Why Richard Carrier probably won’t win over the guild, part 3

NYT Office scene

The guild knows the beat it covers.

This post concludes the series on the frosty reception of Dr. Richard Carrier’s book On the Historicity of Jesus among biblical studies academics, “the guild.” The series is not a book review of On the Historicity, but rather a sketch of a different Bayesian-style investigation parallel to Carrier’s.

Previous posts (link and link) have refined Carrier’s hypothesis specifications, considered alternatives to his permissible but unappealing choice of an initial belief state, and checked his claim to have weighted evidence systematically to disfavor his own preferred hypothesis. This post looks at some reservations about Carrier’s interpretations of ancient texts. Finally, a brief conventional Bayesian sensitivity analysis of his results sheds some light on the potential to open up the guild to new foundational ideas.

Step 3: assessing how the evidence alters the initial plausibility, concluded

Carrier’s critics disagree with him about the meaning of some evidence. Some of that discord is “domain specific,” requiring expertise in the subject to referee. For example, like Doherty, Carrier finds much support for mythicism in the Ascension of Isaiah, an ancient religious text which blends Christian and Jewish mysticism. Here, the guild lacks a well-formed consensus about dates and just how the text has been altered through time (link). Until such matters are settled, it is unrealistic for any paradigm-changing arguments based on the text to command the needed weight.

Sometimes a text simply doesn’t say what Carrier says it does. For example, his misreading of Epiphanius’ Panarion in order to “discover” a new sect of ancient Christianity has been discussed on this blog (link).

A more prevalent problem is that some of Carrier’s readings of crucial texts seem strained. For example, what did Paul mean by the phrase the brother(s) of the Lord (1 Corinthians 9:5 and Galatians 1:19)? Plausible readings point to somebody distinguished among Christians, perhaps a kinsman of Jesus or maybe an important apostle (link).

Carrier’s view is that the phrase means only a baptized Christian, conveying no further distinction from any other formally initiated follower of Christ. This raises questions. For example, if the James whom Paul met in Galatians wasn’t somebody worthy of mention, then why does Paul bother to mention him? Why does Paul use the phrase only twice, only when Peter and all the other apostles are also referenced?

Another notorious Carrierism is the cosmic sperm bank (link). Romans 1:3 seems to say that Jesus was a Jewish man, artfully:

… [God’s] son, having come of the seed of David according to the flesh…

But Carrier says that Paul’s idea was that God had stored some of David’s semen and when the time was right, fashioned a body for Jesus from that. But wait – Paul also suggested there was a woman involved, Galatians 4:4,

But in the fullness of time, God sent his son, come from a woman, come under the law

Carrier would correctly point out that the repeated participle come is an unusual word to use for born. Then again, if the woman was divinely inseminated with King David’s sperm, perhaps an unusual word for her resulting pregnancy isn’t completely unexpected. Regardless of where the sperm came from, however, there’s nothing on the page about the woman being in outer space at any time.

Carrier’s idea, for all its outlandishness, isn’t even slightly incompatible with Jesus having had an earthly mother. No more so than God himself being the father, or Jesus having had no father at all, as in Islam. The partial background information includes that there were some early worshipers of Jesus as a god; maybe this is how some of them thought their god found his way to Earth. As for discerning between the two contending hypotheses, so what?

In both the brothers and the David cases, the intended significance of the proposed reading is to defuse a statement by Paul whose face value favors a historical Jesus. Readings that only their inventor could love cannot be expected to defang plainer readings. What would help is to explore how much sense the statements still make when read plainly, but under the assumption that mythicism were true.

Suppose you did know that Jesus was mythical. How surprised would you be to find Paul, the very fountain of figurative fraternal phrases, calling somebody the brother of the Lord? That’s the probative value of that evidence, not how perfectly it fits with historicism. How surprised should you be to learn that a Jewish mythological figure was said to have had a legendary Jewish ancestor and a mythical Jewish mother?

Criticize the guild for not seriously considering how the world would look if they happened to be wrong. Point to Bayes for support on why such an exercise is obligatory. Hammer home how little ease of comprehension differs no matter which contending hypothesis is true, because Paul would likely use the same language either way.

