[sidlls]

One of the theses of TDD is that the tests guide the design and implementation of an under specified (e.g. unknown) problem, given the requirements regarding the outcomes and a complete enough set of test cases. “Theoretically” one should be able to develop a correct solver without knowing how it works by iterative improvements using TDD. It might not be of good quality, but it should work.

Note: I am quite skeptical of TDD in general.

[nightski]

I don't really use TDD, but I've never heard that TDD would help guide the implementation. I always understood it was about designing a clean interface to the code under test. This being a result from the fact that you are designing the interface based on actual use cases first since the test needs to call into the code under test. It helps avoid theoretical what ifs and focus on concrete, simple design.

Personally I think that one can learn this design methodology without TDD. I find learning functional programming and say Haskell/OCaml/SML/etc.. far more beneficial to better design here than I do TDD.

[sidlls]

It’s both.

In theory TDD drives the interface by ensuring the units under test do what they’re intended (implementation), and that each and every unit is “testable” (interface).

TDD doesn’t really care about “clean” interfaces, only that units of work (functions, methods) are “testable”.

I’d argue this actually creates friction for designing clean interfaces, because in order to satisfy the “testability” requirement one is often forced to make poor (in terms of readability, maintainability, and efficiency) design choices.