My review of Richard Carrier's "On the Historicity of Jesus"

[iskander]

Gday iskander,

Thanks for your reply

http://kapyong.5gbfree.com/Carrier/Elements.html

Element 1. The earliest form of Christianity definitely known to us originated as a Jewish sect in the region of Syria-Palestine.

Element 1 is the explanation of the origins of Christianity. What is the earliest form of Christianity definitely known to us?

I can ask this question in your website, away from the noise, if you prefer that.

Let's discuss here, where others can contribute.
(Seems like this is the premier place for Mythical Jesus discussion.)

Firstly, here is Dr Carrier's section on Element 1, in its entirety :

The earliest form of Christianity definitely known to us originated as a Jewish sect in the region of Syria-Palestine in the first centuries CE. Some historians would challenge this, but their theories have yet to survive peer review or persuade anything near a consensus agreement among experts. Rather than prove it true here, I will simply state it as a given fact of our background knowledge, to be revised only if it clearly disproved.

I'd say Paul is the earliest form of Christianity known to us, (not sure whether I would classify the later Gospels as part of that same earliest form, or a new form)

Anyway -
It seems clear that it originated as a Jewish sect, from Paul's interest and arguments about Jewish vs gentile customs, and the dependence on the Tanakh in many early works.

Did it originate in Syria-Palestine ?
On the face of it - yes.

(Although by the time of G.Mark it seems to have spread quite far already, considering his poor knowledge of the region.)

I don't think Dr Carrier has gone out on a limb here.


Kapyong

Thank you, Kapyong.It is a very good ' element' on which to build on as many important decisions have already been taken . The Inquisition used similar 'elements ' to destroy with.Thanks again.

[timhendrix]

Peter suggested I did a tl;dr version of the argument on OHJ so here goes:

Aim: To examine the argument for the conclusion there is a chance of roughly 1/12000 Jesus did not exist. In particular, how well it conforms to Bayesian probability theory which is what Carrier use to supports the argument.

Outcome: I think there are concrete reasons to believe the argument in OHJ does not work. This is not the same as saying Jesus existed, or another argument could not take it's place

Preliminary remark 1: Outline of the argument
What OHJ does is that it expresses the probability of Carriers myth-theory in terms of other probabilities. Bayes theorem suggests we express the probability of (for instance) historicity in terms of other probabilities which can be classified into two groups: How well our hypothesis of myth (and historicity) explains the evidence and how a-priori plausibly our hypothesis of myth (or historicity) is. Bayes theorem is a consistency requirement between probabilities and not a mean of estimating probabilities which has to be done using external procedures (in OHJ, primarily Carriers expert guess).

This grouping is quite natural. Imagine a trial where we consider guilt vs. non-guilt. Before we consider the evidence, we have an a-priori idea of the accused persons guilt based on general considerations (for instance if he is a career criminal or an 88 year old monk), and then we consider how plausible the evidence is given an assumption the accused is guilty and that he is not. Bayes theorem just express this procedure in a more formal way and using probabilities.

Preliminary remark 2: Stability of Carriers computation
Carriers computation ends up containing about 25 terms that are individually argued in OHJ. One such term could be "Given our general background information and that Carriers theory of historicity is true, how plausible (0-100%) is the Gospels?". Naturally, we can't expect to guess these probabilities completely accurately and so we should consider how inaccuracy will affect our final conclusion. I argue two things:

1) As a rule of thumb, if we have a given level on uncertainty on each of these probabilities (say 5%), then the level of uncertainty on the final conclusion (the probability Jesus existed) will be about 5 times larger.

2) If we are ever so slightly biased this will affect the computation greatly. Suppose a person consistently over-evaluate the evidence in favor of historicity. For instance, if we believe: "The probabilities of the Gospels given the hypothesis of myth" was 20%, he would say it was 5% lower i.e. 19%, then this 'error' (or from the other persons perspective, more realistic evaluation of the evidence) will be magnified by roughly a factor 20(!).

