Scott Alexander, about whom both David Friedman and co-blogger Bryan Caplan have raved, has a lengthy book review of David's The Machinery of Freedom. As I write this, there are 479 comments on his post and I looked only at about the first 100 or so to see if anyone was making the point I want to make. I didn't find it, so I'll make it here.
I don't want to try to defend David's belief in anarcho-capitalism from Mr. Alexander's critique, because I suspect David will do a better job than I would, partly based on David's awesome intellect and partly based on the fact that I'm not sure anarcho-capitalism would work either, which is why I'm not an anarcho-capitalist.
I do want to defend David from one of Alexander's criticisms, however. Alexander quotes the following from David's book:
Suppose that one hundred years ago someone tried to persuade me that democratic institutions could be used to transfer money from the bulk of the population to the poor. I could have made the following reply: "The poor, whom you wish to help, are many times outnumbered by the rest of the population, from whom you intend to take the money to help them. If the non-poor are not generous enough to give money to the poor voluntarily through private charity, what makes you think they will be such fools as to vote to force themselves to take it?
Alexander has a semi-good critique:
I think I have a good answer to this question. Nobody's vote makes very much difference, so people are happy to vote for signaling/psychological reasons rather than financial ones. If casting my vote to help the poor makes me feel like a good person, but losing money in redistribution schemes makes me poorer, well, my vote 100% determines whether I feel good or not, but only 1/300-million determines whether I get poorer. This might also be profitably mapped onto construal level theory, ie Robin Hanson's Near Mode vs. Far Mode.
As I said, it's semi-good. After all, we do have many programs that help the poor, as Alexander goes on to point out.
But why the "semi?" Because it isn't enough to vote for politicians (we rarely get to vote directly to help the poor) who say they will implement programs to help the poor. If we're voting for "signaling/psychological reasons," then we are unlikely to put much time into actually making sure that the programs work. And we're also unlikely to put much time into defending the poor and near-poor from the predatory state. Many people have written about how predatory the Ferguson, Missouri city government is and so I won't take time to give you the links. And I've written about how the drug war does a lot to put people in the bottom one percent.
The argument has often been made that helping the poor is a public good. There's something to that. But if the government sets up programs that ostensibly help the poor, we simply shift the public good problem. Now the public good is monitoring the government. Just as we had an incentive to free-ride on the charitable activities of others, creating too little charity, so also when the government runs programs, we have an incentive to free-ride on the monitoring activities of others, creating too little monitoring.