I mean to write "not a panacea", my bad. That it's not the universal cure people think it is. And people _do_ think that re-factoring will magically solve problems, while it doesn't do all that much in practice, less so when you factor in the costs spent on re-factoring.