If by "almost always" you mean never, then you're right.
Let me put it way: if you wanted to create a terrific blockbuster novel, would you start by scanning in some obscure, cheap title from the discount bin and then editing it, until you had your masterpiece?
These projects are abandoned for a reason. They need to be refactored/modified to go anywhere. If you think it makes more sense to refactor a simple/failed project than rewrite it from scratch, then you're not in the "savvy buyer" category I was talking about.