I took the liberty of performing statistical calculations on a very small
sample - which is always a dangerous thing to do. I bought a HP Scanjet
4670, but alas it was dead on arrival. No power, no light, no motion, no
nothing. Instead the upper left part of the machine got very hot. Serial
number CN417C3393. Withough going into boring excruciating details, HP
agreed to replace it. Just short of 2 weeks later the replacement arrived.
Serial number CN3CCB3375. Also dead on Arrival. OK maybe not dead but
mortally wounded. The light carriage would advance all the way to the right
and then get jammed there while making a loud Grrtrtrtrtr sound. I humored
the technical support person by complying with his instructions as he put me
through hundreds of hoops in order to utterly uninstall and reinstall the
software. It did not work.

Sample size 2, Successes 0 therefore predicted probability of units that
work is 0% with massive 95% confidence intervals spanning 0% to about 75%.
This is implying that the population average of Scanjet 4670s that work is
less than 75%. There is only a 5% chance that it is higher than that. I
guess only Hewlet Packard knows the real dead on arrival rate.

Poor HP. What is the DOA rate? And also, poor me! From those familiar with
manufacturing, what does it take to get 95% of all shipped units to work
when first plugged in?