Umm, there is no contextual reality which was missed.

As a tool, sample surveys are effective and reflective of reality.

The initial claim was that sample surveys are unrepresentative, so if one doubts one sample survey, then all sample surveys are suspect unless you have some specific knowledge about the methodological flaws in this sample survey.

Since you haven't mentioned methodological flaws, either you reject all sample surveys as a tool or accept that the claim that this sample survey is wrong is dubious.

2. I'm curious which exit poll in the last generation has proved to be wrong in your view. Minor differences in final voting percentages is hardly reason to reject exit polls. If we can rely on exit polls, which are self-reported voting preferences, to predict final vote tallies, we should be able to rely on FIES surveys as well. Do you really believe that enough surveyed families are lying to FIES just to hurt India's hunger index rank? Seems like a stretch to me.