Re: 'Industry standard' error rate.
I admittedly haven't worked in manufacturing at anything even resembling that scale, but I did spend a summer working for a company that assembled, sold, and installed PCs to local small businesses and (particularly) school districts. A 4-5% failure rate, at the point of installation, would have been absolutely inconceivable in that job.
Sure, there were lots of failures, most caused by assembly errors, some caused by damage during assembly, some due to component failures, but the vast majority were caught before the trucks were loaded. One guy with a rolling swivel chair and several industrial-sized KVM switches can verify that 500 (what you might expect for several computer labs across a single school district, for school districts the size we were dealing with) or so PCs will more or less work in really just a few minutes. I would assert that this sort of manual sanity checking is the sort of operation that is embarrassingly parallelizable too. Two people can check a batch of 1000 just as fast as one person can check 500. 20 people should be able to do 10,000.
And this was a small operation. Were we larger, it may have made sense to look into more automation.
Between those checks, and when they were hauled into schools and plugged in? Nothing much about a tractor trailer ride is going to make a PC go DOA.
So I guess my question is, what sort of QA are they doing at the factory to get a 4-5% DOA rate? Or if the QA is fine, what the hell is happening during the shipping? Are the devices overly fragile? If they are being damaged during shipping, could the same damage occur in the home?