes works in situations where the probabilities that have been determined by essentially endless of comparable results. It is these very probable probabilities that are used for the valid application of Bayes — the probabilities of a vaccine having an effect are determined by measuring the results of tests on thousands. Finding a lost sub is determined by factoring in probabilities that have been concluded as a result of many years of experience with measuring the effects of tides and relative densities, etc.
Carrier has attempted to justify his use of Bayes in assigning a particular reference class to Jesus by claiming that 14(!!) events justify this, and not only 14, but half of that number that meet 10% of the points on a list that its author said is of random length! The author of that list further said that there is no correlation or cause-effect relationship between those 14 figures and questions of their historicity and explained why by pointing out how the mythical tales of those 14 originated. It is reasonable to conclude, therefore, that Carrier’s use of Bayes in his opening situation of the reference class for Jesus has no comparison with the use of Bayes in the situation described in the page you linked to.
I appreciate your patience, Neil. So let me hopefully finally clarify at least the following point. Consider the situations 1 and 2:
1/ John (J) comes to the medicine doctor (M). M gets an idea to check if J has a rare disease D.
Let A denote the hypothesis “J has D”, and let B denote the hypothesis “J does not have D”.
Given his background knowledge (based on reading the relevant statistics), M knows that the odds A:B are 1:99; we denote this as P(A)=0.01 and P(B)=0.99.
A test T for testing D is applied to J. M knows the statistics, so he knows that if J has D, then the probability of the event E denoting “test is positive” is 0.95; we thus write P(E|A)=0.95.
M also knows that even if J does not have D, then the test might be positive, since P(E|B)=0.04.
Suppose E happens (J is tested positively). For M the situation of assessing the odds A:B clearly changes, but can we quantify this change? Yes, we can, by the Bayes theorem:
P(A|E) = P(A) x P(E|A) / P(E) where P(E) = P(A) x P(E|A) + P(B) x P(E|B).
(I will not do the numerical calculations, but the web-page I referred to gives a concrete example.)
The posterior probability P(A|E) becomes the prior probability P(A) for the next step (M can perform another test or so …).
2/ Recall my mind experiment (in fact, an analogue of standard textbook examples on probabilistic algorithms) where we had
P(A)=0.5, P(B)=0.5, P(E|A)=1, P(E|B)=0.5 (and E was the event “a button-press showed 1”).
We are a neutral judge, so our starting odds A:B are now 1:1, but will they change after E occurs?
Of course, Bayes theorem now says P(A|E)=2/3 (and P(B|E)=1/3), hence we proceed with new odds 2:1.
The only difference between the cases 1 and 2 is that we do not need any statistics to get P(E|A) and P(E|B) in 2, and we use just the starting neutrality for P(A) and P(B).
Neil, you seem to say that because of this difference you “believe in Bayes in the case 1” but you do not believe in Bayes in the case 2 (where you said that my calculations are erroneous, without giving any valid argument).
Can you now understand why I find your reasoning inconsistent?
I think it is you who have the invalid argument and fail to appreciate how probability works meaningfully in the real world.
Ι did not say your calculations are in error.
I have been to trying to explain that your calculations are irrelevant for purposes of making a conclusion about whether hypothesis A or B is correct in the case of your machine. Yes, I can say that “So far, after 3 throws, it is beginning to look like result such and such” — BUT I also know in my mind that that gut feeling is irrelevant and I need to do more than 3 button presses to test a theory.
To decide if your machine is programmed to produce a 1 or a 0 randomly we cannot rely on the evidence of just a handful of button presses. We need the same kinds of data results that informed the doctor about the validity of a test for deciding whether John has a particular disease or not. We need to have the information that comes from many test results.
To make a decision after three button presses is an invalid use or application of Bayes. Bayes was never designed to lead to a conclusion after such limited data. It would be like Thomas Bayes himself having to decide where the billiard ball is on the table after being allowed only three guesses. He would say, no, that’s not what this theory is about.
Historical events have too many unique arrangements of factors going into them that they are always unpredictable — except in hindsight! Carrier cannot “predict” from any Bayesian analysis what evidence he will find — his account where he claims he is predicting evidence of a certain kind is nothing more than a description of the evidence he knows to have existed from the beginning. And it was on the basis of that kind of evidence that the hypothesis of mythicism was formed. So it is simply circular to turn around and say that his hypothesis “predicted” the kinds of evidence he finds.
