I'm putting this part first as it helps clarify things following.
Meaning what? It is by definition a non-negative number. What, then, are you talking about?
Positive = "it happened"
Negative = "it didn't happen"
I have only a guess as to how anybody would "run" a probability, and no idea at all what distinction you are getting at when you contrast "literally" with "conceptually" beginning to perform a physical act on a mathematical abstraction.
As part of my daytime occupation, I am a programmer.
"Run" simply is a habit for me to convey when any computation is accomplished.
"Literal" would be that the value begins as a negative integer; obviously, probability does not do this, so I was clearing up that such was not my intention.
"Conceptually" means that for all intents of History (not Maths), relying on only a probability is to inherit a negative and wait for the probability to indicate a positive (a probability worth noting as "sure, that may have likely happened") before accepting some event or figure into the Historical record.
Do you have evidence, argument or demonstrative proof that probability is inferior to other methods of uncertainty management when the amount of evidence is small?
That wasn't really my point.
The idea of using Bayes' Theorem has been thrown around as if it's the wonder-drug solution to all uncertainties of History and that History should just sweepingly move to this method for establishing Historicity.
No, it should not. It should use probability models (as well as any logical tools which assist) to aid, but not replace the method.
The only point I was making was that the method is an entire method, not one tool. It doesn't really make any sense to replace an entire method with one tool.
Is it not the case that when evidence is overwhelming and copious, it makes little difference what method you use? It is only when evidence is meager or equivocal that the choice of method can matter much. I have already, in earlier posts, listed several difficulties the "historical method," in the style of some posters, has in addressing the on-topic inference problem that I have. Bayesian methods do not have any of these impediments. I choose accordingly.
The point of this section was to highlight that we full well understand that we have a great lack of most of the material regarding the greater chunks of the Ancient record, so relying on probabilities to indicate something to us as the primary method of quantifying Ancient History isn't really that gainful as an alternative to the current Historical Method, as it would be erroneous to negate mass amount of notated History simply because a single probability indicated that a particular event or figure may or may not have occurred or existed; especially when we full well understand that we are absent much material.
Most probabilities ran would result in Historical negatives, not positives.
Currently, we accept the positive and work to prove a negative.
On the average, this method works out fairly well; even if specific cases slip through in both directions occasionally.
Flipping the other way would have us run a probability, which in several cases would result in a probability indicating it didn't occur, and then revising it as we get more information and maybe one day entering the concept into the Historical record.
But there's an issue here, I've mentioned before.
No one would work on the position of the positive if the event or figure wasn't entered into the record in the first place.
Look how long it has taken to get back to actually getting the official Historical record to recognize and re-open the inquiry into the Native American heritage sites and accepting archeological evidence that was shuffled off and dismissed as unreliable for political reasons in the 19th century.
Assuming the negative and hoping that someone will just one day prove a positive meanwhile the tiny piece you do have sits on a back shelf in the belly of the Smithsonian among several countless pieces of material people have forgotten about because they are filed in the "forgery" and dismissed section (the attic of History if you will), is a very uneventful means of proceeding through Historical inquiry.
I'm all for using a variety of tools, but we do actually have to inherit the positive and prove the negative, and not just assume the negative until the probability calculation spits out a probability we deem is worth considering the positive.
Good for you. But my problem is not to increase the output of history, but rather to estimate the truth about a focused question concerning what happened in the distant past, based upon information available to me now. As I have said repeatedly, the very fact that historians are solving a different problem than mine is ample warrant for me to look elsewhere for heuristic guidance in solving my problem.
Conversely, it is all the same to me whether or not historians adopt some "tool." So long as they are interested in achieving goals besides case-by-case truth seeking, like "having more history," then how they pursue their goals can only be sparingly related to how I might achieve mine.
And here's the catch; you have items to estimate the truth about for yourself because the positive is inherited.
If it wasn't, and we went the other way around, then you wouldn't have nearly anything to investigate from Ancient History.
Instead, there would be angry people storming the internet, after a few generations, demanding to know why pieces of information weren't being entered into the public record as Historical and instead were being shuffled off to the belly of museums for no one to request, digitize, or see on the average.
You wouldn't have people making their dissertations on the current Historical recorded line of the Egyptian Kings because most of them wouldn't be accepted as evident; there would just be a big black hole in that section of the History book.
If you wanted to argue that the Indus Script was a root language related to Phoenician you wouldn't even have that idea in mind.
Why not?
Well, the probabilities would strongly suggest (as they do today already) that all of the Indus Script artifacts of name plates that we have found aren't actually indications of a language at all and the Indus Script language would be wiped out from the Historical record.
Instead, you would have no entry on the matter. You may one day look at all of the name plates and see a pattern and think that it looks like there's a language and then propose that to the community in general, but then everyone would just show you the probability indicating that it wasn't an indication of language and to shut up and go away.
Instead, we inherit these items as indications of a language, as impossible as that seems, and folks work passionately on trying to unlock the code of exactly how this language worked and what the language even was, or whether these name plates even have representations of a language, or whether some parts are part of a language and other parts not.
Once, the same thing somewhat happened to the Maya language.
People made the mistake (yes, the mistake) of just thinking the wall decorations were just decorations; clearly the unorganized pattern and massive diversity wasn't a language. It was just art work.
Then the community found out this was a mistake; wrong. Instead, there was a variation in the way of expressing the components of the language which was artistic and the variation was rather wide.
Now Mayan walls are re-examined (and have been) as massive packed walls of written records which were entirely absent previously.
That was rather lucky.
Again, I have no problem with using Bayes Theorem inside of the Historical Method, but I do have a problem with up and replacing the Historical Method with Bayes Theorem (and honestly, I don't understand Carrier to be suggesting this approach for History at large, so the application of such upon any figure specifically in such a manner is queer, to say the least).
