It does not sound backwards at all. That is what TDD is for: To start writing the visible portions of your program to see how it works and adjust accordingly as user needs dictate. Once your program does what the user wants, all you have to do is backfill the "black box" implementation, using test cases created during the discovery phase to ensure that the "black box" provides what the user came to expect. The scenario you present is exactly what TDD was envisioned for.

If you know the exact requirements upfront, you don't really need TDD. You can simply write the code to those requirements and, perhaps, add in some automated acceptance testing to help catch mistakes. TDD shines when you are unsure and exploring options.