> if you publish paper A, and then cite it in paper B which builds on that work, then in paper C you really only need to cite paper B if you're building on the work, not B and A.

Logically I agree with you, but a lot of academics seem to believe differently when it comes to citing other people's work, and if we are to go by that logic (which a lot of people are inevitably forced to do), I don't see why one should treat their own work any differently.