I think that the following quote from your text demonstrates the crux of your misunderstanding of Bayes:
“It would be like Thomas Bayes himself having to decide where the billiard ball is on the table after being allowed only three guesses. He would say, no, that’s not what this theory is about.”
In fact, Bayes would give you quite a different answer: I started with a uniform probability distribution (PD) of the (guessed) position of the ball. After each new input (left, or right), I recalculate the current PD. So after three inputs, the current distribution is such and such.
(I can recommend you to look e.g. here
https://dosreislab.github.io/2019/01/27/ballntable.html
to get an idea. If you cannot install and run the R program, you can at least look at the figure there, after 30 inputs in one concrete case.)
In fact, previously I was presenting you an analogue of the following simpler version of the one-dimensional Bayes’ ball task: The unknown position of the initial ball can be viewed as a real number in the interval [0,1] (0 means the left margin of the table and 1 the right margin).
Suppose person A tells Bayes that the ball is at position 0 (left margin), but B opposes and says that it is at position 0.5 (in the middle).
Bayes should be now expressing his relative confidence in A and B, starting with the distribution P(A)=0.5, P(B)=0.5, which we can express as the ratio 1:1. Now, step-by-step, Bayes repeatedly generates a random number r from [0,1] and learns, from a trustful authority who can see the actual position of the ball, if r is smaller or bigger than this actual position.
I have already shown you how Bayes would be recalculating his confidence ratio (i.e., the respective probability distribution) after each step in a sequence of 10 answers “bigger”. (The posterior P(A) would be in this case bigger than 0.999, and P(B) smaller than 0.001).
Neil, if you do not understand this explanation, then I suggest to finish our discussion.
You keep avoiding my point: Yes, Bayes would make an estimate closer after each piece of new information — of course, he would be more confident that the ball is closer to position X than to some other position with each new piece of input. Is that what you are repeatedly telling me? That with each new piece of input we can revise our estimate to something closer to what will turn out to be the correct answer? That IS what Bayes is all about. But Bayes is also about continuing to fuel our estimates with new information until we get a very close answer to our original question.
A doctor would be remiss if attempting to decide if a patient had a disease on the basis of only three test results to inform her about the reliability of a particular vaccine. (If you cannot see the stark difference between the example you linked to and the one you have proposed to me about having three button presses on a machine, I might suggest you are the one who does not understand the correct application of Bayes or probability in general.)
But I suggest you try to understand my point in return. We have no repeatable tests to apply to any particular historical question that is attempting to answer: Did Plato or Socrates exist, for example? I have stated the reasons. Do you understand my argument? If so, tell me why you believe it is in error.
>Neil, if you do not understand this explanation, then I suggest to finish our discussion.
You are the one trying to persuade me. If you wish to show me that I am wrong and why, you ought to sum up my argument in your own words as clearly as you can. If I can agree to what you are saying is my argument, then you have the way forward to clarifying where I am in error.
Up until now I feel that your replies have each been missing the key point I have been trying to make and you have simply repeated your own argument as if I am a bit dim-witted for not agreeing to it. Can you sum up in word my argument?
Please don’t reply with maths. I do not dispute your maths or their results of probabilities of confidence, etc. To repeat the maths each time is missing my point entirely. I agree with your maths. I disagree with attempting to justify a final conclusion after a limited number of inputs. I disagree with their application to deciding whether Socrates lived, for example.
Neil, your attitude captured by the quote
“You are the one trying to persuade me. If you wish to show me that I am wrong and why, you ought to sum up my argument in your own words as clearly as you can. If I can agree to what you are saying is my argument, then you have the way forward to clarifying where I am in error.”
can hardly lead to any progress. (You might see this by reformulating your quote so that you are the one who tries to “persuade” Richard Carrier to believe that you have exposed significant errors in his peer reviewed academic work.)
I just felt a pity when seeing that you have clearly not understood what Carrier is doing, and you seemed to not have tried to clarify this with Carrier before making your long critical treatise public. I really thought that I could at least somehow influence you before you finish your 4-part treatise. But I have obviously failed; such is life. I can still add a few comments in case you wish to continue (but I have surely no “wish to persuade you” to anything).
