AOL is a big enough target that it's worthwhile for spammers to write custom code. HN isn't, fortunately, so the technique works well here.
Good point.
While I appreciate your insight and experience, you might want to view http://news.ycombinator.com/newest with and without "showdead" turned on for your account: http://news.ycombinator.com/user?id=jaylevitt
Frequently over half the submissions are dead spam, presumably killed by the policy in question. I rarely see false positives. While the quality of the remainder isn't always great, it's a lot better than it would be if there were no controls in place.
ps. Flag some spam while you're there :)
We were wrong. It didn't really block much spam; from the very beginning, spammers had "quality control" bots checking delivery rates of their spam, and they'd adapt to new algorithms within hours, demonstrating just how well they could monitor their black hole rate.
Early this decade, the technology to do efficient blocking "at the edge" was finally implemented, and boom - no more lost mail.
Silent bans are really just a way to exert power. They achieve nothing.