The pivot of the paper is to call any instance of institutional choice based on concern for the welfare of the people who will live within the institution “paternalism.” So, if you prefer institutions or sets of rules that would tend to make everyone involved better off than they would be under the alternatives, you're in favor of “paternalism.” Which is, of course, a position that everyone has agreed with pretty much forever. Because “paternalism” is a term that actually has content narrower than “in favor of a good set of rules,” that can't be what paternalism is.
Usually, paternalism is thought of as essentially involving interference with someone's liberty or autonomy for their own good. As Dan Klein notes in his totally dispositive reply to Sunstein & Thaler, a restaurant manager's putting the dessert bar at the back isn't intereference with autonomy. Nor is eliminating the dessert bar. A guy who wants to order a burger at a sushi place isn't having his autonomy meddled with, even if sushi is better for him.
We call the Founding Fathers “fathers” because the republic is their progeny, not because the separation of powers is good for us. Vernon Smith, to take another example, has studied experimentally how different forms of market institutions result in different outcomes (even if formally identical), because of the way actual people actually interact with the set of rules. Is Smith's program “paternalistic” for promoting the use of market designs that are actually as opposed to notionally efficient? Is an improvement in the floor rules of the Chicago Mercantile Exhchange “paternalistic” because it makes the trader and his principals better off? Of course not.
And, of course, when somebody exercises their liberty, the action isn't therefore “libertarian.” Libertarianism is a poltiical philosophy that is about the minimization of state coercion. When I choose vanilla over chocolate, that's not “libertarian ice cream flavor selection.” And when a restaurant manager, or human relations director, excercises the legitimate powers of her role in an organization, and decides where to put the salad bar, or decides how to structure the benefits package, there's nothing specifically “libertarian” about that, either. None of S&T's example of libertarian paternalism are examples of libertarianism or paternalism. And otherwise, the paper is more or less empty. (“If you're picking rules, pick good ones!”) So you'd think that would be a problem.
I'm not saying anything Klein doesn't. So, if you haven't read his paper, do. And read his rejoinder to Sunstein's evasively glib reply. As Klein points out, it's part of Sunstein's long-term project is to abuse the meanings of words so that his opponents' positions come to appear conceptually impossible.
So, don't let people get away with calling an employer's choice to set their benefit or investment defaults one way rather than another “libertarian paternalism,” since such choices are neither libertarian nor paternalistic. Accuse them instead of making a category error and speaking nonsense. And why not? There is no better way to win friends.