on paternalism in welfare policy

Welfare requires the premise that some people are generally not capable of taking care of themselves. So, it's probably helpful to start by acknowledging that we're talking about the degree to which govt should be paternalistic. (And of course, the govt is the govt, so they can do as they please within political constraints.)

The ethical question is the (ethical) extent to which govt should be paternalistic. Good people will disagree quite a bit here. (It's probably worth noting that nobody wants the govt to paternalistic with them individually; it's always for the *other* people.) 

The first consideration is the extent to which people make systematic mistakes. (And here, we're trying to define mistakes "objectively". If you make a decision that I think is bad-- e.g., spending some money on lottery tickets, going to church, letting your kid play football-- is that a mistake or your best decision given preferences and constraints?) The next consideration is the extent to which people should be protected *by govt* from those mistakes.

The practical question is the extent to which various forms of policy paternalism are effective, short-term and long-term. (For example, to what extent does protecting me from mistakes today lead to more mistakes in my future.) The practice of paternalism certainly requires a higher monetary/regulatory cost. For example, it's more costly to regulate food stamps a bit than just giving cash without strings attached. But what are the benefits?