Okay, former Sierra-employee chiming in again -
Software testing is an INCREMENTAL process - that means that as the product is developed, copies are sent to the QA teams at regular intervals. These are tested, and reports are returned to the developers.
Issues are tracked and tested to ensure that they're fixed in the next version - and then the issue is "closed out" (unless it pops up again). I'm certain that the QA team DID have hundreds of suggestions and bug-fixes; but these took place in the weeks and days leading up to the release.
Then, when its time for the "final" release - SUPPOSEDLY all bugs have been fixed; but the QA teams do 1 final pass through the software to ensure that no bugs have popped up in the "final packaging" of the software, that it installs correctly, etc.
This is why it only takes 3 hours (and in retrospect, only took 1). They don't WAIT to QA until the last second - QA'ing is done continuously as the product develops, between Alpha state and "Gold".
--Noel "HB" Wade
P.S. Yes, lots of programs are still buggy; sometimes due to rushed schedules, sometimes due to bad drivers... I'm not saying it doesn't happen, I'm just trying to educate folks on how the process works.