John Small wonders what the deal is with anti-paternalism. First, in his view, it seems an ad hominem attack on any kind of disliked policies. Second, many "nanny state paternalistic" policies are supported by lots of folks; objection to those policies then seems anti-democratic.
Taking those in turn, while there are some policies where John rightly notes fuzzy boundaries between paternalistic and non-paternalistic policies, for most it's pretty clear-cut. A policy is paternalistic if its main objective is to protect you from yourself. So, yeah, it would just be ad hominem (and absurd) to label, say, laws against murder as paternalistic. We don't have laws against murder solely for the purpose of saving the murder from ex post regret. But it's hardly ad hominem to note that a policy's main goal is to stop you from doing things that might (or might not) hurt yourself.
The main problem is that all sorts of laws that are based mainly on paternalism are framed as being based on externality grounds. Because their proponents know not only what's best for you but also that you'll oppose those laws if they're presented honestly. It's hardly ad hominem to slash through the shonky veneer that gets pasted over such policies.
Small then finds it incongruous that Buchanan would advocate that a popular vote (on a constitutional amendment) could be used to prohibit the use of popular votes on issues like paternalistic restrictions. Buchanan noted the case where weak majorities might be found preferring each of several paternalistic policies (while strong minorities oppose each) but where broad support could be found for restrictions on democratic choice on the set of policies (because different folks are in the minority on different issues).
The whole point of Buchanan's constitutional enterprise is that it is wholly legitimate and even necessary for democracies to bind themselves against certain types of decisions. At the constitutional level, we set restrictions on certain types of choices, requiring that they pass only by means of supermajorities. This doesn't "deny the legitimacy of democratic decision making"; rather, it makes it possible for democratic choice to be legitimate!
Buchanan and Tullock lay it all out in their 1962 Calculus of Consent. If we consent to the constitution, then the rules that follow afterwards also are legitimate. What sorts of constitutions could be supported behind the veil of uncertainty? Well, we want to make sure that we're not going to be on the losing side too often, especially on decisions where fundamental rights are at stake. But it's often better to be on the losing side on some issues if the value of playing the game at all is high enough: all having a veto on everything probably wouldn't work too well. So, what sorts of decisions should pass by supermajorities? Ones where external costs are high relative to decision-making costs. Recall that decision-making costs are increasing in the size of the required majority (it's more costly to get 75% to agree than to get 51% to agree) while external costs -- the costs falling on those who disagree with the decision -- are decreasing in the size of the required majority. So, where are external costs relatively high? Decisions relating to fundamental rights, for starters. Would we want 51% to be able to expropriate the other 49% by popular vote? Does saying no deny the legitimacy of democratic decision making?
Where else might we expect external costs to be relatively high? On decisions that affect personal autonomy where no other issues are at stake. Yes, my autonomy is restricted by having to drive on one side of the road rather than the other, but others' autonomy to hurt me is restricted at the same time. The class of rules that would restrict my autonomy for the sake of preventing my imposing harms on myself (as others deem harms and benefits) seems pretty likely to have pretty high external costs behind the veil. And so it isn't crazy to think that constitutional rules could well be warranted protecting against this kind of legislation.