Evidence for Jesus

[eight bits]

OK, let's work backwards.

Pr(X/Y) changes (unless the evidence is unclear), which is the whole point of doing a Bayesian analysis. If you believe all swans are white(X), and you see a white swan (Y), Pr(X/Y) will increase.

No, not in the same problem, that is, so long as the same individual's beliefs are being represented by the probabilities discussed. Different people will typically have different beliefs, and the same person may change beliefs "between problems." Looking at the problem you pose:

Assuming that 0 < Pr(X) < 1 (you believe it to some degree, but are uncertain), then viewing a white swan can only increase your confidence that all swans are white. In your notation, that is Pr(X/Y) > Pr(X).

Neither Pr(X/Y) nor Pr(X) changes by observing something that makes Y true. Your current degree of confidence in X, however, is no longer Pr(X), but rather Pr(X/Y). Representing that change of belief is "the whole point of doing a Bayesian analysis." What represents the change is satisfying different conditions while maintaining a single, unchanging comprehensive probability distribution over all possible states of the world, evaluated as of the beginning of the problem.

There may be a temptation to denote the current probability of X as Pr(X)... but you've already used that notation for a different quantity, the prior probability of X, and in any case Pr(_) has been used for constants, not variables. So you can't use Pr(X) for something else, unless you declare "Let's call the search for more evidence a new problem." OK, you could do that.

But nature's way of telling you that that would be a bad idea is when six posters have eight ideas about which probability is being discussed, and that's what happened in this thread with the New York Lawyer, and the Apocalyptic Preachers of Old Jerusalem, and so on.

In Hans case ...

I'd like to reserve comment on Hans's case for now, if you don't mind. He's just posted, and probably it's best if you and I stayed with these issues for now.

For example, a racist detective may believe "Suspect A (who's black) committed the crime" has a high probability. His non-biased partner might assign a much lower probability. If they don't straighten it out, the evidence won't be interpreted right.

Different people often have different beliefs. Subjective probability represents those beliefs. So, yes, "the" probability that A committed the crime will differ depending upon whose beliefs the probability represents.

There are some problems where we expect people to agree if they have the same information, as when there is a well understood chance set up, like a bag of marbles. "Whether A committed the crime" has no such chance set-up. We can only expect people to agree if there is hefty shared information, and they may well disagree about how hefty any particular state of information is. The racist may see some pieces of evidence as less bearing than his partner does.

As The Norseman remarked

Cool! Now where is the bag full of green marbles with regards to an historical Jesus?

There isn't any. The Christ-myther may see some pieces of evidence as less bearing than the Christian apologist does. May?

There is no Bayesian way to broker these disagreements. There are two "right" interpretations of the evidence, right because each one faithfully represents the changing beliefs of the person whose interpretation it is. Which is "correct" depends on which person, if either, is correct, and ultimately, in the detectives' problem, whether A did in fact commit the crime. Neither of the detectives knows, nor do we. Hence the name, inference under uncertainty.

If you think about it, it would be magical if a method for representing beliefs caused only correct beliefs to be represented. Bayes offers many wonderful things, but magic is not one of them.

It's the same hypothesis ("Will I pick a green marble out of the bag?"), but the prior probabilities differ because of the different sets of background information.

That's fine if you were talking about two different problems, each with its own background information. But you spoke of some one probability changing (skyrocketing), not differing between two problems, and what occasioned the change was an "added" detail, suggesting a change in the available information within the same problem, which is what usually happens in a Bayesian problem.

Then you did discuss two different believers (OK! Two different problems here we come), but then used the same "k" for their different prior beliefs (huh?), and ... well, it all needed to be sorted out, I thought. Hopefully, we have. or at least made a dent in it.

 

[HansMustermann]

Actually, while you're on the right track about the difference between the probability of something in the real world and the confidence someone has in a particular claim, I think that's an important distinction to bear in mind.

Something doesn't actually become true or correct, just because someone believes it true.

And for Bayes like for any other formula or algorithm: GIGO (garbage in, garbage out). If the data that goes into it is no more than personal beliefs, sure, you could still do the maths on the back of a napkin, but the result wouldn't even be worth the napkin. It's not just another way to be correct.

Or as a different way to put it, for a result obtained by Bayesian reasoning to be worth anything, the same applies as for binary logic: both the reasoning must be valid, and its premise true. Yes, it doesn't mean that X must be true or false, but that Pr(X) and the other probabilities you you punch into the formula should still be based on the real world. Even if you use estimates or probability intervals, they should be defensible based on real world data.

Otherwise, if it's just representing unfounded personal beliefs, it's just exploring the implications of that belief system, rather than saying anything about the real world probability that something is true.