Taken together we have to assume two things:
From 1, that our ability to judge the probabilities used in the computation is about 5 times better than our ability to simply judge "the probability Jesus existed given the evidence".
From 2, that we have a vanishing small bias and, perhaps more critically, that we don't introduce a source of bias accidentally (more on this later)


Preliminary remark 3: Slim vs. fat theories
We have properly all tried to arguing with someone who believes in a conspiracy theory (such as that the moon landings are fake or that 9/11 was an insider job by the US government) and been frustrated of the way evidence seemingly has no effects on his or her beliefs. This behaviour is actually mostly rational and Bayes theorem explains it.

Common for all good conspiracies is that they involve a powerful agent with partly obscure motives. If you present them with a piece of evidence, it is therefore very easy to "explain" why that evidence looks the way it does. This is saying the evidence has a high probability given the conspiracy theory is true. The cost of this "easy explanation" is that conspiracy theories should be regarded as intrinsically implausible (i.e. Occams razor), and this is where people who believe in conspiracies tend to go wrong.
To phrase this in Bayesian terms: For any given hypothesis (i.e. 'a person is not guilty' or '9/11 was not carried out by terrorists'), we can add elements to the minimal hypothesis ("the person is not guilty AND someone planted is DNA on the crimescene" or "9/11 was not carried out by terrorists AND the planes were radio-controlled by the CIA") which makes the evidence more probable. This complexity comes at the cost of lowering the prior probability. In my experience, a good conspiracy crank up the complexity and then 'forgets' (or gives up) on evaluating it's plausibility correctly.

Remark 1: Notice that this 'added complexity' potentially affects all terms in the Bayesian computation, hence complexity which is not accounted for can bias the entire computation in the sense I introduced earlier
Remark 2: Therefore, if we suspect a hypothesis is complex, we absolutely must be careful when evaluating it's probability. We cannot say: Lets just give it a prior of 50% or hand-weave the prior away (hey, it's only the prior, let's discuss how the hypothesis explains the evidence) and expect this won't have a potentially huge effect.

Preliminary remark 4: Carriers hypothesis of myth and historicity
OHJ is often characterized as computing the probability of myth vs historicity. This is not true. As Carrier explains, it examines the probability of Carriers hypothesis of myth, ~h, and a hypothesis of historicity h. Despite the notation these are not binary propositions (it is not true that ~(~h) = h). Focusing on myth, ~h, Carrier defines it as the conjunction of 5 propositions:

• At the origin of Christianity, Jesus Christ was thought to be a celestial deity much like any other.
• Like many other celestial deities, this Jesus 'communicated' with his subjects only through dreams, visions and other forms of divine inspiration (such as prophecy, past and present).
• Like some other celestial deities, this Jesus was originally believed to have endured an ordeal of incarnation, death, burial and resurrection in a supernatural realm.
• As for many other celestial deities, an allegorical story of this same Jesus was then composed and told within the sacred community, which placed him on earth, in history, as a divine man, with an earthly family, companions, and enemies, complete with deeds and sayings, and an earthly depiction of his ordeals.
• Subsequent communities of worshipers believed (or at least taught) that this invented sacred story was real (and either not allegorical or only 'additionally' allegorical)

This is not wrong by any means, but it must be acknowledged that we cannot simply treat ~h as "Jesus did not exist" but we must treat it as the conjunction of these 5 propositions.


The argument:

Part 1: The hypothesis of myth is 'complex':
I think we are justified in saying the hypothesis of myth ~h is complex. Firstly, adding these terms arguably makes it easier to explain the evidence. A single example: If we simply assume the early Christians at some point did not think Jesus existed but was a celestial being (bare myth), they have to do a 180 turn to arrive at a widely taught "historical Jesus" tradition. However if we *assume* they came to take such a turn (as is implied in ~h), *then* we can easily explain such a tradition

Secondly, I suppose there are thousands of communities that once worshipped thousands of celestial beings. However very few later came to satisfy all elements of ~h. This alone should indicate that it is a fairly rare (that is, unlikely) process.

An obvious point to bring up (which Carrier does!) is that the above argument fails to take a lot of information specific to early christianity and the classical world into account. I agree!. So lets just make that a point worthy of notice:

Part 2: When computing the prior probability of ~h, we must take into account how ~h is defined as well as our background information
Remark: If we fail to do this, we fail to do a Bayesian analysis.

