If Land Value is to replace other taxes, an important question is: to what extent?

How great a proportion of government receipts would or could be raised from LVT? Some of the arguments in the paper suggest that it would be substantial, but I think there is an upper limit to the proportion, which depends on the significance of land as a factor of production. It is clearly a much less significant factor of production in a knowledge economy than it is in an agrarian one.

Theoretically there's no upper limit. But in the real world there are limits set by practical politics. There are also other reasons to tax, e.g. 'sin taxes'. If you look at Australia, about 10% of all public revenue (Commonwealth, States, local) comes from land values (that includes rent income from publicly owned land but is mainly LVT/SVR). In Hong Kong, where the government owns virtually all land, about 40% of all revenue comes from land values (not as 'real' LVT - read my paper).
In my view, you need to aim for around 5% initially for it to be worth the cost of doing it at all, but have a target of 15-20% within 10 years.

Land tax is limited by land values. Clearly the rate should not be 100% or anywhere near - making land notionally worthless would destroy the market for land, the basis for valuation, and would be expropriation of property.

So I would expect an upper limit of certainly less than 50%, maybe around 20%. How much is all the land in the country worth, annually?

You might expect the significance of land to go down, but land values have kept going up in spite of the relative insignificance of agriculture to the economy. I guess the desirability of land as an asset is not entirely to do with its productive potential.