Good companies should care about their false negative rate, as if they're smart they will realise that unlike Harvard their pool of potential candidates does not completely refresh each year. As you yourself exemplify, Google are now seeing many candidates who have applied before. While there are more than enough fresh bright young people growing up each year to fill Harvard's interviewing pipeline with first-time-candidates, there clearly aren't enough new top-rate computer scientists (who are interested in working for Google) coming into the field each year to fill Google's interviewing pipeline with only first time applicants.
And if you are reliant on the false negatives coming back, and not being put off by the process, to keep your pipeline full of high-quality candidates, surely that has to be a risky approach longer-term. As Google loses its fast-growing-exciting-upstart reputation, and competition for engineers gets tougher, -- it's gradually going to get harder and harder to convince the best of the false negatives to come back into the process again.
And if you are reliant on the false negatives coming back, and not being put off by the process, to keep your pipeline full of high-quality candidates, surely that has to be a risky approach longer-term. As Google loses its fast-growing-exciting-upstart reputation, and competition for engineers gets tougher, -- it's gradually going to get harder and harder to convince the best of the false negatives to come back into the process again.