I have been buying staff from eBay, and the item have not been arrived in expected time, When I look at tracking on USPS I see the item have been stopped in a place called BILL Garden,Ca (The Black Hole of Bell). after I was reading some website people have same problem and say they lost staff from there too, and one person said that "99.9% of the packages mailed through the USPS eventually get delivered".
My questions:- Is not wrong here to use this probability and instead we have to use bayesian probability because maybe the probability of item that shipped by USPS and sorted in BILL Garden,Ca and not delivered is much more higher?
In this particular context, I think you may be confusing Bayes' Rule with conditional probability.
Assume for the moment that there is any kind of long-term, consistent behavior in any part of USPS. It seems that in the USPS realm, one person thinks that $P(\text{Eventual delivery}) = .999$, whereas you think $P(\text{eventual delivery}|\text{via Bell})$ is quite small.
With sufficient information on USPS shipping patterns, Bayes' Rule would allow you to compute 'inverse' probabilities such as $P(\text{via Bell}|\text{eventual delivery}).$
However, in some sense both of the probabilities you mention might be considered as 'personal probabilities' based on personal belief, not necessarily on data. Sometimes such probabilities are used as 'prior' probabilities is Bayesian analysis.
Another, not totally serious, example of a personal probability is "68.7% of all probabilities are made up on the spot."