I hear what you're saying - I'm guessing my question is a little too general. I guess what I was wondering is whether or not this is a common pitfall into which people fall.
It's not uncommon. In fact, it happens pretty often. Whether it's a pitfall depends on the circumstances.
As a project gets further in development, the cost of implementing changes increases. At a certain point, it becomes unfeasible. I don't know the scope of your project, if you're not even at 'working prototype' yet, it sounds like you've still got room to make significant changes before locking anything in.
Edit: Super Giant just released a
large article on how they changed the implementation of their battle system significantly after hammering on it for some time. Might be relevant. A good read all the same.