Distribution is a huge problem. Often moving food from point A to B is so expensive the economics of it just don't make any sense.
You can double you production. But what good will that do if the cost of moving it to hungry masses is not feasible? So really its not so complex, at the end it all boils down to a term called 'energy' and we just keep going in rounds around it.
Actually, in India, it doesn't. The energy intensity of an ideal distribution network in India has a fraction of the friction of the existing system. For the case provided it boils down to efficacy of governance, corruption, and transparency.
See the previous topic on folks starving in India not because of a lack of food but because of corruption. So I could totally buy into the idea that people will riot and assert their right to be governed in the presence of mass corruption which prevents them from eating. It worked for Mao and it has worked for others. Its easy to incite someone who is hungry and cannot afford food to riot (systemic input). So it becomes a priority for the government to avoid that, lest it lose its authority (systemic response).
People have been predicting future behavior of complex systems for ever, and they are always wrong. Its very strange at first but it arises from the challenge of predicting things you don't know a-priori.
The Indian government could (not that they will but could) nationalize several million hectares of arable land and apply modern farming technology to them. Would double the amount of food. China's modernization program is resulting steadily growing production [2].
The price of food rises, and the production of food rises as well. Complex systems respond to a variety of inputs, not particularly easy to predict their collapse unless you can show something which can force them into failure.
[1] "According to the latest FAO report, the average yield of rice in India is 2.3 tonne/ha as against the global average of 4.374 tonne/ha. " http://articles.economictimes.indiatimes.com/2012-04-02/news...
[2] http://jxb.oxfordjournals.org/content/early/2011/05/06/jxb.e...