There's more than upkeep, there's also risk. Any product that collects people's data has privacy and regulatory consequences that exposes the company to risk. You can't just put a product that has those kinds of risks on "autopilot".
Technically, one or two engineers probably could keep something like Reader going, a drop in the bucket for Google, a rounding error. The true cost is not the upkeep, but the costs of the risks. If not investing any significant resources into upgrading and maintaining Reader means some snafu over personal data leakage, then you're looking at being hauled in front of Senate panels, and EU regulators, and civil suits.
Remember, Reader was written a long time ago by an understaffed and under resourced team.
Technically, one or two engineers probably could keep something like Reader going, a drop in the bucket for Google, a rounding error. The true cost is not the upkeep, but the costs of the risks. If not investing any significant resources into upgrading and maintaining Reader means some snafu over personal data leakage, then you're looking at being hauled in front of Senate panels, and EU regulators, and civil suits.
Remember, Reader was written a long time ago by an understaffed and under resourced team.