This is different than what I typically see Weibull used. I typically see it used in applications for Reliability, where the values within the distribution are the times that it takes something to BREAK, not get fixed.

I have computed Weibull Parameters that describe the distribution of the lengths of time it takes machines to get fixed. They are characteristically less than 1, indicating so called 'infant mortality' is present. Here is my question:

Imagine that we want to estimate the mean time it takes a machine to get repaired. Let's say the mean time is 5 hours. So at T=0, we expect the machine to be fixed at T=5.

Which of the two statements is applicable to my data, noting that infant mortality is present?

A) At T=3, we now expect that the machine will take LESS than 5 MORE hours to fix, so we update the estimated time of repair to T = 3 + x, where x is a number < 5

B) At T=3, we now expect that the machine will take GREATER than 5 MORE hours to fix, so we update the estimated time of repair to T = 3 + x, where x is a number > 5

Thank you for your thoughts in advance!