That's not really accurate. Yes, the total amount of food may not change, but the type of food changes quite a bit. People will eat less desirable, but cheaper food.
The article mentions that it would be better if some of the farmers started growing vegetables rather than grain, but the subsidies don't pay for vegetables.
Same for wheat - low gluten types are cheaper. (Which is basically why they invented the http://en.wikipedia.org/wiki/Chorleywood_bread_process )
However there is a long lead time for changing how many animals there are in the food pipeline. Thus the amount of meat animals does not increase very rapidly to respond to a bumper crop in wheat. (The meat supply can, however, drop rapidly after a drought or flood...)
And the yearly fluctuations in crop yields tend to be fairly large. Certainly larger than the variation we're likely to see in how much people want to eat.
Yeah it's not like there are mechanisms to deal with this sort of volatility (hint - futures).
It is also a technique mostly for businesses, not consumers. How many consumers do you hear about controlling next month's food bill by buying their produce on a future's market? Right, none.
This ignores a subtlety. If one consumer locks in a large fraction of the supply at a low price, then the supply is smaller than expected, you've made the supply-demand problem more extreme for everyone else. So futures allow entities to lessen their own volatility, but that increases everyone else's volatility.
The implication is that even slight excesses result in rock bottom prices, and even slight shortages result in very high prices. This volatility is bad for everyone. Therefore the government steps in with subsidies to even out how much farmers get, and to guarantee a surplus to avoid ruinous food prices for consumers.
But there is simply no good way to do this without perverse consequences somewhere...