Government is just another means of getting things done, but it is seldom the best means. All governmental functions could be performed by private institutions -- in most cases, if not all, with better results for society at large.
Much of what government does it is able to do because of its coercive power -- not because it offers the best deal in town. Goverment either commands resources that could be put to other uses or tells us how to act. Private institutions, by contrast, thrive only to the extent that they provide goods and services we want and are willing to pay for. And, whereas we are compelled to support government, like it or not, we are not compelled to support businesses whose products, services, and rules we may not like.
Are there, nevertheless, gunctions that should be entrusted to government, even where private institutions could perform those functions more efficiently? The answer is "yes," but to get to it I must first give the general argument against government.
The General Case against Government
Every governmental act -- from taxation by local authorities to fighting a foreign war -- disrupts our private affairs and results in the redistribution of wealth and income. Governmental acts therefore make some persons better off and some persons worse off.
Apologists for government argue from the Benthamite principle of "the greatest good for the greatest number" -- as if the individual gains and losses from a governmental act could be summed to show that the act has a "net positive effect on the social welfare function." But there is no such thing as "social welfare"; there are only individual states of well-being, which cannot be summed. To put it another way, one person's happiness is incommensurate with another person's sadness. No matter how much money one person has, it is presumptious of government to take some of that person's money and give it to another person. But that is exactly what government does whenever it acts -- often, quite deliberately and openly.
The Special Case for Government
The proper role of government is to enable each of us to strive to maximize his or her well-being, as long as we don't forcefully or fraudulently diminish the well-being of other persons in our striving.
Protection from force and fraud may be a public good. I say "may" because I'm not sure, but I suppose the following argument can be made: Government is the only resort for "insurance" against force and fraud. Some forms of private insurance can be bought (e.g., bodyguards), but private insurance is "unenforceable" unless government is there to back it up. I would be willing to stipulate that adequate defense forces and criminal-justice systems can be raised through "user fees" (voluntary taxation at a flat rate or amount), the free-rider problem notwithstanding.
Government interference is therefore illegitimate (except in cases of force or fraud) because it presumes to make society as a whole better off when it cannot do so. As long as there are any losers, society as a whole cannot be better off.
There is no way government can compensate the losers without creating other losers. True, but speaking "Paretally" wouldn't the compensation have to be arranged in such a way that only those taxpayers who want to compensate transit users are required to pay compensation, and then only in amounts they are willing to pay? Here, of course, it seems unlikely that the free-rider problem could be overcome, which suggests that public roads, rights of way, and transportation systems ought to be sold at auction to private operators.
Therefore, government should not interfere in private affairs, except in cases of force or fraud. Tough public-policy questions remain; for example: Should government interfere before the fact to prevent force or fraud, or should government wait until there is evidence of force or fraud? How much government interference is "enough" and who should finance it? Can the failure to disclose information be fraudulent? What about the failure to disclose things that some consumers may consider harmful (e.g., the presence of gene-spliced foods) but which the producer does not (a) know about or (b) consider harmful?
