The other day, I got into a Facebook argument over the fairness of a progressive income tax and means-testing of the retirement income assurance system (e.g. Social Security and IRAs). When faced with the assertion that it is unfair to discriminate against relatively wealthy citizens, I responded that true impartiality is impossible given that the state does anything, and therefore the closest thing to fairness is to provide income support to the least well-off members of society.
This is essentially Rawls' "maximin" criterion. As I superficially understand it, Rawls took this as an a priori principle of fairness, and used it as justification for extensive state intervention in a fundamentally free-market (liberal) economy. I want to look at it from the perspective where the state has already intervened in the economy, and we use the maximin principle as a way to minimize the unfairness of the outcome. This is similar to Kevin Carson's maxim that the state "breaks our legs and then gives us crutches", and that we must keep the crutches until our legs have healed. For the sincere statist, this can be seen as a principle of fairness; but for the ruling class managers, it can be seen as a social engineering paradigm to minimize the number of individuals who are willing to rebel despite exploitation.
I start with the premise that every state action produces winners and losers. This is true even if most state actions are positive-sum interventions. To neo-liberal economists, positive-sum interventions should be pursued whenever possible without consideration of fairness, because some of the excess produce can (ideally) be redistributed to the losers to achieve Pareto efficiency. The problem is that it's impossible to identify the winners and losers of each action. Given a large population and a large number of interventions, there will be some people who have repeatedly been on the losing end of these interventions, to the point that their ability to make a living (or save for retirement) has been crippled. Abandoning these people to poverty would be a great injustice.
In this context, we can only observe and manipulate the outcomes. Some people will have done very well for themselves -- through a mixture of effort, benefiting from state policies, and luck (i.e. everything else) -- while others will have done poorly. If we were to select the people who have the least success at life, we would probably also be selecting the people who have been harmed the most by state policies, along with those who exerted the least effort and were otherwise the most unlucky. As such, the provision of an economic safety net can be a strategy for mitigating the cumulative unfairness of those who lose out due to numerous state policies. To take it further, if we're asking the state to compel citizens to provide public goods, then we've already accepted the premise that "we're all in this together", and an economic safety net is justified as a way to mitigate bad luck (which is more in line with the Rawlsian idea).
A typical conservative rejection to the above argument may be that very few (if any) people have been impoverished by the policies that have allowed others to become wealthy, and that differences in effort account for the vast majority of the differences in economic success. For instance, they may say that tax rates on poor people are low, so that even if someone doesn't benefit from state spending, they still have not been crippled by it. Likewise, they may assert that the per-person economic costs of interventions like copyright are minimal, either because they believe that culture would collapse in the absence of copyright or because they believe copyrighted items are essentially frivolous luxuries. While I would point to mass-incarceration for victimless crimes as an example of how state policies "for the greater good" can impoverish some individuals, conservatives would likely assert that the criminal activity (e.g. illegal drug use) was bound to leave the prisoner destitute anyway, or that the criminalization of this activity was only a minor imposition on the criminal, and that their poverty is due to their choice to disobey legitimate authority. Many of the other policies that increase living expenses and depress incomes are so complicated that tracing cause and effect is nearly impossible. Suffice to say, I think that they are sufficient to impoverish many people, through the cumulative effect of many small burdens.
To solidify this viewpoint, I turn to a theory of the exploitative state. The first component of this theory is that while some state actions are positive-sum, many are zero-sum or even negative-sum. The reason that such outcomes are common is that political power is unevenly distributed among the population, and the determinant of state action is that it benefits the powerful -- not that it benefits the population as a whole.
This imbalance of power has a consequence that is even more important than the existence of negative-sum outcomes -- it biases the outcomes such that some people (the powerful) systematically win, while others systematically lose. This scenario does not even require that the powerful consciously exploit the powerless, only that they obstruct any policy that hurts them and promote policies that help themselves (regardless of their impact on the powerless). In this environment, the gains and losses of government action pile up and account for a substantial amount of the wealth variation in the population.
Given this systematic transfer of wealth from the powerless to the powerful (and the consequent construction of both a wealthy and an impoverished class), a person concerned with fairness should not be too bothered by policies that appear to exploit the rich. While an excessively populist attitude can be leveraged by the agents of the state to stuff their own pockets and intimidate critics, adding a clearly defined progressive component to a tax system does not facilitate such shake-downs. By shifting the tax burden from the poor to the rich, we can mitigate the systematic unfairness of government actions. It's only a superficial solution, but it definitely is not the problem.