I didn't have the space, but I originally intended to include a section contrasting the two incremental approaches to welfare reform that are frequently favored by self-described libertarians. One is the road taken: workfare, "learnfare," and other attempts to reduce the rolls through social engineering. It has been pushed mostly by neoconservatives like Myron Magnet and neoliberals like Mickey Kaus, but plenty of libertoids endorsed at least some of their ideas. The other approach is the set of proposals usually known as a negative income tax or a basic income grant, in which the government distributes money directly to its citizens without the intercession of a welfare bureaucracy. Variations of this have attracted support from across the political spectrum, from Milton Friedman to Andre Gorz to Huey Long.
The first approach has the potential to shrink the welfare state by one measurement -- the number of caseloads -- but would probably cost more money and would certainly increase the system's intrusiveness. The second approach, if enacted properly, would be both cheaper and less intrusive, but could increase the number of people getting checks from the government. So there's a tension here. (For the record, I strongly prefer the second option to the first. Someday I might stake out a defense of that position. For now I'll just stick with the article I wrote.)
I didn't mention it in the column, but there's at least one more way to measure the welfare state: the number of employees administering it. Here, once again, the second approach to reform is preferable to the first. Here too is a reason the negative income tax has been enacted as an add-on to the welfare state, in a couple of poorly designed pilot programs and in the Earned Income Tax Credit, but has never been enacted as originally intended -- as a substitute for the rest of the welfare system. Bureaucracies want to perpetuate themselves.