[tomjen3]

Sure, but you have to multiply that with the (perceived) probability of it being an issue with the expected benefit. That is probably a calculation that often comes out as "get it, dammit".