Regarding my maths: I have tried to first concentrate on showing that applying Bayes does not need to have “countless button presses” at our disposal, etc (which implicitly followed from your texts).
Bayes ingeniously showed that we do not need the evidence of “countless” random results (left or right) to show a small interval where the ball is with probability at least 0.95 or so; just having a few dozens of pieces of such an evidence is sufficient. I also showed you an example (A:B) where we can declare the winner with confidence at least 0.95 already when knowing the results of 5 button presses.
You first declared my calculations as erroneous, but later you started to agree with them, with no explanation how this dramatic change of your opinion has occurred. (How can I then sum up your argument?)
Regarding your question on Socrates. Of course, we should first very clearly specify what precisely we mean by the claim “Socrates lived” and then start to weight all the evidence we have. If we can reasonably come only to the conclusion that, based on the evidence we have, the odds are at best 2:1 for this claim being true, then let it be so; we surely cannot say any YES/NO answer with sufficient confidence.
If the justice system comes to a conclusion that the crime was made, beyond reasonable doubt, either by A or by B, but the evidence we have can show the odds A:B only as 8:1 or so, then both A and B should be let free …
(Btw, it was surely some evidence that made A to be one of the suspects but there is no circularity when this evidence is used in assessing the odds A:B.)
I finish by a Jesus example. If I know nothing about the issue, and A tells me that Jesus was a mythical/literary figure, and B tells me that Jesus was a historical person deified by his disciples and later followers, then I have no reason to incline towards any of A or B.
If I then learn that a few decades after this historical man is supposed to have lived, his followers write and believe that he was born from a virgin, then I surely make a slight shift towards A (given my background knowledge, it is no surprise for me that somebody writes about the mythical, literary, figure that he was born from a virgin [I could even “predict” this sort of thing], but it is a bit surprising [not so expected] for me that the reality of the background, family etc., of this supposed historical man did not prevent such claim a few decades later). Bayes would made a similar slight shift towards the right margin if he learns only one result “left”.
Surely it is standard process to demonstrate that one has understood an opposing position before attempting to refute it. If you cannot explain my argument then I suggest you do not understand what you think you are opposing.
Carrier explicitly states that he has no room for discussion with anyone who does not accept his premises — I quoted him here to that effect — so I see no scope for discussing my disagreements with Carrier.
You keep saying that I have not understood Carrier but you have not demonstrated your claim. I have cited Carrier by specific page references to point out I do know exactly what Carrier is arguing. I don’t think you grasp the significance of his insistence that literally every historical claim is a probability claim. Every one — no exceptions.
I used to publicly support Carrier’s use of Bayes until I began to think more deeply about the nature of historical knowledge and research. Yes, I do agree with using Bayes to assess a hypothesis, and to the extent that Jesus was not historical is a hypothesis Bayes is fine — a valid tool. But Carrier goes beyond that.
If you have no wish to persuade me I am wrong then I don’t know why you posted here. But if you want to show me where I am wrong can you please outline what you believe my criticism of Carrier’s method is? Certainly it is not that I have misunderstood how Carrier uses Bayes to argue for the unlikelihood of Jesus’ existence. I understand very well how he has used Bayes to that end. Have I said anything in particular that gives anyone cause to think otherwise?
Here is where you seem to be simply ignoring my argument. Bayes can be used any number of ways but it cannot be used to determine the historicity of what happened. Please try to respond to my replies to you. The calculations you make are simply irrelevant to the determination of historical events.
That is not how historians decide Socrates existed or how they decide any historical event. Nor should they use probability to decide basic historical events, thereby reducing every historical event to a probability statement. That is simply not true of historical events — unless we want to posit that we can know nothing for sure about the world or ourselves. Carrier has simply not demonstrated any awareness of what historians say about their basic “historical events” or the nature of historical knowledge. He is, in effect, in a positivist empiricist camp of historians that went out of fashion in the first half of the last century. (The difference between Carrier and them is more semantic than real.)
That is all very fine. If you do a search on this blog for Bayes you will find quite a few posts where I have presented arguments explaining it and applying it. I am well aware of the process you are describing. But what you are describing is not history. It is a mind-game. Fair enough — it’s fun to do. But it’s not how historians work.