Part 3: The prior
When Carrier computes the prior probability of ~h, he does so by an approximation based on the reference class of Rank-Raglan heros. When a reference class is used to approximate a probability of the form P(A|B) this would typically be done with an expression of the form:

P(A|B) = (elements that matches A and B) / (Elements that matches B)

This type of definition would however imply that we should compute (where RR stands for "Matches at least half of the Rank-Raglan hero criteria"):

p(~h | RR) = (1 + Members of the RR hero class who are ~h, i.e. died and were raised in the supernatural realm etc.)/(2 + Members of the RR hero class)

(the 1 and 2 comes from the rule of succession which Carrier invokes). The computation Carrier actually uses is:

p(~h | RR) = (1 + Members of the RR hero class who are not historical)/(2 + Members of the RR hero class)

There are many problems here. Firstly, that I don't think the members of the RR hero class matches ~h exactly, and so the real prior probability can be expected to be much lower on these grounds alone.

A second problem is that instead of considering all background information, i.e. p(~h|b), we are now only considering b as being composed (or effectively composed?) of RR. This move must be justified in and by itself. There are two ways of doing this: Either we re-write the probability expression to contain the term p(~h|RR) using the rules of probability theory, however in this case the resulting expression becomes much, much more complicated and they are not the expressions used in OHJ.

Alternatively, we argue that b (which would appear to include information about the timing of the composition of the various books) can be reduced to only RR. I find this later move very difficult to accept and the discussion on the two threads Andrew Criddle posted highlights issues that are simply being ignored by such a computation.

The fundamental problem at work is the unsuitability of reference classes as a means to approximate the probability of a complicated expression such as ~h given complicated (and specific!) background information such as b using finite frequencies. The underlying problem here is known as the "reference class problem", and I simply think Carrier fails to take into account how dangerous such a move potentially is.

Conclusion:
Since the hypothesis ~h is complex it's prior probability is important (part 1). We have also assumed the background information contains relevant and important information (part 2). Both of these factors must be taken into account when the prior is computed, however rather than doing this Carrier uses an approximation which archives neither of these goals (part 3). Improperly accounting for a complex hypothesis can (and likely will) bias the entire computation (preliminary remark 3) and even a low level of bias can completely overturn the result (preliminary remark 2). The conclusion of OHJ is therefore not supported.

Part 4: Does this matter?
The obvious problem to ask is how big these problems are. Since we are computing probabilities of an event which we can't know took place or not (Jesus existed) we can't hope to fully falsify the conclusion, however if we can show the computation and the method used to obtain it renders paradoxical results this at least would show something has gone wrong.

Inconsistency 1:
Suppose the only evidence we had about the early church was the Gospels (i.e. no Paul, no external evidence, no Philo, etc.). If we assume the probabilities in OHJ are true then it follows that the probability of ~h given only the Gospels (and general background information) is 93.75%.
This is saying that given only the Gospels we can conclude with near certainty that his followers must have believed Jesus death took place in the supernatural realm and that Jesus did not live on earth. I am happy to be proven wrong, but I think this conclusion can be safely said to be false and so we can conclude that one of the premises in Carriers computation must be false too. This is the first thing I would ask someone who accepts the argument in OHJ: Given only the Gospels and b, does it seem reasonable there is no less than 93.75% chance that Jesus was originally believed to have lived and died in the supernatural realm?.



Inconsistency 2:
What happens if we change reference class? According to Carrier, if we change reference class then Bayes theorem will ensure we arrive at consistent results. My take on this is that it is true *as an ideal statement about unknown probabilities* (see my response to Peter), but when we use crude approximations using reference classes such as Carrier does this will not hold. To take the example I mentioned to Andrew Criddle, suppose we use the class:

L: "is a character found in literature".

Then the computation becomes:

P(~h|b="is a character found in literature") = (Literary characters which satisfy ~h) / (Literary characters)

The class of literary characters includes Zeus, Moses, Julius Caesar, Robin hood, Goofy, Winston Churchill, Mickey-Mouse, etc. etc. Then we can ask how many of these matches ~h. ~h includes that there are "followers" of the character that believe or teach a death and a burial in a supernatural realm (This would include some ancient Gods and with a stretch of imagination also Aslan the Lion), but ~h also includes that subsequent communities came to believe the characters had had an earthly existence with companions, saying, etc (presumably we lost Aslan the lion and most other characters). The bottom line is that very, very few will match ~h whereas the total number of literary characters will be very large and so we get an extremely low prior probability. So how exactly is the information supposed to be accounted for elsewhere? (and how is something similar done in OHJ?).

