I love art. Whenever I travel abroad, I'd love to visit museums. Many of those museums are publicly supported. Meaning, the government collects money from taxpayers and allocate part of it to maintain museums.
When I pay cheap pass to a museum, I feel weird. Back then, honestly, I didn't care: hey, who doesn't love subsidy? But the more museums I visit, the more I think it's unfair to charge me less to enjoy the artifacts in Makassar's Fort Rotterdam, than to stare at Mona Lisa in Louvre. In other day, when I visited The Art Institute of Chicago Museum, I knew some Americans paid partial cost of my pass, through taxation. But not everybody of them loves museum. In short, the costs of art subsidy are very dispersed, while the benefits are concentrated.
My co-host Ujang loves art, too. So do many other goers of Yale University Art Museum. They all benefit if the pass to the museum is subsidized. But who bears the cost? Some art lovers and some art non-lovers (plus, probably some art-haters). Is it fair? Maybe not. But isn't this how every kind of public subsidy works? If that's bad, then every subsidy is bad.
Well, consider national defense. There is no way (at least, there should be no way -- that's the idea) in which the government can exclude certain citizens from national defense/security. This is called "non-excludability". And my consuming the good should not reduce the level that you can consume. This is "non-rivalrous". Art may qualify for the second condition, but almost always not for the first one.
Of course the above definition of "pure" public good is blur. So, let's just do the easy way: Marshallian distribution of wealth. If the aggregate benefits are greater than the aggregate costs, then it's a go. National defense is surely a go. But I'm afraid art is not: the costs to the many taxpayers are likely to exceed the benefits to the minority, museum visitors.
There have been some ways offered around this problem. One of which is to use "local subsidization" of culture. It works like this, more or less: it's alright to collect taxes from Makassar citizens to pay the maintenance of Fort Rotterdam. But not from Javanese. The more able the government to identify who's willing to pay more, the more succesful the program. Nobody says this is easy.
For good and healthy debate, see the special issue of Journal of Behavioral Economics, "Symposium on Subsidization of Cultural Activities", Vol 8 Issue 1, Summer 1979 (esp. the paper by Milton Russel, pg 69-75).