I actually did an interview at MS about a year after Win7 was released (was fresh out of college), and I asked a pretty pointed question about why the release quality seemed so… variable. The manager’s answer was that they had done entirely in-house QA for XP (we didn’t go into WinMe), outsourced the vast majority for Vista, and brought it entirely back in house for 7. He further mentioned they were taking a hybridized approach for 8. I remember questioning the decision, given the somewhat clear correlation between release quality and QA ownership, and got some business buzzword gobbledygook (which I took as “the real answer is so far above my pay grade that I can do absolutely fucking nothing about it”).
TL;DR: it was largely just profit-driven quality cuts done too aggressively, so they had to backstep and reinvest a couple times to normalize it for the user base.
Well, it was more than that.
I actually did an interview at MS about a year after Win7 was released (was fresh out of college), and I asked a pretty pointed question about why the release quality seemed so… variable. The manager’s answer was that they had done entirely in-house QA for XP (we didn’t go into WinMe), outsourced the vast majority for Vista, and brought it entirely back in house for 7. He further mentioned they were taking a hybridized approach for 8. I remember questioning the decision, given the somewhat clear correlation between release quality and QA ownership, and got some business buzzword gobbledygook (which I took as “the real answer is so far above my pay grade that I can do absolutely fucking nothing about it”).
TL;DR: it was largely just profit-driven quality cuts done too aggressively, so they had to backstep and reinvest a couple times to normalize it for the user base.