[Synoptic-L] Re: [GPG] Matthew and "Q"
- BRUCE: I think a little better than Dave here seems to of statistics; a good
statistical result, at least in some cases, comes complete with a label as
to its probable accuracy or range. Very grownup.
In responding to Dave I. I said that if all we had to go on what the statistical results, then my statements would be a guess (at best). What I meant was, that I had made statements supposing that Matthew might have written around 85, etc... Even with other data, statements like this may be little more than a guess, but the statistical results are completely silent here. The point was just that I was considering other data, in addition to the statistical results.
As far as our relative confidence in statistical results in general, I'm not sure how much disagreement we really have. - If I have complete assurance the model is correct, then I agree absolutely that things like "95% confidence" are completely applicable. They are rigorous mathematical statements in that case. The problems tend to slip in when the models are good, but not good enough.
Here is a short story illustrating the sort of thing I worry about, which given its connection to current events might not bore everyone too much.
I did a degree similar to a masters degree in financial engineering, and most of the classes presented us with nice clean mathematical models. You can calculate why an option on a stock should be priced the way it is, for example. Stock movements are generally modeled as being log-normally distributed. This means they have the same chance of going up 1% in a given day as the do of going down 1%. Small moves are more probable than large moves, etc...
We also had a risk management class. Here you are focused on the lowest tail part of the distribution, i.e. "what is the worst that can happen?" But there is a problem - the distribution that seems to work so well most of the time does not fit so well in the tail. The tails are "too fat". The distribution describes normal market moves very well, but seems to have problems with "exceptional events". If we were very careful, we might make statements like "assuming the model holds, the risk of a loss this large is less than 1%". But in real life the CEO might then demand to know the "total risk", including the risk that the model is wrong. But the problem here is we don't know what we don't know. All we can say is that we know what to expect where experience is a guide. If asked what to expect if experience ceases to be a guide, we should be silent. But, given that this is not an option, risk management is full of mathematical "hand-waving" with no firm footing at all, making it look like they can answer the unanswerable. The result is things like a housing crash.
The author of the "Black Swan" also focuses on this issue, if anyone is interested in the problems of risk management.
> What you propose is clearly not impossible. But it is certainly not as good
> as the original. What you present here is a brief general call to
> repentance, followed by scenarios of people eager to repent. In our extant
> Luke there is a warning of wrath and fire, so that by verse 10 one can sense
> the crowds feeling guilty and ready to make amends. It parallels an
> evangelistic meeting where there is a lengthy build-up of emotion before a
> challenge to commitment. Luke was a good storyteller!
'Better', is a subjective judgment. Personally I'd prefer a version without the fire and brimstone, but that's just my subjective judgment.
Consistency, and sticking with a theme is a less subjective measure.
>I really don't see this as much of a jump, if any. v3 mentions forgiveness (group not named), and the quote from Isaiah's 'punchline' is about salvation (for all). "Forgiveness -> salvation" seems like theme continuity to me.
> But there remains the jump in the opposite direction, between verses 3 & 4.
> In the extant text vv. 4-6 represent a temporary departure from the theme of
> repentance so that Luke can portray the Jewish scriptures as hinting at the
> salvation of the Gentiles (v.6).
I've not had a chance to look at the rest yet. Probably Tuesday, if not before then.