I have hesitated getting involved in this, but...

One is that while valuing site values in central cities can be done with not 
too much difficulty,
the outcome can be somewhat arbitrary.  I have looked at data series on 
vacant land sales
in urban areas, and while there are thick markets and stable values near the 
edge of cities,
there can be enormous variation of values near the center where markets are 
thin, variations
up to an order of magnitude for essentially similar lots at similar times. 
So, the fact that
assessors or appraisers may not vary that much does not prove that they are 
accurately
reflecting the market.  Nevertheless I have always had sympathy for the land 
tax argument
in urban areas.  Clearly it does not provide the negative incentives that 
other taxes do.

The main problem with the old George proposal to replace all other taxes 
with a land or site
tax is that governments are now simply too large to be funded out of the 
income that goes just
to land.  Maybe that would have worked in the 19th century, but no more.

Regarding this issue about the nature of property in land, the old 
institutionalists are right on
this.  It is a very complicated matter of legal traditions and customs, and 
there is much more
of a spectrum from full-blown state ownership to pure private ownership, 
involving all the
various aspects of what one can do with land.  So, even in the case of the 
US that Deirdre
admires with supposedly full-blown private ownership, the state still 
retains the right of
eminent domain to take land for public uses, supposedly with appropriate 
compensation,
of course.  In the original 13 colonies, the states are the residual owners 
of "unowned" land,
and there is such land here in Virginia, where I live, whereas the federal 
government is the
residual owner in the rest of the country, over 80% in Nevada and Alaska. 
In Virginia,
the US Forest Service has a deal with the state that the state will give 
such land for free to
the Forest Service in the areas supposed to become national forests, if the 
feds bear the
transactions costs of verifying that the land is indeed unowned, which is a 
non-trivial issue
given the weird state of old deeds from the colonial period and the 400 year 
history of VA.

I would also note that not only can site ownership be separated from 
structures ownership,
but so can underground ownership, even to the legal system used.  So, in St. 
Louis, MO,
surface and structure rights follow the US/British system, with the squares 
and rectangles
laid down by Jefferson for the Northwest Territories Ordinance, whereas the 
subsurface
rights follow the French land tenure system, the French having founded St. 
Louis.  That
system, followed on the surface in parts of Wisconsin such as near Green Bay 
and Prairie
du Chien, has long narrow strips of land that go to the nearest river, which 
was, of course,
the main route of transportation in the colonial era.

Barkley Rosser