Very good arguments can (and should!) be brought against this computation: Most obviously, it is at best approximating p(~h|L) and not the true prior p(~h|b). However these and similar arguments will also apply to the computation in OHJ. This is not consistent. I mention two other examples in my review and discuss Carriers arguments for his reference-class method based on the examples.

[andrewcriddle]

About the possibility B (a mythical Jesus on the Earth and not in sublunar realm) that would be the view that Paul had (according to Roger Parvus).
If I remember well, in a brief note of OHJ (I go to memory) Carrier recognizes that possibility as a specific sub-option of the possibility named C by Andrew.
But maybe Andrew means as B the possibility that a mythical Jesus lived an ENTIRE life on Earth and was considered historical. Correct?

I think you are understanding me correctly.
B is the possibility that Jesus was originally considered to be a historical figure living and dying on earth although in reality there was no such historical Jesus.

Andrew Criddle

[GakuseiDon]

About the possibility B (a mythical Jesus on the Earth and not in sublunar realm) that would be the view that Paul had (according to Roger Parvus).
If I remember well, in a brief note of OHJ (I go to memory) Carrier recognizes that possibility as a specific sub-option of the possibility named C by Andrew.
But maybe Andrew means as B the possibility that a mythical Jesus lived an ENTIRE life on Earth and was considered historical. Correct?

I think you are understanding me correctly.
B is the possibility that Jesus was originally considered to be a historical figure living and dying on earth although in reality there was no such historical Jesus.

That is along the lines of GA Wells' theory. I think a case could be made that, using Dr Carrier's odds alone in OHJ, that the case for GA Wells' ahistorical Jesus is stronger than that of Carrier's.

Wells' theory is that Paul believed in a historical Jesus, but one that died a century or more before Paul wrote. That explains why Paul didn't include any historical details about Jesus, because by the time of Paul such details had been forgotten.

I give a summary of the odds in Section 5 of my review of Carrier's "On the Historicity of Jesus" here: http://members.optusnet.com.au/gakuseid ... l#Section5

Carrier breaks the odds down into four sections:
1. Extrabiblical evidence
2. Evidence of Acts of the Apostles
3. Evidence of the Gospels
4. Evidence of the Epistles

In nearly every case, the odds for Wells' theory would appear to not differ from the odds for Carrier's theory. But Wells' theory would be superior to Carrier's (according to the odds provided by Carrier in OHJ) when it comes to "made from sperm" (2/1) and "made from a woman" (2/1). So Wells' theory is at least four times more likely than Carrier's, as "best case".

[MrMacSon]

B is the possibility that Jesus was originally considered to be a historical figure living and dying on earth although in reality there was no such historical Jesus.

That is along the lines of GA Wells' theory ...

Wells' theory is that Paul believed in a historical Jesus, but one that died a century or more before Paul wrote ...

Well's current theory or a previous one?? There has been some commentary that Wells has gone from being 'ahistorical' to being 'historical' (?)

[GakuseiDon]

B is the possibility that Jesus was originally considered to be a historical figure living and dying on earth although in reality there was no such historical Jesus.

That is along the lines of GA Wells' theory ...

Wells' theory is that Paul believed in a historical Jesus, but one that died a century or more before Paul wrote ...

Well's current theory or a previous one?? There has been some commentary that Wells has gone from being 'ahistorical' to being 'historical' (?)

AFAIK Wells' current theory is that Paul's Jesus lived a few centuries in the past, and the Gospels may have been influenced by an itinerant preacher who lived in the early First Century, with no connection between the two. From here: http://infidels.org/library/modern/g_a_ ... liest.html

• I have treated both the Galilean and the Cynic elements less skeptically in The Jesus Myth, allowing that they may contain a core of reminiscences of an itinerant Cynic-type Galilean preacher (who, however, is certainly not to be identified with the Jesus of the earliest Christian documents).

