The devil's in the details on this stuff, and unfortunately we don't have the details so it's hard to say exactly how much devil there is in them.
How do they lower the false negative rate? A sibling comment mentioned Google's practice of giving diversity candidates a second interview if they fail the first. This would mean if you have a false negative rate of `n` your false negative rate would become `n^2` (which is lower because n is hopefully much less than 1). However, this also increases your false positive rate from `p` to `1 - (1 - p)^2`. So in effect, this is lowering the bar as it's giving certain groups a better chance of being hired when they're not qualified than others. I would be very interested to hear about a hiring practice that lowers false negative rate without affecting false positive rate. I can't think of one right now but it seems like it should be possible.
That's not really moving the bar though - false positives are undesired errors. Also, at Google hiring decisions are made by committees using feedback collected from the interviewers. Presumably they don't discard the first interviews, but instead consider all of them.
How do they lower the false negative rate? A sibling comment mentioned Google's practice of giving diversity candidates a second interview if they fail the first. This would mean if you have a false negative rate of `n` your false negative rate would become `n^2` (which is lower because n is hopefully much less than 1). However, this also increases your false positive rate from `p` to `1 - (1 - p)^2`. So in effect, this is lowering the bar as it's giving certain groups a better chance of being hired when they're not qualified than others. I would be very interested to hear about a hiring practice that lowers false negative rate without affecting false positive rate. I can't think of one right now but it seems like it should be possible.