[panda88888]

Here's my understanding of Google's hiring practice. In essence, Google optimizes to minimize false positive (i.e. hiring unqualified engineers). Hence, even if an applicant meets the bar (i.e. qualified), it is still a crapshoot whether the person is hired or not. By lowering the false negative rate (i.e. rejecting qualified applicant), the effective hiring rate for the target demographic group is higher if all other factors are assumed to be constant (and that's a big if and debatable).

Edit: Another way to say this is, if an applicant simply meets the bar, the person is most likely not hired. Conversely, an applicant needs to far exceed the bar in order for interviewers to vouch for the person and improve odd of hiring.

On the other hand, applicant that fits into the targeted demographics are probably afforded additional consideration, even if the person's interview did not wow the interviewers as much, hence the "lowering the bar" argument. It is not lowering the standard bar, but rather the "score", if you will, the applicant needs to clear the bar.

The best analogy I've read so far is that the candidate's score and its error bar must clear the bar. But affirmative action applicants are extended the courtesy of second look, which reduces the error bar, hence lower score is needed for the score and error bar to clear the Google bar.