E.g., in the example with the racist detective, any data that comes from just his unfounded prejudices, is just garbage in. And you get garbage out. The result represents no more than the logical (or I should say Bayesian) conclusion of that system of falsehoods in his head. It doesn't mean he's correct too, it means he just calculated garbage.

And really, it's not just for Bayes. As I was saying, GIGO applies just as well to any formula. We could be having this discussion about Pythagoras instead, and basing a calculation on fantasy still wouldn't be worth anything.

E.g., I could calculate the number of shingles I need for my roof based on the actual numbers, and get a result worth anything. Or I could base it on just a delusional belief that the height is an imaginary number. Or on a belief that it's a ten-dimensional roof, because Yog Sothoth told me so in a vision. (Mind you, I might have been nearly terminally drunk and probably asleep at the time, but still, hey, if you can't trust a multi-dimensional outer god about a multi-dimensional roof, who can you trust, eh? ) But if I do any of the last two, it isn't a way to be correct too, and I shouldn't wonder much if I have too many or too few shingles.

But basically I'll agree with you, if I understood right, that ultimately it's calculating just someone's confidence in it, rather than any real probability, but I'll disagree that then everyone is correct. The ones plugging fantasy data in it, are still getting an incorrect result.

 

[HansMustermann]

In Hans case, I think he's leaving out important background information: the abundance of messianic prophets at the time, and Paul's letters referencing Jesus. If there were no actual person, it would require torturous explanations that don't fit with our background knowledge of how cults form.

In Hans's case, he just wasn't talking about that in the argument you jumped into

And you'd have probably noticed it if you actually read the argument, instead of just jumping feet first to defend Jesus.

Because the argument here wasn't as much whether OTHER considerations might support Jesus, but whether there's any merit to such silliness as that because Jerusalem or carpentry are real, that somehow makes the core "Jesus existed" claim more likely. Which it can't.

Even if I were to grant that other arguments like the plethora of messianic pretenders have any bearing on Jesus's existence (although see the marbles example of Last of the Fraggles for what's wrong with that), it still doesn't mean that every single piece of nonsense is valid just as longe as it makes Jesus real. Even if argument X were sound, that doesn't mean that any random arguments Y and Z are sound too.

Even if you could make Jesus real by such references to hordes of messianic pretenders and whatnot, and let's call that argument X, it doesn't mean that nonsense argument Y, where the existence of Jerusalem or carpentry somehow makes Jesus real, also has merit too.

 

[Fudbucker]

OK, let's work backwards.



No, not in the same problem, that is, so long as the same individual's beliefs are being represented by the probabilities discussed. Different people will typically have different beliefs, and the same person may change beliefs "between problems." Looking at the problem you pose:

Assuming that 0 < Pr(X) < 1 (you believe it to some degree, but are uncertain), then viewing a white swan can only increase your confidence that all swans are white. In your notation, that is Pr(X/Y) > Pr(X).

Right.

Neither Pr(X/Y) nor Pr(X) changes by observing something that makes Y true. Your current degree of confidence in X, however, is no longer Pr(X), but rather Pr(X/Y).

Right, Pr(X) changes after the evidence is evaluated.

Instead of Pr(X), let's use Pr(H), where H is the hypothesis (All swans are white). E is the evidence of a white swan. Before the evidence, Pr(H) might be 99%. After the evidence, Pr(H) will increase. So I don't understand when you say

n Bayesian terms, none of the probabilities change during a problem, because they're all conditional or prior (unchanged by evidence, by definition)

I don't know why you wrote "during the problem". Not even sure what that means. Before the evidence, there's Pr(H). After some piece of evidence, Pr(H) usually increases or decreases. It's still the same hypothesis.

Representing that change of belief is "the whole point of doing a Bayesian analysis." What represents the change is satisfying different conditions while maintaining a single, unchanging comprehensive probability distribution over all possible states of the world, evaluated as of the beginning of the problem.

This is not clear. I've only had a few courses in logic.

There may be a temptation to denote the current probability of X as Pr(X)... but you've already used that notation for a different quantity, the prior probability of X, and in any case Pr(_) has been used for constants, not variables. So you can't use Pr(X) for something else, unless you declare "Let's call the search for more evidence a new problem." OK, you could do that.

This is also not clear. Probabilities just represent degrees of belief (in Bayesian calculus). If H is the hypothesis that all Swans are white, why would you use another variable for H at any other time? Pr(H) will be constantly changing as new evidence is gathered. Do you want to call it Pr(H/E1), Pr(H/E2)...? That seems needlessly complicated.

Different people often have different beliefs. Subjective probability represents those beliefs. So, yes, "the" probability that A committed the crime will differ depending upon whose beliefs the probability represents.

Right. My point was that background knowledge and beliefs have to be evaluated to be sure a proper prior degree of belief in a hypothesis is established before the evidence is considered. The racist cop's Pr(H) will be much different than his non-biased partner, and more to the point, the racist cop's Pr(H) will be wrong (or rather, "unjustified").

