I left the following thoughts in Megan McCardle's comments, but I thought I would share them here, along with some additional thoughts about liberalism and libertarianism . . .
Governments use coercion to make things happen. Government coercion can be legitimate, but it has to meet certain conditions. One of the traditional conditions is that a majority of the citizens who are going to be coerced, or a majority of their representatives, have to agree to it. However, if policies are structured so that we can't see whether or how we are being coerced, then we can't freely endorse them in the democratic process. So those policies fail the test of legitimacy.
Part of the issue here is a big principle-agent/incentive compatibility problem between representatives and the citizens they represent. Politicians want to get re-elected. If they can subsidize interest group A at group B's expense without group B really noticing due to the hidden transfer, then that will sound like a real winner to a politician. Which is just to say that the incentives politicians face encourage them to violate the very conditions of transparency and public justification that make their coercive powers legitimate. That sounds like a problem to me.
Politicians would have a constant incentive to try to violate and work around an explicit transfer requirement. Which is exactly why we need one. It would give anyone in the group from which resources are being appropriated standing under the Constitution to file suit in order to repeal the law licensing the hidden transfer. The whole class wouldn't need to notice the hidden transfer, and then fight it off politically. Only one person would have to notice and win in court. If the politicians can succeed in passing a law that makes the transfer explicit, then that's OK–it satisfies the minimal conditions of legitimate government power.
It occurs to me that Charles Murray's “just give everyone $10,000 a year” plan is a lovely example of a rule that satisfies both the explicit transfer/public justification/democratic transparency requirement and Buchanan's generality requirement.
For those of us, like Murray and me, who are Hayek/Friedman/Buchanan-style classical liberals as opposed to Rand/Rothbard-style libertarians, things like generality and transparency matter. There is no natural rights beef against transfers per se, but a fundamentally liberal beef against institutional forms that undermine the conditions for liberal legitimacy. If the conditions of liberal legitimacy are met, then the result is the free-market, minimal welfare state. This is a “libertarian” result according to the vernacular, if not according to orthodoxy. It is surely a classical liberal result. (Is Samuel Freeman right that Rand/Rothbard libertarianism is not really a kind of liberalism?)
Here is a hypothesis for debate: The cause of classical liberalism as a really existing possibility for political reform has been harmed by bundling free markets with a ban on transfers. This package deal has influenced people who think justice requires transfers to eschew free markets. If we had spent the last forty years hammering away at liberal fundamentals like transparency and generality instead of the natural right to not be taxed, our society would now be closer to the free market, limited government ideal.