When we talk about the demand curve for a good, what we mean is how the quantity consumed for that exact same good changes with the price of that good while holding everything else constant (such as the consumer's income, the price of other goods, etc). A moment's reflection makes it obvious that the customer who purchases the high-price prostitute would demand just as much or more of her services if she were willing to do all the same things but at half the price. Similarly, the customer who chooses the low-price prostitute would also consume more of her services if her price were halved. If this is the case, the demand for prostitutes indeed slopes downward, just like the demand for virtually every other good known to mankind.
So how is it that rice in rural China might violate this rule? How could it be that when the price of rice rises, people actually consume more of it? Two factors are critical.
First, rice makes up a large share of the total expenditures of these Chinese peasants. Second, if they were richer, these peasants would prefer to eat less rice and more of other things, like meat; it is just that they are too poor right now to afford much meat and they have to eat something. When rice becomes more expensive, one effect of the higher price is to make the peasants want to consume less of it (just as johns do with prostitutes who raise their prices).
Now I'm someone who thinks prostitution should be decriminalized. But even so, I recognize the nastiness of that life for the majority of sex workers, and to compare the "consumption" of a prostitute's services to consuming rice is so incredibly offensive. But even more offensive is some of the commenters on the Freakonomics blog - none of whom, I'll wager, have ever had to worry about selling their bodies to survive - or being sold into sex slavery by their own parents, or kidnapped into it - the fate of thousands or even millions of girls around the world.
What do Steve D. Levitt and a douchebag have in common? What don't they?