In Hans case, I think he's got the initial Pr(historical Jesus) so low, that evidence that would normally be persuasive isn't moving the needle much. So he gravitates to odd explanations that don't make much sense (specifically about Paul's letters). The simplest theory is that the early Christian communities Paul was writing to were started by a messianic prophet names Jesus. Anything else is needlessly multiplying entities.

 

[Fudbucker]

In Hans's case, he just wasn't talking about that in the argument you jumped into

And you'd have probably noticed it if you actually read the argument, instead of just jumping feet first to defend Jesus.

Because the argument here wasn't as much whether OTHER considerations might support Jesus, but whether there's any merit to such silliness as that because Jerusalem or carpentry are real, that somehow makes the core "Jesus existed" claim more likely. Which it can't.

Even if I were to grant that other arguments like the plethora of messianic pretenders have any bearing on Jesus's existence (although see the marbles example of Last of the Fraggles for what's wrong with that), it still doesn't mean that every single piece of nonsense is valid just as longe as it makes Jesus real. Even if argument X were sound, that doesn't mean that any random arguments Y and Z are sound too.

Even if you could make Jesus real by such references to hordes of messianic pretenders and whatnot, and let's call that argument X, it doesn't mean that nonsense argument Y, where the existence of Jerusalem or carpentry somehow makes Jesus real, also has merit too.

Edited for civility

 

[eight bits]

Hans

Something doesn't actually become true or correct, just because someone believes it true.

That has been my experience.

And for Bayes ...

... it is a belief representation. Its operations are consistent with, and ensure compliance with widely accepted (but not universally accepted) heuristic principles of uncertain inference. Which is all there is in contingent uncertainty, heuristic principles of inference.

I don't really know you, Hans, but I think I know you well enough to say that if you think Bayes sucks, then you'd like the leading alternatives even less.

... that Pr(X) and the other probabilities you you punch into the formula should still be based on the real world....

As I mentioned in my post, there are problems where you would expect little disagreement about the probabilities, usually because there is a well understood chance setup, or occasionally the question is logical (If all the marbles in the bag are green, then any marble drawn from the bag must be green).

There's another category of problem where maybe people can be "well calibrated." That means, for example, that it rains on 79,7% of the days when the weather forecaster says "the probability of rain today is 80%." Obviously, this category requires some kind of repeated trial to measure.

Then there's everything else people have beliefs about. No objective probability model, no repeated situation to count hits, just something that is true or false, maybe once ever. Guess which category Hisotrical Jesus falls in.

These are questions about the real world. Evidence is things that happen in the real world. But the trajectory of some beliefs happens nowhere except in people's heads. If there's enough evidence, then it will serve to coordinate people's beliefs. A few scraps of debatable evidence? Then there will be unresolvable disagreement forever.

... rather than saying anything about the real world probability that something is true.

That's right. There is no "real world probability" that a proposition is true, unless it falls into one of the special categories, which Historical Jesus doesn't.

I'm not sure what your point is in the elaboration of the other poster's racist cop example. Offhand, I don't see any advantage in having Bayes misrepresent his beliefs. As I said in my post, it is magical thinking to suppose that a belief represntation could correct his beliefs.

So, the racist cop probably does see a lot of things differently than I do. I could have guessed that with or without Bayesian analysis. Its usefulness lies elsewhere.

if I understood right, that ultimately it's calculating just someone's confidence in it,

Yes.

rather than any real probability,

Well, if there is a real probability in the problem, then I would expect Bayes to agree with it. For example, my confidence in a fair coin toss is equal for either outcome because my beliefs are in fact consistent with the real probability, and Bayes correctly represents my beliefs.

but I'll disagree that then everyone is correct.

I didn't say everyone's beliefs are correct. I said that the Bayesian representation of the beliefs is (or can be) correct, meaning Bayes accurately portrays the beliefs. Whose beliefs, if anybody's, are factually correct is what we don't know.

The most decisive end position would be when almost everyone is in close agreement, usually because there's lots of evidence. Those who aren't in agreement have some distinctive story which explains their disagreement, like the racist cop being the last one to say that a black suspect almost certainly didn't do the crime.

There's really no way to do better than that, which isn't Bayes' fault.

 

[HansMustermann]

In Hans case, I think he's got the initial Pr(historical Jesus) so low, that evidence that would normally be persuasive isn't moving the needle much. So he gravitates to odd explanations that don't make much sense (specifically about Paul's letters). The simplest theory is that the early Christian communities Paul was writing to were started by a messianic prophet names Jesus. Anything else is needlessly multiplying entities.

Skipping over all else, I don't think that "bayesian" means whatever handwaving you feel like doing. It kinda centers around Bayes's theorem. It also is not just a blank check to cricumvent the basic skepticism of needing some evidence.

