Julian Sanchez has some criticisms here (hat tip to commenter HPG) of Metcalf on Nozick and libertarianism. They seem fair, although I don't have time to evaluate them in detail (it's been a long time since I read Anarchy, State and Utopia):
Nozick is here setting up a dilemma: Under these idealized circumstances, from what is stipulated to be a perfectly just starting distribution by your preferred theory, a series of free choices yield a very different, and much more unequal pattern. On what we might call a strictly orstrongly patterned theory, then, Nozick observes that a highly counterintuitive conclusion follows: That from a perfectly just starting point, we quickly progress to an unjust state by a series of moves themselves involving no apparent injustice, but only people’s voluntary deployment of the holdings to which your favorite pattern theory entitles them. Alternatively, one’s preferred pattern might be loose enough to permit this transfer without dubbing it unjust—in which case one acknowledges that a pattern is not all there is to it, and one must know something about the history of holdings and transfers, not merely the overall distribution, to know whether it meets standards of justice.
Unfortunately, Sanchez doesn't address Metcalf's central point. The question is whether "liberty is the only value the state should concern itself with," to quote another (poor, ill-conceived) critique by the Cato Institute. If the state can't concern itself with liberty without concerning itself with other values, the libertarianism is in trouble. This is an empirical question: if refraining from redistributive measures lead to essentially authoritarian political conditions by concentrating wealth and power among a few, then libertarianism simply doesn't work. There is ample economic and historical evidence to support this view -- although I'd like to see a substantive challenge, rather than more conversations about the extent to which Nozick repudiated his views in Anarchy (which is a relatively trivial historical or scholarly question).
Update: This piece by Mark Thompson at The League of Ordinary Gentlemen is the best response to Metcalf I've come across thus far (other than Sanchez' more limited responses). It addresses the central issues head on, and makes clear how Metcalf has misused Nozick. Importantly,
Moreover, Nozick explicitly disclaims the idea that absent “patterned redistribution” or even government interference, recompense will naturally line up with talent and hard work.
And the point of the Wilt Chamberlain argument:
An important underlying point here, which makes Nozick’s argument so subversive, is that he’s suggesting that by reducing the concept of justice to a set formula, a “pattern,” one assures that inequality produced for undeniably just reasons is treated as indistinguishable from inequality produced entirely unjust reasons. Indeed, Nozick gives much space to advocating for an alternative, “historical” principle of justice, in which an inquiry ought to be made as to the justness of each particular transaction.
None of this makes the Chamberlain argument an indisputably successful rebuttal to the existence of a welfare state, though, and I don’t think Nozick would have suggested as such. Instead, in effect, he is simply arguing here that a system of redistribution in the name of justice must be able to point to specific unjust transactions it is rectifying if the system is to avoid causing injustice itself. It is simply not possible to say that deviations from a particular idealized pattern of distributive justice are inherently unjust, even if some or even most such deviations are a result of unjust transactions. Except for very rare circumstances, any such pattern will ultimately have to either allow for deviations or impose otherwise unacceptable restrictions.