If you don't mind a broader answer...
When I have an idea which is new to me and off on a tangent (for example, the idea of a 3D Solar System sim) then I'll just bang away at it any old how, but knowing that at some point I will probably decide 'yes, this is working' (or no, it's not) and know that I will have to stop, go back to the beginning and restructure and properly design the code.
'Proof of Concept' is an important, but can be a very quick, stage.
When I have an idea which is new to me and off on a tangent (for example, the idea of a 3D Solar System sim) then I'll just bang away at it any old how, but knowing that at some point I will probably decide 'yes, this is working' (or no, it's not) and know that I will have to stop, go back to the beginning and restructure and properly design the code.
'Proof of Concept' is an important, but can be a very quick, stage.