(And no, just wanting to believe that whoever called himself Paul (even Luke doesn't think it was his real name) in a story must be totally telling the truth, ain't evidence. Otherwise I'd have to also conclude that The Call Of Cthulhu was totally written by Francis Wayland Thurston, because he says so himself right from the start, and he was a bona fide witness to some craziness about some ancient god. I mean, why would he invent all the people like the distinguished professor George Gammell Angell and H. A. Wilcox and detective Legrasse as characters who confirm his story? He must have really gotten the story from them, like Paul got it from whatever Jerusalem apostle gang, right? Isn't the simplest explanation that he's just telling the truth? )

But that's a digression, since I wasn't talking about that.

If you want to address my use of probabilities, please do. But just doing a pseudo-intellectual version of "oh Hans just doesn't want to believe", is still not the same as supporting your case.

But if you want to talk Bayes, by all means, go ahead and show your maths. again, "bayesianism" ain't quite the same as a blanket excuse to commit an argument from personal incredulity. Bayesian reasoning is just maths, not an excuse to handwave why you want to believe Paul. You have to show the maths leading to that, if you want to claim to be right by way of Bayes.

And, yes, it's based on imperfect knowledge, and it's a probability conditional of that knowledge being true, but it's still not just an excuse to handwave personal incredulities that Paul would lie or whatever. You still have to have something that's both valid and based on defensible premises, even if it's applying a formula instead of binary logic. You still have to show why that P(X|Y) is greater than 0.5, and now you also have the problem that P(X|Y) generally doesn't say jack about P(X) unless Y is absolutely true (as in As in P(Y)=1) or X actually doesn't depend on Y at all and has the same probability for non-Y too (i.e., the conclusion is actually orthogonal to the supposed premises, i.e., it's a non-sequitur.) Otherwise you're not showing that X is the most likely, but just that it's the best fit for your unsupported beliefs on the topic. Well, I'm not going to demand a clean 1.0, but you still have to defend that premise Y.

I.e., it's not really escaping that need for evidence. Just moving onto Bayes still doesn't get skepticism out of the way.

But ok, I'm game. In fact it would be refreshing to see someone actually use "probably" correctly about Jesus. If you think that what you're doing is correct by Bayesian reasoning, please do show the maths and defend the values you're using.

It would sure beat just speculating about how my prior must be too low. Show me yours and explain why it has that value, and believe me, I'll accept maths in a jiffy.

 

[HansMustermann]

I don't really know you, Hans, but I think I know you well enough to say that if you think Bayes sucks, then you'd like the leading alternatives even less.

[...]

Then there's everything else people have beliefs about. No objective probability model, no repeated situation to count hits, just something that is true or false, maybe once ever. Guess which category Hisotrical Jesus falls in.

[...]

That's right. There is no "real world probability" that a proposition is true, unless it falls into one of the special categories, which Historical Jesus doesn't.

I guess I might be a hardliner bayesian or something, because I see no problem with treating it as real world probabilities, if calculated correctly. Even if it was something that only happened once.

E.g., let's assume I were the only nutter in the universe who's ever made a pair of dice in the shape of a great icosahedron and numbered the tips. (That thing ain't gonna land with a face up, I think.) And rolled that pair of dice exactly once. You have absolutely no frequentist statistics of any other such dice being ever rolled. You could still calculate which total value is more likely to have come up, innit? It will of course be conditional of stuff like each actually being reasonably enough close to a perfect great icosahedron, and rolled on a flat surface, and that they're not loaded, and all, but nevertheless, you can calculate a real life probability even for something that happened only once.

In other words, if there is any objective reason to believe that Jesus existed, I think we could make a probabilities model for it. An that it would be real world probabilities, even if those dice were only rolled once. We might have to work with intervals for the values instead of knowing the exact probability to be a disciple of John The Baptist, but still, it would be possible.

And if it isn't even possible to do a model of that at all, well, then I'd say that there is no rational way to that conclusion, for those arguing it.

That said, you're indeed right that I'd dislike misuses of frequentism even more

 

[eight bits]

I guess I might be a hardliner bayesian or something, because I see no problem with treating it as real world probabilities, if calculated correctly.

The main difference is that everybody has their own probabilities, if the probabilities' sole function is to represent a person's beliefs, which is what a Bayesian probability is. It is nice that when there are real world probabilities, that Bayes will perform exactly as well as the person being represented performs, but there are many, many problems where there is no real world probability, or where real-world and inner-world probabilities are both present.

Closest to the topic, there is no real world probability that John the Baptist had a star pupil who got crossed in Jerusalem and was sighted as being alive a few days later. About that, the only objective thing to say is that it happened or else it didn't. What's beyond that is that there are different people with different levels in confidence in whether that is true. Some of that variety is because of different information bases, but even people with the same information will disagree.

