There's a difference between seeming to work properly when "hooked up" and being tested. Testing is a rigorous process, and rigor is something that's seldom seen in conjunction with PC's.

You're right, of course, in that there are many things that work much of the time and then fall apart under some application-specific timing or sequence. Rigorous testing exposes those, while superficial "try it and ship it" often doesn't. That's why the test has to be designed before the circuitry. That way you design the circuit to meet the required test criteria rather than trying to invent criteria that your circuit will survive once the boards are assembled.

Then, there's the business of testing under other-than-normal operating conditions. What, for example, is it going to do if someone starts a transfer before the cables are plugged in? What happens when they're plugged in after the process is started?

See what I mean?

RE