This is true also, though I can't imagine getting more than the 2-4 patches we get every year given the time they take -- even if they were "free". I'm guessing (with no numbers to back it up) that the development costs are the biggest part of the cost of making a patch. (In practice those costs might not count the same to bean counting types because they don't change the bottom line -- they just eat at the development time for the next year's product).