Step 4: sensitivity analysis

Carrier estimated final odds of between 2:1 and 11,999:1 in favor of his preferred mythicist hypothesis. Although the figure 11,999 leaps off the page, it is the 2:1 that has more diagnostic importance. Those same odds, 2:1 in favor of the mythicist hypothesis, were also the upper end of Carrier’s starting odds for his flavor of mythicism, 2:1 up to 15:1.

That the final odds interval includes the initial odds interval means that some arguably reasonable persons might conclude that all the evidence except Rank-Raglan is a wash, favoring neither of the two hypotheses. This inclusion of the initial interval within the final interval would persist even after moderate adjustments to allow for such lapses as neglecting that the Gospels do somewhat support the bare-bones historicist position, or that Epiphanius didn’t write what Carrier misattributes to him.

This outcome is not a quirk of Carrier’s specific analysis. Whenever two parties agree what the background information and evidence are, the source of any disagreement between them must be how they interpret what they know and can observe.

The available evidence and background information in this controversy are especially prone to interpretation. There’s not a lot of either, there’s not likely to be more evidence any time soon, and what there is has been delivered to us in less than pristine condition. That the odds interval widened rather than narrowed as additional evidence was taken into account is symptomatic of an outsized role for purely personal interpretations and estimates of bearing.

There are also difficulties peculiar to when an analyst is nearly certain. Near certainty is what much of the guild feels toward the historical Jesus hypothesis. Near certainty is stubborn, resistant to change by new contrary information, and should be stubborn according to Bayesian theory.

Example The maximum probative value (maximum likelihood ratio) of Carrier’s interpretation of the evidence is about 800:1 in favor of his preferred hypothesis (final maximum odds: 11999:1 divided by initial maximum odds: 15:1 gives a maximum probative value of ~ 800:1).

Suppose a guild member read On the Historicity with an initial belief state 1000:1 favoring the historical Jesus hypothesis (99.9%, tepid compared with Carrier’s nearly 12,000:1 the other way). Our member disagrees with Carrier about Rank-Raglan, and declines to change her own initial confidence. Also, her initial belief state is heavily based on the Gospels, which don’t affect Carrier’s 800:1. But then: epiphany! She buys Carrier’s analysis of all the other evidence without reservation, and adopts the whole of his possible 800:1 probative value. What’s her new state of belief?

She still finds the historical Jesus is more likely than not (1000:1 initially times 1:800 evidence equals 1.25:1 final; still the same direction as before, about 56% confidence).

And what if that guild member merely agreed that Carrier made several good points, and adopted a generous probative value of 100:1 against her initial position? She remains highly confident about a historical Jesus, 90.9% (10:1).

The moral of the example is two-fold: Novel interpretations of the available evidence, the sort of thing Carrier offers:

– are unlikely to flip the guild consensus to mythicism, not even to attain a widespread more-likely-than-not level of confidence,

might measurably reduce the guild’s confidence about a historical Jesus, but only if they were far more impressed by the new arguments than their reception of the work so far.

Finally, there is an impersonal difficulty in the nature of the hypotheses themselves: they “approximate” each other. That is, how different in effects objectively observable after two millennia is

some people worshiped a man as a new divinity soon after he died


some people soon came to understand a new divinity as a recently deceased man?

Neither of the two situations happens often. Which is more likely and by how much, given that some early Christians honored both the mortal humanity and the divinity of their Jesus (as many still do)? What amount and kind of evidence would decisively distinguish one situation from the other?

Is the guild immovable, then, until and unless radically new, powerfully diagnostic evidence is discovered?

Not necessarily. Bayes offers another way forward. Challenge the guild’s initial confidence by exploring a wider range of seriously possible hypotheses. Call their attention to the inability of the existing background information or evidence to discriminate among hypotheses that “approximate” each other. Mere mortals should consider avoiding taking sides with near certainty in such a situation.

The heuristic teaching of the consensus denier cuts both ways. If the available evidence and background information are inadequate to support a foundational hypothesis that has withstood intense scrutiny for centuries, then it probably doesn’t strongly support any single novel alternative either. Somebody would likely have noticed by now.

Which should be fine. Acceptance of uncertainty is the prerequisite for learning. Ideally, learning is what scholarship is about. The critic of complacent consensus has a useful role to play in this affair, but heroically instituting a new paradigm without new evidence is unrealistic in any established field.

Photo credit: Inside the New York Times Newsroom, Marjory Collins, 1942


Leave a comment

Filed under Inference and choice, Knowable historical Jesus

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.