There is a much simpler rule by which we can confront the Jesus figure in history. That is: there is no evidence for his historical existence. Full stop. Therefore it is pointless to try to discuss him as a historical enterprise.
We have theological writings about Jesus that are not independently testified until well into the second century. We have a statement in Josephus that is questionable but worse, cannot be said to have been written until at least 2 generations after said Jesus. No historian gives a second glance at “evidence” like that for events of the early first century unless there are independent contemporary confirmations of some kind.
Carrier, as you know, responded to these posts of mine by saying that I was claiming he had predicted the exact form of Clement’s letter. I tried to point out to him that I had not. I fully understood that he claimed that his hypothesis predicted that type of evidence — not the specific form of the evidence. I also pointed out that a hypothesis that was built on that kind of evidence cannot be said to predict that kind of evidence.
I have posted over 50 times on Bayes Theorem — if you skim through some of these posts I think you will surely agree that I do have a fundamental understanding of how Bayes’ Theorem works. But probability has a limitation when it comes to human studies — both the social sciences and history. Again, check out the Willer and Willer explanation.
Now I am reacting just to your
“If you have no wish to persuade me I am wrong then I don’t know why you posted here.”
I would characterize my motivation as an effort to understand more precisely what is going on here. It is true that as far as I understand (or do not understand) you so far, I think that you have misinterpreted Carrier (to which he could surely “help” by his usage of the terminology “prediction, expectation, …” etc.), but I cannot exclude that I will change my mind after a clarifying discussion. So I still plan to write a sort of a summary of my understanding here, but in an “unpredictable :-)” future (unless you discourage me to do so).
I would further greatly appreciate a clarification on what you think my argument and disagreement with Carrier actually is. In what way, specifically, do you believe I have misunderstood Carrier? If you look at some of my early posts on Carrier and Bayes Theorem I am sure you must accept that at least in the past I have had a clear understanding of both. In comments I have attempted to refute others who have attempted to argue Carrier’s actual process of using Bayes was wrong. Carrier certainly uses Bayesian reasoning and probability assessments correctly in themselves. But it is his attempt to apply them to historical claims generally (not just hypotheses) that is problematic — and that is far from being just my view: it is the view that has been taken more generally among the social sciences since the 1960s.
What would you think about a theory of Christianity emerging out of the Idumeans?
1. The Idumeans disappear from history after AD 70.
2. An Idumean Temple are Maresha has a depiction of Kaus that kind of resembles a Cross.
3. Talmudic Traditions associated Christianity with Edom.
4. Josephus list of the four leader of the Idumeans in Jerusalem during the Revolt include a Simon and a pair of brother named Jacob(James) and John.
5. Revelation 1 described Jesus hair with a word that means Goat Fur also used of Esau in the LXX.
6. Forms of of the Greek word Petros becoming a name for a leader of the community could come from Petra.
7. Esa the name for Jesus in the Qurran seems more related to Esau then it does Yeshua.
8. The “brothers of the Lord” seem to be named after Jacob and three of Jacob’s sons. The author of Galatians refers to his more Jewish rivals as “they of Jacob”.
9. After Hadrian banned Jews from living not just in Jerusalem but anywhere Jerusalem was visible from, it could have been largely Idumeans who settled in those areas becoming the genealogical ancestors of the Palestinian Christians who happen to mostly be in that same region.
I sometimes come across ideas comparable to this with lists of items that point to some parallel in the New Testament. They are not “theories” but speculations based on disconnected items. Someone has sought out as many items to compare as possible, it seems, without any interest in pausing to think them through and how they could really have come together and “caused” Christianity.
The same process was at work in listing many points to spread the “theory” that Paul McCartney was dead back in the 1960s, or the notion that something mysterious was behind the assassinations of Kennedy and Lincoln supposedly because of many parallels, and so forth.
To make the parallels meaningful one would have to explain how they actually came to contribute to Christianity. I can’t imagine how the place Petra could have been linked to a leading disciple named Peter, for instance. Why? What would be the point? And what reason do we have to think there was an association?
Others draw lists of parallels between the gospels and astrology.
One can always find some kinds of parallels if one looks hard enough. It’s called confirmation bias at work — only looking for details that add to a list of speculative points.