For background, see prior posts here and here
What I would really like to see, though it will never happen, is a controlled test of FSA servicing crop insurance versus the companies servicing crop insurance. Choose 10 counties and have half the current insureds serviced by FSA and half by the companies. Run the test for 3 years and compare results.
Why won't it happen? Disregarding the political realities, the practical consideration is: to service crop insurance you need to develop software. To do the job right, you need the software to be integrated with FSA's existing or to-be systems. That takes time and money, which you can't justify for a limited test. (Though it's what we did with CAT insurance over a couple years. Unfortunately FSA management in DC spun its wheels for some months so the first year was pretty grim. The second year was significantly better. And then the big shots proclaimed that CAT was available through insurance agents through the whole country.)
Of course, if the result of the test were to show a big advantage for FSA (which most people who worked for FSA would like to believe) and you extrapolate that over the country and over 40 years, then it's worthwhile to do the test. But that's not how we do things in this country.