I had/have the same aversion to selling things that don't exist that you have. But honestly, I think that the next time I do a startup, I'm going to do it that way. I get really sick of speculatively building things that nobody wants. Why not make sure they want it first and then build it?
Plus, two of my favorite startups - DropBox and RescueTime - both used this approach (though perhaps with a little less dishonesty - they clearly marked it "demo"), and their products are better for it.