"Sure. But if you give them all away, what do you sell?"
The anecdotal evidence Guy Kawasaki bases his advice for self-publishers on is that giving away books is always good. My publishers have signing events at conferences and give away case-loads of my current book at these events. Amazon sales always spike afterward.
In practical terms, that tipping point seems to be very far out there.
I guess my question is, where do you draw the line? You said, there is a lot of evidence that promotional copies sell books. Sure. But if you give them all away, what do you sell? (Assuming that you want to sell any at all.)