Indeed. I suspect that, despite the sensibilities of the average HN’er, Google’s metrics tell them these garbage results are actually satisfying the average user more than results that alternative weightings would produce.
So do mine of course, and so do yours. Metric design is a hard, hard problem.
I've come to the conclusion that I should keep secret metrics from people so that they aren't captured by Goodhardt's law, but apart from the ethical implications, I haven't managed to implement this anywhere yet.