Suppose ...

Actually that's much like a favorite hypothetical of philosophers' discussions of probability. First, there is no problem at all for belief representation whether you're going to roll it once or repeatedly. For those who prefer to avoid belief representation (I am not sure why, since all the numbers will be the same, and you will do the same things with them), there is something called the "propsensitist" interpretation of probability. And it will be just as you say, you calculate your probabilities based on your engineering model of the physical situation.

How selecting a physical model isn't an expression of belief, and analyzing that model isn't belief formation, I leave to the philosophers. I suppose they'd say those are meta-beliefs instead of beliefs. OK, then.

In other words, if there is any objective reason to believe that Jesus existed, I think we could make a probabilities model for it.

You could try, but the track record for these things is poor. There's a lot of interest in it, too. Sports betting is the same kind of problem, and there's money in it if you can predict better than a bookie. So far, either not so much, or else the people who've done it don't advertise. Equities and commodities markets? Same situation.

The proverb is to never say never, but I wouldn't hold my breath waiting for one, either.

We might have to work with intervals for the values

It doesn't help. It's the same interpretations. You can create "probability theories" that have no numbers at all, if you like. You won't gain any expressive capability. (An interval model is a special case of a point valued model coupled with a particular kind of what's called "sensitivity analysis." So, if you built an interval model, you could have built a point model instead, and whether you actually did the sensitivity analysis would be up to you.)

Moreover, if there is no real uncertain probability, then there is no real interval that contains it, either. If you want objective probabilities, your interval contains both 0 and 1 (it did or didn't happen, after all, so one of those is the only available objective probability for Jesus). That's not helpful - really not, since any evidence short of revelaton cannot shrink that interval. You remain completely uncertain forever. Unless Jesus comes back, of course.

Better, I think, to take a chance and go with some "best guess."

And if it isn't even possible to do a model of that at all, well, then I'd say that there is no rational way to that conclusion, for those arguing it.

Rational just means that you don't contradict yourself. There is no way to guarantee that you have only "correct" beliefs in an uncertain contingency, except to have no such beliefs. That's not a popular choice. All probability can accomplish is to represent what you have instead, and make it easier to work with than talking to yourself would be.

That's not nothing. Probability gives you the use of a lot of attractive heuristics at no extra charge, and formality makes it possible to "check your work," which may reduce the number of inadvertant contradictions. It's also often useful to be able to explain how you arrived at a conclusion, and formality supports that, too.

That said, you're indeed right that I'd dislike misuses of frequentism even more

Yeah, I figured that would be a good bet .

 

[tsig]

The main difference is that everybody has their own probabilities, if the probabilities' sole function is to represent a person's beliefs, which is what a Bayesian probability is. It is nice that when there are real world probabilities, that Bayes will perform exactly as well as the person being represented performs, but there are many, many problems where there is no real world probability, or where real-world and inner-world probabilities are both present.

Closest to the topic, there is no real world probability that John the Baptist had a star pupil who got crossed in Jerusalem and was sighted as being alive a few days later. About that, the only objective thing to say is that it happened or else it didn't. What's beyond that is that there are different people with different levels in confidence in whether that is true. Some of that variety is because of different information bases, but even people with the same information will disagree.



Actually that's much like a favorite hypothetical of philosophers' discussions of probability. First, there is no problem at all for belief representation whether you're going to roll it once or repeatedly. For those who prefer to avoid belief representation (I am not sure why, since all the numbers will be the same, and you will do the same things with them), there is something called the "propsensitist" interpretation of probability. And it will be just as you say, you calculate your probabilities based on your engineering model of the physical situation.

How selecting a physical model isn't an expression of belief, and analyzing that model isn't belief formation, I leave to the philosophers. I suppose they'd say those are meta-beliefs instead of beliefs. OK, then.



You could try, but the track record for these things is poor. There's a lot of interest in it, too. Sports betting is the same kind of problem, and there's money in it if you can predict better than a bookie. So far, either not so much, or else the people who've done it don't advertise. Equities and commodities markets? Same situation.

The proverb is to never say never, but I wouldn't hold my breath waiting for one, either.



It doesn't help. It's the same interpretations. You can create "probability theories" that have no numbers at all, if you like. You won't gain any expressive capability. (An interval model is a special case of a point valued model coupled with a particular kind of what's called "sensitivity analysis." So, if you built an interval model, you could have built a point model instead, and whether you actually did the sensitivity analysis would be up to you.)

Moreover, if there is no real uncertain probability, then there is no real interval that contains it, either. If you want objective probabilities, your interval contains both 0 and 1 (it did or didn't happen, after all, so one of those is the only available objective probability for Jesus). That's not helpful - really not, since any evidence short of revelaton cannot shrink that interval. You remain completely uncertain forever. Unless Jesus comes back, of course.

