And the smaller and more detailed the tasks in your list, the more likely they are to become obsolete/wrong before you get to actually implementing them.
Change is fine. Keep in mind we are talking about a task list for estimates, not implementation level requirements. Estimates for smaller pieces of work are generally more accurate than estimates for larger pieces of work and usually lead to more accurate overall project schedules.
Without adding some detail to our discussion, I am not sure how much we agree (or disagree). This will vary a lot from project to project, but here are my thoughts:
* if the overall project is person-months of effort, try go down to the level of detail of days per task
* if the overall project is person-years of effort, try go down to the level of detail of weeks per task
Do you think this is too detailed?