A few points:
Most people have no idea what HR is. By this I mean, have you ever heard of someone majoring in HR in college? There aren't many formal educational programs for HR. Sure, there are training programs, both internal to a company, and provided by third parties, but they generally aren't comprehensive, nor are they widely used. Most HR people aren't formally trained for most aspects of their jobs.
Add to that that IT is a young, dynamic industry. Hiring doctors, for example, hasn't changed much. Sure insurance programs have changed, and there are new techniques, etc, but fundamentall, doctors have a degree, and board certification, and then experience. The strucutre of their job remains the same, even if the knowledge might progress. 30 years ago, you didn't have formal QA, and build engineers, you just had "software guys." Doctors may prefer different techniques, but it's ok if they are different. Take someone who only knows XP and put him on a team of RUP folks, and there will be some initial costs. This problem didn't even formally exist 20 years ago when methodologies were much less complex.
Then add a simple communication barrier. I've met many HR people who don't really understand the different between
Java and C, other then that they are different languages, and so a "foo" position must have someone with "foo" on his/her resume. If they understand the differences, they might realize the differences may be surmountable, in light of the other skills required. In my last company, I sat down with the HR folks and sorted resumes with them, to her train them as to what I'm looking for. I'd explain why I accepted or rejected each one. not a perfect method, but it did improve our system.
Haggling is value added (from the companies perspective), but if that's the only value add (aside from maybe filtering on key words), then it would be cheaper to train other managers to haggle. (Heck, everyone should know how to do that, for their own careers!)
If you read my essay
here, you'll see that I expect HR to be able to fully gauge ability and projected revenue. Sadly, I've yet to meet anyone who can do this. Granted, it's hard, but that doesn't mean we shouldn't try--partial information is typically better then no information.
--Mark
PS Please feel free to keep your topics open even if they've been answered; most people do. You never know when future ranchers feel like picking up the discussion.