Consider the following only slightly exaggerated version of an argument that occurs in a lot of political philosophy

1. Institution X is unjust or bad.
2. Therefore, X should be abolished or reformed.

What's wrong with this? Well, two things, actually, which I describe below.

First, identifying some feature(s) of X as bad or unjust doesn't give any reason, or at least no particularly strong reason, to believe that an alternative institution will be better or less unjust. A joke illustrates the problem. A Roman Emperor asked to hear the best singers in his kingdom. The finalists were narrowed down to two. The emperor heard the first one, was unimpressed, and promptly announced that the award goes to the other finalist, because the next singer must be better than the first one. Of course, that’s wrong: the second one could be no better or worse. The emperor needs to hear both singers to make a proper judgment.

  Call this first problem the nirvana fallacy. The nirvana fallacy is not that common in my view among political philosophers; they usually mention, at least implicitly, some alternative institution that is supposed to eliminate or lessen the alleged evil or injustice. (*For a whopping example of the nirvana fallacy, see below.)

  However, simply identifying some alternative institution is not sufficient, for one must also specify what processes or mechanisms of the alternative institution are likely to bring about an improvement. And failing to do this is, alas, fairly common in political philosophy. (For some prominent examples, see the excerpt from chapter 1 of my book: http://www.cambridge.org/servlet/file/item_9780521677936_excerpt.pdf?ITEM_ENT_ID=2348506&ITEM_VERSION=1&COLLSPEC_ENT_ID=6)

 Call this second problem the failure to engage in comparative institutional evaluation. 

  Failing to do comparative institutional evaluation runs the risk of failing to compare like with like, with jumping between the real and the ideal. Arguments that an institution is bad or unjust are typically arguments about real institutions. So an argument for changing it must specify a real or realistic alternative. (The reason for the modifier “or realistic” and the challenges it raises for libertarians will be mentioned below).

  This suggests the following better argument for institutional change:

1. Institution X manifests or produces injustice or social evil E.
2. Institution Y has some processes or mechanisms that make it likely that it will lack E or manifest or produce less of E than X does.
3. If an institution produces or manifests more injustice or evil than a feasible alternative, it ought to be abolished.
4. Therefore, we should abolish or alter X and bring about Y.

Even this isn't quite right, because even if Y produces or manifests less of E, the change from X to Y might produce or manifest such side effects or so much injustice that it would be wrong to change X and try to bring about Y. But the above argument is certainly an improvement from the original defective argument and gives a rough idea how to do comparative institutional evaluation.

Two lessons can be drawn from the need for comparative institutional evaluation, one for political philosophy in general and one, a challenge specifically for libertarians. The general lesson is that doing political philosophy–at least if one is arguing for institutional change–requires social science. To argue that one institution is superior to or less bad or unjust than another one must compare how the different institutions work (or realistically would work) and that requires social science. A lot of political philosophy ignores social science; hence a lot of political philosophy is defective.

 The challenge for libertarianism is to avoid over-idealizing their alternative to statist institutions. Sometimes there simply aren’t any real alternatives to statist institutions; that’s why I said the comparison sometimes is real-realistic, rather than real-real.

I don’t have a general rule for how to avoid over-idealization.  Often history is a great help. The lack of a welfare state will lead to the immiseration of the poor? Nope, take a look at David Beito’s excellent book on mutual aid societies.  The market can’t provide certain public goods? Again, the empirical trumps the a priori. (For a long time the lighthouse was considered a paradigm example of a good the market couldn’t provide till Ronald Coase pointed out it had been provided without the state. http://www.econlib.org/library/Enc/bios/Coase.html) Sometimes using arguments about similar or analogous institutions will help. So for example, there is nothing libertarians can point to that exists or existed that is genuinely free market health insurance (the claim that the US private health insurance is somehow an exemplar of market health insurance is utterly false—see pp. 42-53 of my book or David Henderson “Myths of US Health Care” in Better Medicine: Reforming Canadian Health Care, but one can use arguments about how market insurance works in other areas where the state’s presence is not as overwhelming as it in health care to show how it would work in health care. Still, one person’s realistic alternative is another person’s unfair use of an ideal market to compare to a real statist institution. This is an issue which libertarians need to give more thought to, I believe.

* Here's the example: Ronald Beiner, “What Liberalism Means,” Social Philosophy and Policy 13 (Winter 1996), 203, says the following: “A liberal is someone who says that the present social order in contemporary, Western, democratic, individualistic and pluralistic societies is basically okay, apart from a need for improvements in equality of opportunity and more equitable social distribution. A critic of liberalism like myself will say this is nonsense. To this, the liberal will reply: ‘Okay, this isn’t good enough; what’s your alternative?’ It is both necessary and legitimate for me to claim that I don’t need to answer this question. . . . That’s not my job. My job as a theorist is to criticize the prevailing social order.”