Glen Whitman says, for example, that Wilkinson's argument only holds if public spending is such that it's only effective in large amounts -- if, that is, Buffett's theoretical $1000 is only effective if enough people contribute that it becomes worthwhile. If, on the other hand, each dollar of government spending is worth the same benefit, then Buffett should be giving more to government than to private charity if he really believes that government spending is more effective.
But that's not necessarily true. Suppose that Buffett believes that $1000 it would create $1500 in benefits if it was given to charities he selected; $1200 in benefits if it was given to government; and $800 in benefits if it was given to the charities everyone would actually choose (because in practice a ton of people are going to give to scams, or to charities which he believes do more harm than good). So if it's just his $1000, he'll give it to the charity he supports (over government) -- while still preferring a system in which he's constrained to pay that $1000 in taxes as long as everyone else is. In other words, he'll accept using his own money sub-optimally (in his own view) if it will buy everyone else contributing.
Indeed, that logic applies even if Buffett believes that the actual mix of charities people would give to if they were required to give produced more benefits than government did, if he also believes that forcing people to pay taxes is OK but forcing them to give to charities is improper. Why might someone believe that? Arnold Kling, apparently imagining an improbable world in which voluntary contributions would be sufficient, says:
I think it is worth imagining a world in which government competes on a level playing field with other charities. That is, imagine a world in which government relied on voluntary donations. In such a world, government would be smaller and other providers of public goods would be larger. To me, that sounds like a win-win.Is it win-win? Only if you assume that "other providers of public goods" are inherently more benign than governments. Is that true? I can understand why someone might believe it, but I certainly don't. Non-governmental organizations such as, say, churches, have a long history of all sorts of nasty things when they are the dominant organizations within a polity, and I can't think of any organization which isn't vulnerable to some sort of pathology when it gets large enough to provide the public goods that a very large nation would benefit from. Which is not to say that governments are necessarily more benign, either; it's just that there's an argument to have here, not the assumption that (some) libertarians buy in to.