[extropy-chat] calling all bayesians

[Mike Lorrey]

--- spike <spike66 at comcast.net> wrote:
> 
> Guys help me eff this real-life effing problem:
> 
> I build 150 droobs and use 131 of them in my freem.  I
> test the remaining 19 spares destructively and find that
> all are good.  From that information only, what is the
> probability that all 131 droobs are good?
> 
> I have four Monte Carlo sims chewing on this problem
> but they are giving me puzzling results.  A closed-form
> solution to this would be impressive, winning my
> undying respect.

Depends what the MTBF was for each of the 19 spares, as well as the
standard deviation in the distribution of MTBF for them. You must
compare this against how much operation time under similar conditions
is considered "good". Do you mean just good out of the box? You also
have problems of the various components that the droobs are made of,
their MTBF. What sort of performance is considered 'good'? For example,
there are at least three standards of performance for resistors in
terms of resistance variation from the labelled value.

Without knowing any of this, the only metric I have to go by is
population statistics in which the opinion of 200 people supposedly can
quite accurately predict the opinion of 20,000 within less than 1%
(according to FSP statisticians). Based on this, the behavior of 19
components could thus predict the behavior of 1900. Since you only have
131, you should be able to say they are all good out of the box if you
test only 2 spares and find them good. 19 is overkill, so you can
pretty conclusively say that all 131 should be good out of the box, but
nothing more about MTBF.

Mike Lorrey
Vice-Chair, 2nd District, Libertarian Party of NH
"Necessity is the plea for every infringement of human freedom.
It is the argument of tyrants; it is the creed of slaves."
                                      -William Pitt (1759-1806) 
Blog: http://intlib.blogspot.com


		
Yahoo! Mail
Stay connected, organized, and protected. Take the tour:
http://tour.mail.yahoo.com/mailtour.html