always work on the step you know you can take next. There are lots of unknowns in every project, but there are usually a few knowns. Usually the problem is not pulling the trigger to move forward on the knowns. You will have to re-write, you will have to backtrack, you will end up wasting oodles of code and time, but steps backwards are steps forward when learning is a part of it, and learning is always a part of it.