Google errs on the side of rejection because one bad hire has a much bigger impact than one good hire.
We do not hire at random, each hiring goes through multiple interviews and the results of each of those interviews are also reviewed by multiple people (interviewers are required to record all notes/code produced during the interview). Then a decision is made. If we feel not super certain we err on the cautious side and turn down the candidate, even though we know there is a good chance for false negatives.
As far as I can tell, our false positives rates are very low, and everyone I've worked with here at Google are incredibly qualified at their job. Are there people who don't perform? For a company this size that's an obvious yes, but I think if you judge interview's goal of eliminating false positives at the expense of producing false negatives, then we've been pretty successful.
I've advised people on a job search that the way it works with tech hiring is they ask you a few questions and either you'll happen to be able to get the answer quickly or you won't, and if you go on enough interviews eventually you'll get one where you get all the answers quickly and you will look really smart as a result. Consequently, you can't pin your hopes on or predict whether you would be able get hired by any particular company.