Better, I think, to take a chance and go with some "best guess."



Rational just means that you don't contradict yourself. There is no way to guarantee that you have only "correct" beliefs in an uncertain contingency, except to have no such beliefs. That's not a popular choice. All probability can accomplish is to represent what you have instead, and make it easier to work with than talking to yourself would be.

That's not nothing. Probability gives you the use of a lot of attractive heuristics at no extra charge, and formality makes it possible to "check your work," which may reduce the number of inadvertant contradictions. It's also often useful to be able to explain how you arrived at a conclusion, and formality supports that, too.



Yeah, I figured that would be a good bet .

Probability means nothing when we're talking about a historical fact. I happened or it didn't and no amount of Bayes bashing can change it.

 

[HansMustermann]

Well, probably the most correct way to say what I'm trying to say isn't that some aren't beliefs, but that some beliefs are grounded in real world data, while others are based on just fantasy, and a lot are in between. If I believe that the sun rises because of Earth's rotation, that's a belief grounded in reality, while if I believe that Ra changes into a cosmic beetle and pushes it up the dome of the sky, that's just fantasy.

Of course, one can still do reasoning (including formal logic, informal logic, induction or bayesian reasoning) from both. It's just that the one based on garbage in will produce garbage out. If someone fails to produce the evidence on which that reasoning is built, well, what I'm trying to say is that I still won't take it seriously even if they claim to go by Bayes.

That said, I fail to see why the interval of probabilities for anything would have to contain zero and one. Since we're already discussing the probability for something happening or not happening, I don't see why 0 and 1 need to be even considered for probabilities.

E.g., any given person can either be schizophrenic (to a degree that merits such a diagnostic) or not, but for the probability of it, both 0 and 1 are actually unreasonable values. If I'm talking, say, "what's the probability of a guy being schizophrenic in the 1st century AD", then allowing a 0 probability involves allowing that their brains could have been so different from ours, that absolutely nobody was schizophrenic. Which is actually a rather unreasonable possibility to allow. And allowing 1 involves allowing the possibility that every single person was schizophrenic, presumably from birth. Which again is rather unreasonable.

ETA: or for an even better example: what is the probability that a random person is below the relative poverty line? Any given person either is or isn't below the poverty line. Yet allowing for that probability being 1, is allowing a mathematical impossibility. Any country for which I'm assuming it could be 1, is a country where I'm assuming it could be possible that 100% of the population could be earning less than half the median wage. It's mathematically impossible that everyone be on the same side of the median, so there is no point in allowing 1.0 as one end of the interval.

 

[Brainache]

Ebionites, probably yes. But if they were responsible for the DSS, are you dating these scrolls to the period 30 to 70 AD? Do you have evidence that they were assembled over this short period? (I presume you're not assigning any of them to the period 70 to 120 AD.)

I think there was a community of "The Poor" there under various leaders from about 100bc through to 70ad. I think it is possible that Jesus was just one of those leaders and that the practices and beliefs of these people were not entirely static over the period they lived there. Unfortunately providing evidence for these ideas is a bit beyond my abilities at present and I am fully prepared to accept that I could be wrong.

 

[Craig B]

I think there was a community of "The Poor" there under various leaders from about 100bc through to 70ad. I think it is possible that Jesus was just one of those leaders and that the practices and beliefs of these people were not entirely static over the period they lived there. Unfortunately providing evidence for these ideas is a bit beyond my abilities at present and I am fully prepared to accept that I could be wrong.

Well I suppose it's not impossible, but some positive evidence would be nice. No, it is indispensable! There was a Jesus (son of Stada) in 100 BC and a Jesus (son of Ananias) in 70 AD, and it is extremely probable that the stories about Jesus the Christ contain material derived from the biographies of these other Jesuses. But that there was yet another Jesus who was a leader of a postulated "Poor" sect, possibly as early as 100 BC, and that the stories about the Christ are based entirely on this Jesus, is less well founded on surviving evidence.

 

[Brainache]

Well I suppose it's not impossible, but some positive evidence would be nice. No, it is indispensable! There was a Jesus (son of Stada) in 100 BC and a Jesus (son of Ananias) in 70 AD, and it is extremely probable that the stories about Jesus the Christ contain material derived from the biographies of these other Jesuses. But that there was yet another Jesus who was a leader of a postulated "Poor" sect, possibly as early as 100 BC, and that the stories about the Christ are based entirely on this Jesus, is less well founded on surviving evidence.

In one of the DSS they describe themselves as "The Poor" or "The Poor Ones" so I don't see it as a big stretch. But you are right, not being impossible is not the same as being likely or probable.

Time to get back to work on that Time Machine I spose...

 

[Fudbucker]

Probability means nothing when we're talking about a historical fact. I happened or it didn't and no amount of Bayes bashing can change it.