The point I make in my review is that Carrier sees "made from woman" and "made from sperm" as counting towards historicity; but actually they don't. They count towards 'earthliness' (which would be against Carrier's celestial Jesus anyway), but there are many examples of 'earthly' people who were not historical. I think that, even using Carrier's odds, Wells' theory scores higher than Carrier's.

[MrMacSon]

AFAIK Wells' current theory is that Paul's Jesus lived a few centuries in the past, and the Gospels may have been influenced by an itinerant preacher who lived in the early First Century, with no connection between the two. From here: http://infidels.org/library/modern/g_a_ ... liest.html

• I have treated both the Galilean and the Cynic elements less skeptically in The Jesus Myth, allowing that they may contain a core of reminiscences of an itinerant Cynic-type Galilean preacher (who, however, is certainly not to be identified with the Jesus of the earliest Christian documents).

Cheers, G'Don. I know Wells is somewhat off-topic, yet in looking to see if wikipedia encapsulates his position (though that is not where I got my 'impression' from) I came across an interesting point that does seem to align with Carrier -

Since the late 1990s, Wells has said that the hypothetical Q document, which is proposed as a source used in some of the gospels, may "contain a core of reminiscences" of an itinerant Galilean miracle-worker/Cynic-sage type preacher.[2^] This new stance has been interpreted as Wells changing his position to accept the existence of a historical Jesus.[3] In 2003 Wells stated that he now disagrees with Robert M. Price on the information about Jesus being "all mythical".[4] Wells believes that the Jesus of the gospels is obtained by attributing the supernatural traits of the Pauline epistles to the human preacher of Q*.[5]

4 George Albert Wells (2003) Can We Trust the New Testament? ISBN 0812695674 pp. 49–50

5 ibid p.43

* That statement of Wells view about Jesus, ironically, does seem to align with Carrier's view (or aspects thereof; whether Carrier believes in Q or not).

The point I make in my review is that Carrier sees "made from woman" and "made from sperm" as counting towards historicity; but actually they don't.

They don't have to, I agree; but it would seem to depend on the context in which those claims were made.

[Ben C. Smith]

Wells' theory is that Paul believed in a historical Jesus, but one that died a century or more before Paul wrote.

By highlighting the century mark, you are leaning heavily into Alvar Ellegård, I believe. Wells himself — at least in his early, classic stuff (the stuff that influenced me in college) — did not press exactly how long before Paul the figure of Jesus was imagined to have lived: all that mattered was that it was not in recent, contemporaneous or nearly contemporaneous history. Peter Kirby summarizes his view accurately, I think: "Paul, for instance, wrote before any gospel existed, and his Jesus lived on earth as a shadowy figure of the indefinite past."

[Kapyong]

Gday all,

Well, I'm only an amateur minnow here, but I am pleased as punch to have been cited by name in a scholarly paper on Jesus Mythicism.

Thank you for citing me timhendrix.

Yes friends, I was one of those three Amazon reviewers who timhendrix quoted as praising Dr Carrier's use of Bayes' Theorem.

No such thing as bad publicity, right ?


Kapyong

[GakuseiDon]

Wells' theory is that Paul believed in a historical Jesus, but one that died a century or more before Paul wrote.

By highlighting the century mark, you are leaning heavily into Alvar Ellegård, I believe. Wells himself — at least in his early, classic stuff (the stuff that influenced me in college) — did not press exactly how long before Paul the figure of Jesus was imagined to have lived: all that mattered was that it was not in recent, contemporaneous or nearly contemporaneous history. Peter Kirby summarizes his view accurately, I think: "Paul, for instance, wrote before any gospel existed, and his Jesus lived on earth as a shadowy figure of the indefinite past."

Thanks Ben. Yes, you are right. I do remember reading that Wells placed Paul's Jesus a couple of centuries before, but on googling this Wells doesn't say this that I can find. I suspect I got it from Doherty's review of Ellegard, where Doherty writes: http://www.jesuspuzzle.humanists.net/BkrvEll.htm

• Professor Wells has always maintained that this is the way Paul regarded his Christ Jesus, as a heavenly, pre-existent figure who had come to earth at some uncertain point in the past and lived an obscure life, perhaps one or two centuries before his own time.

But even here the emphasis is on an uncertain point in the past.