Toto

There is an hour long podcast here that is worthwhile. Carrier talks about Baysian method and also the Argument for the Best Explanation, which he says is actually Baysian.

avi

Thanks, Toto, I still have not perused the last link you provided, shame on me. Sorry to be so lethargic. For years I blamed the poor mosquitoes--> but finally the truth emerged, I had drunk a lot of quinine in Africa (with some other medication mixed in, naturally), so I cannot in fact attribute my progressive mental ineptitude to malarial sequelae.

Is there a text link available, for I am not very fond of "podcasts", though, I did watch War of the Worlds, as a young adult....And, yes, as you would have predicted, I did fall for Orson Welles' trick....

Umm, my humorless hostility to employing Bayesian inferences to identify, establish, or verify probabilities in an environment of impure data stream, is unabated, and unlikely to change, no matter what Dr. Carrier, or anyone else writes.

Has Dr. Carrier performed a simple test to demonstrate the utility and accuracy of employing Bayesian statistics with biblical source material?

He needs some parameter, universally accepted, i.e. some question, the answer to which is found in every extant edition of one of the gospels, with not so much as a change in spelling, between various versions, in order to run such a confirmatory test....

In my (very limited) experience, that's not going to be easy. If the data streams from each of the various editions are in even minor disagreement, it will be impossible, in my view, to draw meaningful conclusions about the particular passage under investigation.

Mathematics is not terribly forgiving, about errors of even modest dimension.

avi

jgoodguy

It is going to be peer reviewed! Yea. Estimated publish date is 12/12.

avi

Did Jesus travel through Sidon, on his way to Lake Galilee, OR, did he travel to Lake Galilee without passing directly through the town of Sidon?

Was Sidon a precise destination for him, on his journey, or was Sidon simply a town proximate to the route on which he had been traveling, a town well known to people throughout the Mediterranean region, a town which serves to identify in a general way, the region through which he had passed, but not an explicit destination, per se.

The answer to this question is not obvious.

The answer depends entirely on which version of Mark 7:31 one reads:

The Byzantine Majority text, in agreement with (both the 1550 and 1894) Textus Receptus present Sidon as a mere place marker, while the Alexandrian text, and the Hort&Westcott version both present Sidon as an explicit locus through which JC passed en route to Lake Galilee.

One can perform statistical analysis til the cows come home, but, it doesn't matter how many peers of the realm review Carrier's text, this issue is not amenable to Bayesian analysis, because the original data lacks uniformity.

Imagine five weather stations, situated adjacent to one another, and two of the five emit data conformant with a prediction of rain, while the other three emit data suggesting sunshine without rain....

Imagine further that the owner of the two stations predicting rain, had fiddled with the output, so that a particular activity, scheduled to meet outdoors, would have reduced attendance....

That's not an environment conducive to performing Bayesian analysis.

Peer review makes sense, if the folks reviewing the study understand the mathematics behind the assertion that meaningful data can emerge from employing this mathematical tool in conducting biblical research.

Thus far, no one on this forum has addressed my objections. Maybe we lack sufficient exposure to mathematics.....

avi

Toto

I don't think I understand your objections, avi. Do you accept the idea that Baysian statistics is a way of dealing with uncertain information?

avi

Thanks Toto.

There are clearly two problems.

1. My writing is obscure;
2. The topic is not intuitive.

Performing any kind of statistical analysis on ANY data, demands pure data.

We lack pure data in examining the four gospels, or Paul's letters.

here's a simple example:

data: 3, 8, 0, 5, 1, 7, 5, 7, 6, 4.

What is the probability that the next number to appear in this stream of data will be minus 2, i.e. -2 ?

What is the probability that the next number to appear in the data stream will be 582?

Neither number is likely, because the only data that we have seen have been positive integers, with a value less than ten.

However, suppose that the original data stream,

3, 8, 0, 5, 1, 7, 5, 7, 6, 4.

had been manipulated by someone, or some organization, for some specific purpose or reason.

Imagine then, that the REAL data, the actual data, was not

3, 8, 0, 5, 1, 7, 5, 7, 6, 4.

but rather, this:

-6, 463, 5, 93, 4, -82, 508, 6, -19, 3024.

Now, the probability that the next number to appear will be -2 or 582 is not negligible.

In fact, it is quite possible that such a value would be seen.

How likely?

That is the value of the Bayesian inference method. It permits one to predict probabilities, given sufficient previous information (data).

In the case of new testament data, one is feeding text (i.e. alphanumeric characters, in this case Greek alphabetical characters) into a computer which then must create tables, and perform computations on the tabular data.

If this tabular data is corrupted, the output of the computer program will be meaningless.

hope this helps.....

The Bayesian method demands unadulterated data, of certain validity.

Looking at the example above, we don't KNOW what the next number will be, but we MUST know, that the raw data representing the previous values is pure and not corrupted.

avi

squiz

Lies, damn lies, and statistics!
I would also like to see a short written (I also dislike podcasts) summary of how he proposes to estimate his probabilities and what he hopes to calculate with it. All I have seen so far (and I haven't read it all yet) is a paper which was big on blinding us with a nice looking mathematical formula but not very big on explaining how it should be used. I hope that Richard Carrier can make some useful and sound conclusions with this tool and doesn't fall into some sort of creationist 2nd law of thermodynamics fallacy.

jgoodguy

That is a real danger, a real complicated method to find 'truth' in historical documents or to show no 'truth' in historical document. The apologists will have a field day.

mountainman

For those who prefer reading, there is a pdf available from Carrier's website entitled The Twelve Axioms of Historical Method (March 2010) that is quite relevant to the OP, and to how the author sees Bayes Theorem fitting in.

Here is a list of the first 5 axioms: I think this is laying a good foundation ....


We might quip Carrier was not engaged in doing ancient history according to Axiom 3 when he recently tore into Sources of the Jesus Tradition -- see separate thread

aa5874

But, you will show from your own examples that statistical analysis does NOT need pure data because the statistical analysis can ONLY reflect the data presented.

You have shown that your analysis is quite in order because you used the DATA that was presented. You did NOT presume that the Data was invalid you simple made a statistical analysis of the DATA available.

Once you become aware that the data was corrupted you will only have to change your conclusion after APPLYING the uncorrupted data. That is all.

You have shown again that your statistical analysis is in order based on the DATA that you used. Your statistical analysis ALWAYS follows the DATA.

All that is needed is Data and the probabilities will change when the DATA is changed.

It can be seen that pure data is NOT needed for a statistical analysis but it is correctly analysing the data available that is most important and necessary.


I personally do not think the Bayesian method is much use for the basic analysis of figures of history with respect to the NT.

However, we do have EXTANT Data from antiquity and it can be used to ADVANCE theories about the existence or non-existence of characters found in the NT.

It is the DATA as it is found that should be analyzed not what it might have been.

When one discovers a cave writing or a fragment of some ancient text one may not even be able to determine the "purity" of the actual data but one may be able to determine what is written.