Probability (how it's been used in this thread) represents our confidence level that such and such event happened/didn't happen.

You're right, historical events either happened or they didn't.

 

[maximara]

And Paul's letters to early Christian community's referencing Jesus. Our background knowledge is such that it's more probable there was an actual person who inspired the early Christian sects.

The King Arthur, Robin Hood, and John Frum Cargo cult say otherwise.

The John Frum Cargo shows the idea that the myth came first and that somebody tried to plug themself into that myth is just as likely.

King Arthur and Robin Hood show that you can have composite characters--legendary stories that are not of one person but of several people woven together until what you get is a person that in reality never existed.

When you get right down to it our background knowledge of Jesus is more suggestive of a composite allegorical figure draped in the garb of history then of an actual person.

Ignore the Gospels for a moment and concentrate on what Paul tells us of the man Jesus for moment. There is next to nothing regarding Jesus as a person and other then an off the hand comment about meeting his "brother" (spiritual brother, delusional, or conman...we don't know) we know nothing about the family of Paul's Jesus.

For an actual person this is very particular but using John Frum as a reference it is exactly what you would expect regarding an composite allegorical figure. Perhaps someone was inspired to become this figure but that doesn't mean that person is the "historical" Jesus any more then Manehivi is the "historical" John Frum.

In fact, John Frum is the perfect comparison because in 1991 (60 years) we were at roughly the same point from his supposed February 15, 1931 appearance as a vision as to when Josephus supposedly wrote about Jesus and here we have a vague history the gives us next to no details.

At about 30 years (when Paul is thought to have written his part of the NT) there was a variant of the John Frum cult where Prince Philip was his brother. Around this time people started to display photos of who they believed to be John Frum even if these photos didn't match each other (mirroring 2 Corinthians 11:1-4)

AFAIK no Church father talks about Jesus the man until the early part of the 2nd century which is about where we are with John Frum right now. It wouldn't surprise me if c2040 the first attempt to collect the stories of John Frum occurred (mirroring Marcion's attempt) and that several other attempts would follow making the already existing fragmentation evident.

Compare this fragmentation with cults that had known founders like Mormonism and you immediately see the differences. What we know of it Christianity of the 2nd century was nearly as diverse as it is now with even the idea of what God himself varying wildly much as one see with the Cargo cults.

In fact, some of these cults picked real people as their founder and teacher even if they had never been to the island or taught anything like they were saying - the long defunct FDR cargo cult case in point (they wanted to pay for his return).

 

[maximara]

Probability (how it's been used in this thread) represents our confidence level that such and such event happened/didn't happen.

You're right, historical events either happened or they didn't.

More times then not history is not that simple. Take the old myth Christopher Columbus sailed West to prove the world was round.
Strictly speaking historical event never happened as while Christopher Columbus did indeed sail West the reason was to find a cheaper route to the spices in the Far East.

Paul Revere's ride ala Longfellow is another myth that strictly speaking is not historical. Longfellow's poem is filled with inaccuracies and the Revere depicted is actually a composite of Paul Revere, William Dawes, and Samuel Prescott. Yes the poem is based on history but it in of itself is not history (or at least not accurate history)

 

[Lowpro]

More times then not history is not that simple. Take the old myth Christopher Columbus sailed West to prove the world was round.
Strictly speaking historical event never happened as while Christopher Columbus did indeed sail West the reason was to find a cheaper route to the spices in the Far East.

Paul Revere's ride ala Longfellow is another myth that strictly speaking is not historical. Longfellow's poem is filled with inaccuracies and the Revere depicted is actually a composite of Paul Revere, William Dawes, and Samuel Prescott. Yes the poem is based on history but it in of itself is not history (or at least not accurate history)

Sort of like the Huckleberry Finn fallacy? Sure the Mississippi exists and sure young boys may have traveled down it and had an adventure but that doesn't make them Twain's Huck Finn nor validate his story to being necessarily true even though elements of them may be true such as the Mississippi existing.

 

[eight bits]

tsig

Probability means nothing when we're talking about a historical fact. I happened or it didn't and no amount of Bayes bashing can change it.

It happened or not, but our problem is that we don't know which. Probability does not tell us which. We tell probability what we think about the uncertainty, and probability represents that for us in a convenient way.

So, probability means something in the context of a historical fact, it's just not the same set of meanings that probability has when there is a well understood chance set-up (like that bag of marbles) or some stable series of repeated trials (like insurance risk).


Hans

There's nothing in Bayes that denies that some beliefs are better than others, or that some people will make better use of the same body of information than others. Bayes offers precautions against some kinds of mistakes, and makes it easier to "reality check" beliefs. But it's the user who drives the inference, not the representation. A car may have safety equipment, but if there's a loose nut behind the wheel, then there will be accidents. Same thing with Bayes.

That said, I fail to see why the interval of probabilities for anything would have to contain zero and one.

Funny, I thought I showed it can't, not that it would have to. The issue arose because you seemed to want some objective probability for the historical question, and those were the only probabilities for which any sort of objective case could be made with any generality.

You wouldn't ever use zero and one as any part of a representation of your belief about something uncertain. There is no uniformly accepted Bayesian representation of "complete prior ignorance," that your only belief about an event is that it happened or else it didn't. This cannot be a surprise, since there is also no Bayesian account of original (prior) belief formation, either. (Compare evolution by natural selection. It does not account for the first lifeform; the theory is about something else, what happens when there are lifeforms. So it is with Bayes and beliefs.)

But that's fine, because as you say, you usually do believe something about that historical event, and so you are sure that your beliefs are not what zero or one would represent about your beliefs. So, you don't use them to estimate an appropriate measure of your belief state. You seem to like to express your estimate as an interval. That's fine, too.

However, nothing in that interval, whatever it turns out to be, is objectively correct. What is somewhere inside that interval is a fair measure of your confidence in what you believe about the question, not anything that is objectively correct to believe about the question.


max

You have no hard issue with anyone who acknowledges that the question is uncertain. Cargo cults don't show that all stories precede their protagonists, just that some do. Thus, you have shown by example that it is possible that Jesus stories preceded anybody thinking Jesus lived, even that it is seriously possible. Yes, it is. That's one aspect of what "The question is uncertain" means.

It is also possible for stories to accumulate around a real person. Dead people seem to be especially sticky in that way. For example, Mason Locke ("Parson") Weems' fanciful The Life of Washington appeared in 1800, shortly after its protagonist died. A later President, Abraham Lincoln, has recently added "vampire hunter" to his already lengthy list of more realistic imputed accomplishments. (Although it is not the first supernatural story about him.)

AFAIK no Church father talks about Jesus the man until the early part of the 2nd century

As Fudbucker points out, there are Paul's letters, which you seem willing to date, at least as a possibility, to the mid-First Century. Paul includes the eucharstic institution narrative in 1 Corinthians 11: 23-26. That is, Paul depicts Jesus being involved in an unsatisfactory social interaction and on that same night handling food, speaking with men, and discussing his physical vulnerability to injury. These are all human passtimes. One can hardly count the number of occasions when Paul speaks of Jesus dying, a once-in-a-lifetime thing that all men, and only living beings, do, but 11: 26 will serve.

Taking the letter as a whole, I understand Paul to be telling his readers that a man died, and then a few days later ceased being dead. The point of that is to invite readers to believe that they will either not be indefinitely dead themselves, or maybe they will even avoid dying altogether. Jesus comes up in this conversation as a proof of concept or precedent. In Paul's own thinking, moreover, he considers the "end of days" to be a real event, and it is belief in the actual occurrence of a real man's temporary death which persuades Paul that the "end of days" is really occurring as he writes.

Paul, in my view, is writing about a man, and was plausibly understood by his first readers to be writing about a man, a human being like themselves. If you read the letter differently, then that's fine, since that, too, is one aspect of what "The question is uncertain" means.

 

[HansMustermann]

In one of the DSS they describe themselves as "The Poor" or "The Poor Ones" so I don't see it as a big stretch. But you are right, not being impossible is not the same as being likely or probable.

Time to get back to work on that Time Machine I spose...

Being poor was a title routinely taken by different people or groups in that time. You can't assume that it was the only one. Not to mention that one is basically a rank of the Essenes, while the other is a group with an ideology rather at odds with the Essenes. I don't see how you can take it as necessary or even probable that the group some two hundred years later is the same one.

In fact, it would be quite unusual to have a group that splits off a main group and takes their name from a rank in the old group. It's like having a group split off the Catholic church and call themselves The Laymen (actually a rather accurate analogy for what the Ebionites would have had to do) or The Choir Boys. It's more common to have the Fan Dumb phenomenon in which they make some claim of being the only ones right and the rightful continuation of the old group, because they're the true fanboys. E.g., the Sedevacantists.

Anyway, you need more than that. As Craig said, you kinda need evidence for that. Otherwise it's like saying that the Tea Party in our time are the same group as those who did the Boston Tea Party.

Furthermore there's still the fly in the ointment that the ancient church fathers say they got their name from their founder and ideologist, the heresiarch called Ebion. I.e., some individual who called himself "The Poor", as in, singular. And they address individual claims and arguments of that one guy. That's quite different from being the continuation of any pre-Christ group by that name.

Of course, the church fathers COULD be talking out the ass and just inventing stuff about the Ebionites or their origin. In fact, it's a good bet that they are. But once you decide that they're just making up BS, you're left not knowing much about the Ebionites in their time at all, because everything else about them could be just as made up.